A GRASP algorithm for solving large-scale single row facility layout problems

•We address large-scale instances of the single row facility layout problem (SRFLP).•A GRASP algorithm with a diversification phase based on frequency memory, called GRASP-f, is proposed.•GRASP-f uses techniques to accelerate the evaluation of the gain of the objective function.•Computational experiments for SRFLP instances with up to 1,000 facilities are reported.•GRASP-f improved the best-known values for 29 out of 93 instances. The single-row facility layout problem (SRFLP) is an NP-hard problem, which consists of arranging facilities along a straight line in order to minimize the weighted sum of distances between all pairs of facilities. In this paper, an implementation of the Greedy Randomized Adaptive Search Procedure (GRASP) meta-heuristic is presented to solve large-scale instances of the problem. The novelty of the proposed implementation is that it is endowed with a diversification phase based on frequency memory. The algorithm was tested on 93 instances with sizes up to one thousand facilities. To date, in the literature, there is only one work, which has considered instances of such magnitude. Computational experiments demonstrate the efficiency of the proposed algorithm, which improved the best-known values for 29 out of 93 instances.