Scribd
Upload
324 views1 page

A Network Flow Model For The Vehicle Routing Problem With Time Windows and Multiple Routes

The document describes a network flow model and exact solution algorithm for solving the Vehicle Routing Problem with Time Windows and Multiple Routes (MVRPTW). The network flow model represents each vehicle workday as a path in an acyclic directed graph where nodes correspond to time instants and arcs correspond to feasible routes. The algorithm discretizes time using rounding procedures and embeds the model in an algorithm that uses disaggregation to correct any issues from rounding. Computational experiments showed the method clearly outperforms the only other exact approach and reduces computational time by around 95% in cases where both methods find a solution.

Uploaded by

Semana de Engenharia '2010
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
324 views1 page

A Network Flow Model For The Vehicle Routing Problem With Time Windows and Multiple Routes

The document describes a network flow model and exact solution algorithm for solving the Vehicle Routing Problem with Time Windows and Multiple Routes (MVRPTW). The network flow model represents each vehicle workday as a path in an acyclic directed graph where nodes correspond to time instants and arcs correspond to feasible routes. The algorithm discretizes time using rounding procedures and embeds the model in an algorithm that uses disaggregation to correct any issues from rounding. Computational experiments showed the method clearly outperforms the only other exact approach and reduces computational time by around 95% in cases where both methods find a solution.

Uploaded by

Semana de Engenharia '2010
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 1

A NETWORK FLOW MODEL FOR THE VEHICLE ROUTING PROBLEM WITH TIME

WINDOWS AND MULTIPLE ROUTES R&D CENTRE


ALGORITMI

RITA MACEDO*
University of Minho Supervisors: Cláudio Alves, J.M. Valério de Carvalho
School of Engineering
Algoritmi Research Center
* [email protected]

Introduction We developed a new network flow model, and an exact solution This algorithm ensures that our solution method is exact.
algorithm to solve this problem. In the network flow model, every It consists of disaggregating some nodes of our graph whenever it is
The Vehicle Routing Problem (VRP) is a combinatorial optimization workday (sequence of routes assigned to one vehicle for one planning necessary. This means that we go back and consider the real values
problem that has been widely studied in the literature. Generally period) corresponds to a path in an acyclic directed graph. It is of some eventually conflicting arc extremities in order to correct the
speaking, it is the problem of scheduling a fleet of vehicles to visit a formulated as a minimum flow problem, whose nodes correspond to solution.
set of customers, to whom they must deliver a demanded quantity of time instants of one workday and whose variables correspond to
goods. The problem consists of finding the best set of routes, feasible routes.
according to a given objective function, such that all operational
constraints of the vehicles are respected, and the set of customers is
covered. The VRP is well-know to be NP-hard, and so are most of its As the nodes of the graph represent time instants, a discretization of
variants. Its solution methods include several heuristic and time is required. We do it by using some rounding procedures.
metaheuristic approaches, as well as some exact methods, mainly
based on branch-and-bound techniques.
We address a variant of the classical VRP, the Vehicle Routing
Problem with Time Windows and Multiple Routes (MVRPTW). Despite
its apparent practical relevance (delivering perishable goods, for
example), this variant of the classical VRP has not been the subject of
a large number of studies. It considers that a same vehicle can be
assigned with more than one route per planning period. This means
that it can perform one route, reload at the depot and leave to the
following route until the end of the workday. In addition, all customers
must be served within a predefined time period that corresponds to
their time windows. All vehicle routes are previously generated. The integer model is
then solved with a commercial software (CPLEX), explicitly
SOLUTION METHOD considering all its variables. Variables represent feasible vehicle
routes in the network flow model. Reducing the number of arcs
reduces the size of the model, increasing its efficiency. We derived
four propositions defining dominance rules that allow us to discard
some routes that are never interesting when compared to some other
one. With them, we were able to reduce approximately 29% of
variables to include in the model.

Computational Results
We conducted some computational experiments on a set of
benchmark instances and compared our algorithm with the only other
exact approach to this variant proposed in the literature, to our
knowledge, which is a branch-and-price algorithm.

Our method clearly outperforms the branch-and-price method in


Disaggregation for unfeasibility I Disaggregation for unfeasibility II most cases. In the cases where the two methods find a solution, our
We embedded our model in an algorithm that allows us to correct method dramatically reduces the computational time needed. This
some problems that may occur due to those rounding procedures. reduction is around 95%.

Example of one path in the network flow graph

Engenharia para a Qualidade de Vida Semana da Escola de Engenharia -11 a 16 de Outubro de 2010

You might also like