Download PDFOpen PDF in browser

Evolutionary Algorithm Outperforms MILP Solver on the Double-Floor Single-Row Facility Layout Problem

EasyChair Preprint 11716

15 pagesDate: January 7, 2024

Abstract

We consider facility planning and design in the single-row facility layout problem (SRFLP) [1] for double-floor spaces to minimize land and material handling costs by efficiently arranging facilities along a given transport side of double-floor spaces. Previously, this problem has been solved based on a mixed-integer linear programming formulation. Our fast (1+1) EA operates by initially constructing a solution that aligns with the characteristics of SRFLP. It then proceeds to determine mutations using a heavy-tailed distribution function. To generate new solutions, we employ local insert and swap operators and incorporate a restart strategy into the process. We demonstrate the practical value of the fast (1+1) EA for DF-SRFLP on numerical experiments (9-80 facility-scale datasets) and real-world instances, including assembly workshops (size 14) and footwear manufacturing workshops (size 54). The experimental results show that under the numerical experiments: the solve time of the MILP solver (GUROBI 10.0.1) grows exponentially with the problem size, while (1+1) EA is relatively stable. And the large-scale, MILP solver is not able to give suitable values in a reasonable time. In a practical application involving a 14-scale assembly line shop, the 1+1 EA achieves a layout that reduces material handling costs by 24 % compared to the original layout form. Furthermore, in the context of a 54-scale shoe factory, our proposed algorithm not only successfully addresses the DF-SRFLP but also offers a valuable reference for establishing an improved layout.

Keyphrases: Facility planning and design, Mixed Integer Linear Programming, double-floor single-row facility layout problem, fast evolution algorithm, real-life case

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@booklet{EasyChair:11716,
  author    = {Junqi Liu and Benjamin Doerr and Zeqiang Zhang and Martin S. Krejca and Berna Ulutas},
  title     = {Evolutionary Algorithm Outperforms MILP Solver on the Double-Floor Single-Row Facility Layout Problem},
  howpublished = {EasyChair Preprint 11716},
  year      = {EasyChair, 2024}}
Download PDFOpen PDF in browser