Robust uncapacitated multiple allocation hub location problem under demand uncertainty: minimization of cost deviations

Aleksejs Lozkins, Mikhail Krasilnikov, Vladimir Bure

Research output

Abstract

The hub location–allocation problem under uncertainty is a real-world task arising in the areas such as public and freight transportation and telecommunication systems. In many applications, the demand is considered as inexact because of the forecasting inaccuracies or human’s unpredictability. This study addresses the robust uncapacitated multiple allocation hub location problem with a set of demand scenarios. The problem is formulated as a nonlinear stochastic optimization problem to minimize the hub installation costs, expected transportation costs and expected absolute deviation of transportation costs. To eliminate the nonlinearity, the equivalent linear problem is introduced. The expected absolute deviation is the robustness measure to derive the solution close to each scenario. The robust hub location is assumed to deliver the least costs difference across the scenarios. The number of scenarios increases size and complexity of the task. Therefore, the classical and improved Benders decomposition algorithms are applied to achieve the best computational performance. The numerical experiment on CAB and AP dataset presents the difference of resulting hub networks in stochastic and robust formulations. Furthermore, performance of two Benders decomposition strategies in comparison with Gurobi solver is assessed and discussed.

Original languageEnglish
JournalJournal of Industrial Engineering International
DOIs
Publication statusE-pub ahead of print - 1 Jan 2019

Fingerprint

Costs
Decomposition
Telecommunication systems
Freight transportation
Uncertainty
Experiments

Scopus subject areas

  • Industrial and Manufacturing Engineering

Cite this

@article{7443ec01665e4bf49389033c5ca142e8,
title = "Robust uncapacitated multiple allocation hub location problem under demand uncertainty: minimization of cost deviations",
abstract = "The hub location–allocation problem under uncertainty is a real-world task arising in the areas such as public and freight transportation and telecommunication systems. In many applications, the demand is considered as inexact because of the forecasting inaccuracies or human’s unpredictability. This study addresses the robust uncapacitated multiple allocation hub location problem with a set of demand scenarios. The problem is formulated as a nonlinear stochastic optimization problem to minimize the hub installation costs, expected transportation costs and expected absolute deviation of transportation costs. To eliminate the nonlinearity, the equivalent linear problem is introduced. The expected absolute deviation is the robustness measure to derive the solution close to each scenario. The robust hub location is assumed to deliver the least costs difference across the scenarios. The number of scenarios increases size and complexity of the task. Therefore, the classical and improved Benders decomposition algorithms are applied to achieve the best computational performance. The numerical experiment on CAB and AP dataset presents the difference of resulting hub networks in stochastic and robust formulations. Furthermore, performance of two Benders decomposition strategies in comparison with Gurobi solver is assessed and discussed.",
keywords = "Absolute deviation, Benders decomposition, Hub location problem, Pareto-optimal cuts, Robust solution, Stochastic programming",
author = "Aleksejs Lozkins and Mikhail Krasilnikov and Vladimir Bure",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/s40092-019-00329-9",
language = "English",
journal = "Journal of Industrial Engineering International",
issn = "1735-5702",
publisher = "Springer",

}

TY - JOUR

T1 - Robust uncapacitated multiple allocation hub location problem under demand uncertainty

T2 - minimization of cost deviations

AU - Lozkins, Aleksejs

AU - Krasilnikov, Mikhail

AU - Bure, Vladimir

PY - 2019/1/1

Y1 - 2019/1/1

N2 - The hub location–allocation problem under uncertainty is a real-world task arising in the areas such as public and freight transportation and telecommunication systems. In many applications, the demand is considered as inexact because of the forecasting inaccuracies or human’s unpredictability. This study addresses the robust uncapacitated multiple allocation hub location problem with a set of demand scenarios. The problem is formulated as a nonlinear stochastic optimization problem to minimize the hub installation costs, expected transportation costs and expected absolute deviation of transportation costs. To eliminate the nonlinearity, the equivalent linear problem is introduced. The expected absolute deviation is the robustness measure to derive the solution close to each scenario. The robust hub location is assumed to deliver the least costs difference across the scenarios. The number of scenarios increases size and complexity of the task. Therefore, the classical and improved Benders decomposition algorithms are applied to achieve the best computational performance. The numerical experiment on CAB and AP dataset presents the difference of resulting hub networks in stochastic and robust formulations. Furthermore, performance of two Benders decomposition strategies in comparison with Gurobi solver is assessed and discussed.

AB - The hub location–allocation problem under uncertainty is a real-world task arising in the areas such as public and freight transportation and telecommunication systems. In many applications, the demand is considered as inexact because of the forecasting inaccuracies or human’s unpredictability. This study addresses the robust uncapacitated multiple allocation hub location problem with a set of demand scenarios. The problem is formulated as a nonlinear stochastic optimization problem to minimize the hub installation costs, expected transportation costs and expected absolute deviation of transportation costs. To eliminate the nonlinearity, the equivalent linear problem is introduced. The expected absolute deviation is the robustness measure to derive the solution close to each scenario. The robust hub location is assumed to deliver the least costs difference across the scenarios. The number of scenarios increases size and complexity of the task. Therefore, the classical and improved Benders decomposition algorithms are applied to achieve the best computational performance. The numerical experiment on CAB and AP dataset presents the difference of resulting hub networks in stochastic and robust formulations. Furthermore, performance of two Benders decomposition strategies in comparison with Gurobi solver is assessed and discussed.

KW - Absolute deviation

KW - Benders decomposition

KW - Hub location problem

KW - Pareto-optimal cuts

KW - Robust solution

KW - Stochastic programming

UR - http://www.scopus.com/inward/record.url?scp=85071439193&partnerID=8YFLogxK

U2 - 10.1007/s40092-019-00329-9

DO - 10.1007/s40092-019-00329-9

M3 - Article

AN - SCOPUS:85071439193

JO - Journal of Industrial Engineering International

JF - Journal of Industrial Engineering International

SN - 1735-5702

ER -