Mathprog-ORlib

Commonly used data sets for hub location problems include:

The Civil Aeronautics Board (CAB) instances are due to Prof. Morton E. O’Kelly and are mirrored here for convenience (see below)
The Australia Post (AP) instances (see below)
The Turkish Network (TR) instances are available from Prof Bahar Yetis Kara at this website
Less commonly used but much larger, the USA423 data set by J. Peiró, A. Corberán A. and R. Martí (see UrApHMP). For convenience this instance has been mirrored here

Australia Post (AP) dataset for p-hub median and capacitated hub location problems ==================================================================================

File: APdata.zip

The Australia Post data set was first published in:

Ernst, Andreas T., and Mohan Krishnamoorthy. “Efficient algorithms for the uncapacitated single allocation p-hub median problem” Location Science 4, no. 3 (1996): 139-154.
Ernst, Andreas T., and Mohan Krishnamoorthy. “Solution algorithms for the capacitated single allocation hub location problem” Annals of Operations Research 86 (1999): 141-159. (With capacities & fixed costs)

This download contains the AP (Australia Post) data set for p-hub median and capacitated hub location problems. The files contain the following:

APdata200 - Data file for a full 200 node problem with 8 hubs
generate.c - C program for generating smaller data sets

USAGE: generate n p newdata

This creates a new problem with n nodes and p hubs
20.3 - A sample data set with 20 nodes produced by

generate 20 3 20
FcostX.NN - Fixed cost file for NN nodes where X = 'T' or 'L' (the T problems tend to be more difficult). Fcost files are not relevant to the p-hub location problems
CapY.NN - Node capacity file for NN nodes where Y = 'T' or 'L' (the T problems are more tightly constraint) . Capacity files are not relevant to the p-hub location problems
Solutions-*.txt - Optimal solutions for all combinations of capacitated (CSAHLP) and uncapacitated single allocation (USApHMP) and multiple allocation (UMApHMP) hub median problems, with n in {10,20,25,40,50} and multiple options for hubs or capacities. In the result files we use the nomenclature “NNXY” to refer to the problem generatedfrom data files NN, FcostX.NN and CapY.NN. The objective is given as well as the allocation of hubs. Eg an allocation vector 2,2,4,4,4 means that nodes 2 & 4 are hubs, node 1 is allocated to hub node 2 and nodes 3 & 5 to hub node 4.

Data file format for nodes file:

<n>                         Number of nodes
<x[1]> <y[1]>               x & y coordinates of node 1
:
:
<x[n]> <y[n]>               x & y coordinates of node n
<w[1][1]> <w[1][2]> ... <w[1][n]>    flow from node 1 to all others
: : :
: : :
<w[n][1]> <w[n][2]> ... <w[n][n]>    flow from node n to all others
<p>        Number of hubs (for p-hub median problems)
<δ>        Collection cost
<α>        Transfer cost
<χ>        Distribution cost

All of the costs are per unit (euclidean) distance, per unit flow volume divided by 1000.

The costs and and capacity files contain one number for each node (in the same order as in the nodes file). For FcostX.NN this represents the cost of making the node a hub. For CapY.NN the numbers are the capacity on incoming commodities (including from the node itself) if that node is made a hub.

Civil Aeronautics Board (CAB) Hub Location Instances ==================================================================================

This is really just a 25 node instance with smaller instances created by taking the first n nodes (typically for n=10,15 & 20) and by varrying the transfer cost (typically 0.2 to 1.0 in steps of 0.2). These instances have been contributed to the research community by M.E. O’Kelly.

This is the 25 node data set that has been used extensively with hub and spoke location models. It is sometimes referred to as the CAB data set. Full references to the source and prior results for these data can be found in several previous paper, including

O’Kelly, Morton E. “A quadratic integer program for the location of interacting hub facilities.” European Journal of Operational Research 32, no. 3 (1987): 393-404.
O’Kelly, Morton E., Deborah Bryan, Darko Skorin-Kapov, and Jadranka Skorin-Kapov. “Hub network design with single and multiple allocation: A computational study” Location Science 4, no. 3 (1996): 125-138.

This data is provided in two formats

25.3.4 - same format as the AP data sets
CAB25.txt - number of nodes, n x n matrix of flow (w), n x n matrix of distances

File: CABdata.zip