Article ID: | iaor20023103 |
Country: | United Kingdom |
Volume: | 29 |
Issue: | 5 |
Start Page Number: | 489 |
End Page Number: | 504 |
Publication Date: | Apr 2002 |
Journal: | Computers and Operations Research |
Authors: | Chou Shuo-Yan, Chen Yi-Kuang, Lin Shih-Wei |
Keywords: | multidimensional scaling |
Multidimensional scaling (MDS) is a useful mathematical tool that enables the analysis of data in areas where organized concepts and underlying dimensions are not well developed. In this paper, MDS algorithms are used as a dimension reduction tool which arranges facilities in a two-dimensional space while preserving the adjacency relationship between facilities. The output of MDS is a scatter diagram and is in turn used as the input or location references for developing into the final block layout. The bay structures of layout are considered where the given floor space is first partitioned horizontally or vertically into bays, which are subsequently partitioned into the blocks. Rotating the scatter diagram about the origin results in different layouts in the bay structure. A simulated annealing approach is adopted to rotate the scatter diagram so that the total cost of traveling between facilities and shape violation in the final layout is minimized.