A Benders decomposition based framework for solving cable trench problems

Hatice Calik, M. Leitner, Martin Luipersbeck

Research output: Contribution to JournalArticleAcademicpeer-review

Abstract

In this work, we present an algorithmic framework based on Benders decomposition for the Capacitated p-Cable Trench Problem with Covering. We show that our approach can be applied to most variants of the Cable Trench Problem (CTP) that have been considered in the literature. The proposed algorithm is augmented with a stabilization procedure to accelerate the convergence of the cut loop and with a primal heuristic to derive high-quality primal solutions. Three different variants of the CTP are considered in a computational study which compares the Benders approach with two compact integer linear programming formulations that are solved with CPLEX. The obtained results show that the proposed algorithm significantly outperforms the two compact models and that it can be used to tackle significantly larger instances than previously considered algorithms based on Lagrangean relaxation.
Original languageEnglish
Pages (from-to)128-140
Number of pages13
JournalComputers and Operations Research
Volume81
DOIs
Publication statusPublished - 2017

Funding

This work is supported by the Vienna Science and Technology Fund (WWTF) through project ICT15-014. Part of this research has been performed while M. Leitner was a research fellow at the Department of Computer Science, Université Libre de Bruxelles (Brussels, Belgium) where he was supported by the Interuniversity Attraction Poles P7/36 COMEX initiated by the Belgian Science Policy Office. The research of H. Calik is supported by the e4-share (Models for Ecological, Economical, Efficient, Electric Car-Sharing) project funded by Innoviris, the Brussels Institute for Research and Innovation, via JPI Urban Europe and COMEX. M. Luipersbeck is supported by the University of Vienna through the uni:docs fellowship programme. These supports are greatly acknowledged.

FundersFunder number
COMEX
Vienna Science and Technology FundICT15-014
Belgian Federal Science Policy Office
Universität Wien
Innoviris
Université Libre de Bruxelles

    Keywords

    • Benders decomposition
    • Integer linear programming
    • Location
    • Network design

    Fingerprint

    Dive into the research topics of 'A Benders decomposition based framework for solving cable trench problems'. Together they form a unique fingerprint.

    Cite this