Geir Hasle

Invited Review: Industrial aspects and literature survey: Combined inventory management and routing

This paper describes industrial aspects of combined inventory management and routing in maritime and road-based transportation, and gives a classification and comprehensive literature review of the current state of the research. The literature is contrasted with aspects of industrial applications from a constructive, but critical, viewpoint. Based on the status and trends within the field, future research is suggested with regard to both further development of the research area and industrial needs. By highlighting the industrial aspects, practitioners will hopefully see the benefit of using advanced decision support systems in complex situations related to combined inventory management and routing in their business. In addition, a classification and presentation of the research should help and motivate researchers to further focus on inventory management and routing challenges.

Review: Industrial aspects and literature survey: Fleet composition and routing

The purpose of this paper is to describe industrial aspects of combined fleet composition and routing in maritime and road-based transportation, and to present the current status of research in the form of a comprehensive literature review. First, presents a classification of problems, and then focuses on a basic definition of combined fleet composition and routing: the fleet size and mix vehicle routing problem. A basic mathematical formulation from the literature is presented. Further, the literature of extended and related problems is described and categorized. Surveys of application oriented research in road-based and maritime transportation conclude the review. Finally, we contrast the literature with aspects of industrial applications from a critical, but constructive stance. Major issues for future work are suggested.

Metaheuristics for the Vehicle Routing Problem and its Extensions: A Categorized Bibliography

We provide a categorized bibliography of metaheuristics for solving the vehicle routing problem and its extensions. The categories are based on various types of metaheuristics and vehicle routing problems.

A multi-start local search algorithm for the vehicle routing problem with time windows

In this paper a multi-start local search (MSLS) heuristic is proposed for the vehicle routing problem with time windows (VRPTW). In VRPTW the objective is to design least cost routes for a fleet of identical capacitated vehicles to service geographically scattered customers within pre-specified service time windows. The suggested approach is based on a MSLS framework and several new improvement heuristics. A new speedup technique is introduced for the construction heuristics, and the results of the MSLS are post-optimized by a threshold accepting post-processor. Experimental results on 358 benchmark problems from the literature show that the suggested method is highly efficient and competitive.

An Effective Multirestart Deterministic Annealing Metaheuristic for the Fleet Size and Mix Vehicle-Routing Problem with Time Windows

This paper presents a new deterministic annealing metaheuristic for the fleet size and mix vehicle-routing problem with time windows. The objective is to service, at minimal total cost, a set of customers within their time windows by a heterogeneous capacitated vehicle fleet. First, we motivate and define the problem. We then give a mathematical formulation of the most studied variant in the literature in the form of a mixed-integer linear program. We also suggest an industrially relevant, alternative definition that leads to a linear mixed-integer formulation. The suggested metaheuristic solution method solves both problem variants and comprises three phases. In Phase 1, high-quality initial solutions are generated by means of a savings-based heuristic that combines diversification strategies with learning mechanisms. In Phase 2, an attempt is made to reduce the number of routes in the initial solution with a new local search procedure. In Phase 3, the solution from Phase 2 is further improved by a set of four local search operators that are embedded in a deterministic annealing framework to guide the improvement process. Some new implementation strategies are also suggested for efficient time window feasibility checks. Extensive computational experiments on the 168 benchmark instances have shown that the suggested method outperforms the previously published results and found 167 best-known solutions. Experimental results are also given for the new problem variant.

Industrial Vehicle Routing

Solving the Vehicle Routing Problem (VRP) is a key to efficiency in transportation and supply chain management. The VRP is an NP-hard problem that comes in many guises. The VRP literature contains thousands of papers, and VRP research is regarded as one of the great successes of OR. Vehicle routing decision support tools provide substantial savings in society every day, and an industry of routing tool vendors has emerged. Exact methods of today cannot consistently solve VRP instances with more than 50–100 customers in reasonable time, which is generally a small number in real-life applications. For industrial problem sizes, and if one aims at solving a variety of VRP variants, approximation methods is the only viable approach. There is still a need for VRP research, particularly for large-scale instances and complex, rich VRP variants. In this chapter, we give a brief general introduction to the VRP. We then describe how industrial requirements motivate extensions to the basic, rather idealized VRP models that have received most attention in the research community, and how such extensions can be made. At SINTEF Applied Mathematics, industrial variants of the VRP have been studied since 1995. Our efforts have led to the development of a generic VRP solver that has been commercialized through a spin-off company. We give a description of the underlying, rich VRP model and the selected uniform algorithmic approach, which is based on metaheuristics. Finally, results from computational experiments are presented. In conclusion, we point to important issues in further VRP research.

