Multiple time window VRP dataset

Explanation of the Multi-Time Windows Dataset for the Vehicle Routing Problem with Time Windows Dataset

Introduction

This document provides an overview of the newly created dataset that extends the classic Solomon benchmark instances for the Vehicle Routing Problem with Time Windows (VRPTW). The dataset introduces multiple time windows for each customer, enhancing the realism and applicability of routing algorithms in logistics and supply chain management.

Dataset Structure

The dataset consists of six CSV files, each corresponding to a specific problem type from the original Solomon datasets. The first (most) preferred time window is READY_TIME_1 - DUE_TIME_1. The second preferred time window is READY_TIME_2 - DUE_TIME_2. The third (least) preferred time window is READY_TIME_3 - DUE_TIME_3. Each file contains data for 100 customers, including the depot, and includes the following columns:

CUST_NO: Customer ID (0 represents the depot)

XCOORD: X-coordinate of the customer's location

YCOORD: Y-coordinate of the customer's location

DEMAND: The demand of the customer (in units)

READY_TIME_1: Start time of the first time window

DUE_TIME_1: End time of the first time window

READY_TIME_2: Start time of the second time window

DUE_TIME_2: End time of the second time window

READY_TIME_3: Start time of the third time window

DUE_TIME_3: End time of the third time window

Example of the New Dataset

The following table illustrates a sample of the new dataset, showing the first six customers, including the depot:

CUST_NO

XCOORD

YCOORD

DEMAND

READY_TIME_1

DUE_TIME_1

READY_TIME_2

DUE_TIME_2

READY_TIME_3

DUE_TIME_3

0

40

50

0

0

960

0

960

0

960

1

45

68

10

690

750

570

630

900

960

2

45

70

30

750

810

480

540

570

630

3

42

66

10

780

840

540

600

720

780

4

42

68

10

480

540

810

870

750

810

5

42

65

10

690

750

810

870

480

540

Original Solomon Dataset Analysis

The following table summarizes the analysis of the original Solomon datasets, highlighting key metrics for each problem type:

Problem Type

Count

Mean Ready Time

Mean Due Time

Mean Duration

Min Duration

Max Duration

C1

900

316.33

637.32

320.99

29

1136

C2

800

1114.85

2035.36

920.51

160

3291

R1

1200

59.46

146.42

86.95

10

215

R2

1100

241.34

695.08

453.73

27

985

RC1

800

64.26

149.69

85.42

10

225

RC2

800

258.90

628.66

369.75

60

945