You are on page 1of 40

Logistics

Logı́stica empresarial

Jorge Oyola
Universidad de Córdoba

Jorge Oyola Logı́stica 1 / 24


Logistics Outline

Today’s plan

1 Routing problems

2 Arc routing problem

3 Traveling salesperson problem

4 The vehicle routing problem

Jorge Oyola Logı́stica 2 / 24


Routing problems

Routing problems

Jorge Oyola

Universidad de Córdoba

Jorge Oyola Routing problems 3 / 24


Routing problems Introduction

Some applications to real problems

Milk collection.
Forest harvesting.
Distribution of soft drinks.
Shipping.
Livestock collection.
Movement of observation satellites.
Solid waste collection.
Beer distribution.

Jorge Oyola Routing problems 4 / 24


ARP

Arc routing problem

Jorge Oyola

Universidad de Córdoba

Jorge Oyola ARP 5 / 24


ARP The bridges of Königsberg

A little bit of history

A question from the 18th century: Does


it exist a closed walk that crossed exactly
once each of the seven bridges of the
river Pregel? [Gribkovskaia et al., 2007]

Jorge Oyola ARP 6 / 24


ARP The bridges of Königsberg

A little bit of history

A question from the 18th century: Does


it exist a closed walk that crossed exactly
The bridges were on the river Pregel.
once each of the seven bridges of the
river Pregel? [Gribkovskaia et al., 2007]

Jorge Oyola ARP 6 / 24


ARP The bridges of Königsberg

A little bit of history

A question from the 18th century: Does


it exist a closed walk that crossed exactly
The bridges were on the river Pregel.
once each of the seven bridges of the
river Pregel? [Gribkovskaia et al., 2007] Leonhard Euler proved that the walk
did not exist.

Jorge Oyola ARP 6 / 24


ARP The bridges of Königsberg

A little bit of history

A question from the 18th century: Does


it exist a closed walk that crossed exactly
The bridges were on the river Pregel.
once each of the seven bridges of the
river Pregel? [Gribkovskaia et al., 2007] Leonhard Euler proved that the walk
did not exist.
For such a walk to exist, the graph
must be connected and every vertex
must have an even degree.

Jorge Oyola ARP 6 / 24


ARP The bridges of Königsberg

A little bit of history

A question from the 18th century: Does


it exist a closed walk that crossed exactly
The bridges were on the river Pregel.
once each of the seven bridges of the
river Pregel? [Gribkovskaia et al., 2007] Leonhard Euler proved that the walk
did not exist.
For such a walk to exist, the graph
must be connected and every vertex
must have an even degree.
Euler did not prove that these
conditions were sufficient. This was
proved later.

Jorge Oyola ARP 6 / 24


ARP The bridges of Königsberg

A little bit of history

A question from the 18th century: Does


it exist a closed walk that crossed exactly
The bridges were on the river Pregel.
once each of the seven bridges of the
river Pregel? [Gribkovskaia et al., 2007] Leonhard Euler proved that the walk
did not exist.
For such a walk to exist, the graph
must be connected and every vertex
must have an even degree.
Euler did not prove that these
conditions were sufficient. This was
proved later.
If such walk exists the graph is said to
be unicursal or eulerian.

Jorge Oyola ARP 6 / 24


ARP Conditions for a connected graph to be Eulerian

Conditions for every type of graph [Eiselt et al., 1995a]

Undirected

Every vertex must have an even degree.

Jorge Oyola ARP 7 / 24


ARP Conditions for a connected graph to be Eulerian

Conditions for every type of graph II

Directed

Jorge Oyola ARP 8 / 24


ARP Conditions for a connected graph to be Eulerian

Conditions for every type of graph II

Directed

The number of arcs entering and leaving each vertex must be equal.

Jorge Oyola ARP 8 / 24


ARP Conditions for a connected graph to be Eulerian

Conditions for every type of graph III


Mixed

Jorge Oyola ARP 9 / 24


ARP Conditions for a connected graph to be Eulerian

Conditions for every type of graph III


Mixed

Every vertex must be incident to an even number of directed and


indirected arcs.
For every subset S ∈ V , the difference between the number of
directed arcs from S to V \S and the number of arcs from V \S to S,
must be less than or equal to the number of undirected arcs joinning
V \S and S.
Jorge Oyola ARP 9 / 24
ARP Capacitated Arc Routing Problem (CARP)

CARP [Eiselt et al., 1995a, Eiselt et al., 1995b]

Graph (vertices and arcs).


