This page contains a collection of benchmark instances for the Uncapacitated
Facility Location Problem (UFLP), also known as Simple Plant
Location Problem (SPLP) or the (Uncapacitated) Warehouse Location Problem. This problem
from Operational Research has been studied intensively, and
here a collection of standard benchmark problems can be
sources and binaries of benchmark generators for different operating systems and
code for several solution algorithms are available.
If you find any errors or like to comment on this page, or if you have additional benchmarks or code material, please write an e-mail to Martin Hoefer.
|13.07.2007||Optimal solutions were added for 16 of 18 instances in the K-median package, only 2500-10 and 3000-100 remain unsolved. For the symmetric and asymmetric Koerkel-Ghosh packages the instances gs250a-1 and ga250a-1 have been solved. Thanks to Cesar Beltran-Royo, Jean-Philippe Vial, and Antonio Alonso-Ayuso for providing the solution files. This paper discusses a solution technique and some additional results.|
The evaluation is completed. Improved solutions exist for
5FPP_11S.txt - opt: 36226 - prev. reported: 36228
23FPP_17S.txt - opt: 54557 - prev. reported: 54558
1331GapBS.txt - opt: 42169 - prev. reported: 42171
For all instances not solved to optimality .bub-files were included that contain an upper bound on the optimum solution.
|06.08.2004||Some of the .opt-solutions could slightly be improved. Thanks to Pascal van Hentenryck for bringing this to my attention. I started to re-evaluate all the solutions provided on this page.|
|21.10.2003||New: Chessboard, Eukldian, Perfect Code, Uniform and Koerkel-Ghosh benchmarks|
Finally all optimal solutions of the Large duality gap and Finite projective planes
benchmarks are included in the packages.
New: Large duality gap and finite projective planes benchmarks
More information and generators can be found on Yuri Kochetov's web page.
For everyone, who does not know what we are talking about, here's a tiny
The uncapacitated facility location problem can be stated as follows:
There are a number of