In this paper we qualitatively and numerically compare basic modeling
approaches to the distance-based facility layout problem and the types
of result obtained from them. In the process we observe that when the
facility layout problem is modeled as a QAP, one will not necessarily
obtain an optimal solution to the layout problem even if the QAP is s
olved to optimality. We also propose an alternative distance measure,
namely, EDIST, which is based on the expected distance between two dep
artments. We illustrate how the EDIST measure addresses some of the we
ll-known shortcomings associated with the centroid-to-centroid distanc
e measure.