Depot.
Set of arcs with non-negative
demand.
Traveling cost from vertex to
vertex.
Fleet of vehicles with equal
capacity.
Each arc is serviced exactly
once, but can be traversed as
many times as required.

Jorge Oyola ARP 10 / 24


ARP Capacitated Arc Routing Problem (CARP)

CARP [Eiselt et al., 1995a, Eiselt et al., 1995b]

Graph (vertices and arcs).


Depot.
Set of arcs with non-negative
demand.
Traveling cost from vertex to
vertex.
Fleet of vehicles with equal
capacity.
Each arc is serviced exactly
once, but can be traversed as
many times as required.

Purpose: to determine the service routes with the minimum cost

Jorge Oyola ARP 10 / 24


ARP Capacitated Arc Routing Problem (CARP)

Rural Postman Problem [Eiselt et al., 1995b]

Jorge Oyola ARP 11 / 24


ARP Capacitated Arc Routing Problem (CARP)

Rural Postman Problem [Eiselt et al., 1995b]

There is a set of towns. Their streets


must be served by the postman.

Jorge Oyola ARP 11 / 24


ARP Capacitated Arc Routing Problem (CARP)

Rural Postman Problem [Eiselt et al., 1995b]

There is a set of towns. Their streets


must be served by the postman.
There is a set of edges, that do not
require service, but they can be use to
travel from one town to another.

Jorge Oyola ARP 11 / 24


ARP Capacitated Arc Routing Problem (CARP)

Rural Postman Problem [Eiselt et al., 1995b]

There is a set of towns. Their streets


must be served by the postman.
There is a set of edges, that do not
require service, but they can be use to
travel from one town to another.
The total traveled distance is
minimized.

Jorge Oyola ARP 11 / 24


ARP Capacitated Arc Routing Problem (CARP)

Chinese postman problem [Eiselt et al., 1995a]

Jorge Oyola ARP 12 / 24


ARP Capacitated Arc Routing Problem (CARP)

Chinese postman problem [Eiselt et al., 1995a]

Every edge must be served.

Jorge Oyola ARP 12 / 24


ARP Capacitated Arc Routing Problem (CARP)

Chinese postman problem [Eiselt et al., 1995a]

Every edge must be served.


In case the graph is not unicursal, it
must be extended (adding extra
edges) at minimal cost.

Jorge Oyola ARP 12 / 24


ARP Capacitated Arc Routing Problem (CARP)

Chinese postman problem [Eiselt et al., 1995a]

Every edge must be served.


In case the graph is not unicursal, it
must be extended (adding extra
edges) at minimal cost.
The route that traverse all edges must
be found.

Jorge Oyola ARP 12 / 24


TSP

Traveling salesperson problem

Jorge Oyola

Universidad de Córdoba

Jorge Oyola TSP 13 / 24


TSP Introduction

Complexity

Traveling salesperson problem (TSP)


Find the shortest route that starts at the depot, visit all the customers and
returns to the depot. The distance between each point is given by the
Euclidean distance, i.e. the direct line between the points.

How many possible solutions may this problem have?


Jorge Oyola TSP 14 / 24
TSP Introduction

How difficult can it be?

This problem with 51 customers has 51! possible solutions


(51 · 50 · 49 · · · 2 · 1 = 1.55 · 1066 )

Jorge Oyola TSP 15 / 24


TSP Introduction

How difficult can it be?

This problem with 51 customers has 51! possible solutions


(51 · 50 · 49 · · · 2 · 1 = 1.55 · 1066 )

For comparison:
Some estimate the number of grains of sand on earth to be close to 70 · 1021 .
Some estimate the number of atoms in the observable universe to be close to 1080 .
Jorge Oyola TSP 15 / 24
TSP Methods

Some simple examples


Transportation problems are difficult to solve. For some, an optimal solution may not be found even with a super fast computer.
There are several strategies to solve such problems, not necessarily to optimality.

Problem with one depot Select a random order to


and 21 cities visit the cities.