Dynamic And Stochastic Vehicle Routing In Practice

The VRP is a key to efficient transportation logistics. It is a computationally very hard problem. Whereas classical OR models are static and deterministic, these assumptions are rarely warranted in an industrial setting. Lately, there has been an increased focus on dynamic and stochastic vehicle routing in the research community. However, very few generic routing tools based on stochastic or dynamic models are available. We illustrate the need for dynamics and stochastic models in industrial routing, describe the Dynamic and Stochastic VRP, and how we have extended a generic VRP solver to cope with dynamics and uncertainty

GPU computing in discrete optimization. Part II: Survey focused on routing problems

In many cases there is still a large gap between the performance of current optimization technology and the requirements of real-world applications. As in the past, performance will improve through a combination of more powerful solution methods and a general performance increase of computers. These factors are not independent. Due to physical limits, hardware development no longer results in higher speed for sequential algorithms, but rather in increased parallelism. Modern commodity PCs include a multi-core CPU and at least one GPU, providing a low-cost, easily accessible heterogeneous environment for high-performance computing. New solution methods that combine task parallelization and stream processing are needed to fully exploit modern computer architectures and profit from future hardware developments. This paper is the second in a series of two. Part I gives a tutorial style introduction to modern PC architectures and GPU programming. Part II gives a broad survey of the literature on parallel computing in discrete optimization targeted at modern PCs, with special focus on routing problems. We assume that the reader is familiar with GPU programming, and refer the interested reader to Part I. We conclude with lessons learnt, directions for future research, and prospects.

GPU Computing in Discrete Optimization Part I: Introduction to the GPU

In many cases there is still a large gap between the performance of current optimization technology and the requirements of real world applications. As in the past, performance will improve through a combination of more powerful solution methods and a general performance increase of computers. These factors are not independent. Due to physical limits, hardware development no longer results in higher speed for sequential algorithms, but rather in increased parallelism. Modern commodity PCs include a multi-core CPU and at least one GPU, providing a low cost, easily accessible heterogeneous environment for high performance computing. New solution methods that combine task parallelization and stream processing are needed to fully exploit modern computer architectures and profit from future hardware developments. This paper is the first part of a series of two, where the goal of this first part is to give a tutorial style introduction to modern PC architectures and GPU programming. We start with a short historical account of modern mainstream computer architectures, and a brief description of parallel computing. This is followed by the evolution of modern GPUs, before a GPU programming example is given. Strategies and guidelines for program development are also discussed. Part II gives a broad survey of the existing literature on parallel computing targeted at modern PCs in discrete optimization, with special focus on papers on routing problems. We conclude with lessons learnt, directions for future research, and prospects.

A lower bound for the Node, Edge, and Arc Routing Problem

The Node, Edge, and Arc Routing Problem (NEARP) was defined by Prins and Bouchenoua in 2004, although similar problems have been studied before. This problem, also called the Mixed Capacitated General Routing Problem (MCGRP), generalizes the classical Capacitated Vehicle Routing Problem (CVRP), the Capacitated Arc Routing Problem (CARP), and the General Routing Problem. It captures important aspects of real-life routing problems that were not adequately modeled in previous Vehicle Routing Problem (VRP) variants. The authors also proposed a memetic algorithm procedure and defined a set of test instances called the CBMix benchmark. The NEARP definition and investigation contribute to the development of rich VRPs. In this paper we present the first lower bound procedure for the NEARP. It is a further development of lower bounds for the CARP. We also define two novel sets of test instances to complement the CBMix benchmark. The first is based on well-known CARP instances; the second consists of real life cases of newspaper delivery routing. We provide numerical results in the form of lower and best known upper bounds for all instances of all three benchmarks. For three of the instances, the gap between the upper and lower bound is closed. The average gap is 25.1%. As the lower bound procedure is based on a high quality lower bound procedure for the CARP, and there has been limited work on approximate solution methods for the NEARP, we suspect that a main reason for the rather large gaps is the quality of the upper bound. This fact, and the high industrial relevance of the NEARP, should motivate more research on approximate and exact methods for this important problem.