Find the nearest neighbor Start from the depot and


for each city and combine select always the nearest
these into a round trip. city from where you are.

Optimal solution found by using more advanced


heuristics.

Jorge Oyola TSP 16 / 24


TSP Methods

Let’s see a solver in action

This problem has 75 customers and a depot.

Jorge Oyola TSP 17 / 24


VRP

The vehicle routing problem

Jorge Oyola

Universidad de Córdoba

Jorge Oyola VRP 18 / 24


VRP

The Vehicle Routing Problem


(VRP)[Dantzig and Ramser, 1959, Toth and Vigo, 2002]

Undirected graph (vertices and edges).


Depot.
Set of customers with non-negative
demand.
Traveling cost from vertex to vertex.
Fleet of vehicles with equal capacity.
Each customer is visited exactly once.

Jorge Oyola VRP 19 / 24


VRP

The Vehicle Routing Problem


(VRP)[Dantzig and Ramser, 1959, Toth and Vigo, 2002]

Undirected graph (vertices and edges).


Depot.
Set of customers with non-negative
demand.
Traveling cost from vertex to vertex.
Fleet of vehicles with equal capacity.
Each customer is visited exactly once.

Purpose: To design low cost delivery routes

Jorge Oyola VRP 19 / 24


VRP

Integer programming model

X X
min cij Xijk (1)
(i,j)∈A k∈K

subject to
X
X0jk ≤ 1, ∀k ∈ K (2)
j∈N
X X
Xijk = 1, ∀i ∈ N \0 (3)
j∈N k∈K
X X
Xijk = Xjik , ∀i ∈ N , k ∈ K (4)
j∈N j∈N
X
V0k = di Xijk , ∀k ∈ K (5)
(i,j)∈A

Vjk ≤ Vik − dj + Q(1 − Xijk ), ∀i ∈ N , j ∈ N \0, k ∈ K (6)


V0k ≤ Q, ∀k ∈ K (7)
Xijk ∈ {0, 1}, ∀i ∈ N , j ∈ N , k ∈ K (8)
Vik ≥ 0, ∀i ∈ N , k ∈ K (9)

Jorge Oyola VRP 20 / 24


VRP

Complexity

Increases when the problem becomes more realistic.

The vehicles can be of different type and size.


Customers may only be visited at certain
times.
Vehicles will both deliver and collect goods
on the same trip.
Goods can be delivered from different depots.
The size of the delivery is uncertain and may
vary from day to day.
Several objectives are optimized.
Vehicles use trailers.
And many others.

Jorge Oyola VRP 21 / 24


VRP Solution methods

Methods commonly used to (attempt to) solve VRPs


[Cordeau et al., 2002, Laporte, 2009]

Branch and cut


Branch and bound
Column generation
Dynamic programming
Constructive heuristics
Improvement heuristics
Local search
Evolutionary algorithms
And others

Jorge Oyola VRP 22 / 24


References

References I

Cordeau, J.-F., Gendreau, M., Laporte, G., Potvin, J.-Y., and Semet,
F. (2002).
A guide to vehicle routing heuristics.
The Journal of the Operational Research Society, 53(5):512–522.
Dantzig, G. B. and Ramser, J. H. (1959).
The truck dispatching problem.
Management Science, 6(1):80–91.
Eiselt, H. A., Gendreau, M., and Laporte, G. (1995a).
Arc routing problems, part I: The chinese postman problem.
Operations Research, 43(2):231–242.
Eiselt, H. A., Gendreau, M., and Laporte, G. (1995b).
Arc routing problems, part II: The rural postman problem.
Operations Research, 43(3):399–414.

Jorge Oyola VRP 23 / 24


References

References II

Gribkovskaia, I., Halskau, Ø., and Laporte, G. (2007).


The bridges of königsberga historical perspective.
Networks, 49(3):199–203.
Laporte, G. (2009).
Fifty years of vehicle routing.
Transportation Science, 43(4):408–416.
Toth, P. and Vigo, D. (2002).
Models, relaxations and exact approaches for the capacitated vehicle
routing problem.
Discrete Applied Mathematics, 123(1-3):487 – 512.

Jorge Oyola VRP 24 / 24

You might also like