Adaptive iterated local search for the parallel row ordering problem

This study addresses the parallel row ordering problem (PROP), which aims at arranging a number of facilities along two parallel rows in a plant. The PROP assumes prior assignment of facilities, i.e. some specific facilities should be allocated to one row, and the remaining facilities to the other row. The objective of the problem is to determine the ordering of the facilities on the rows so as to minimize a cost function. The problem has several applications (e.g. flexible manufacturing systems design) and is NP-Hard. We present a solution approach based on Iterated Local Search (ILS) that departs from its standard implementation. The proposed ILS algorithm is adaptive, alternating periods of intensification and diversification. Moreover, the ILS algorithm utilizes a technique to speed up the calculation of the problem objective function. The proposed algorithm improves on the best-known values for 34 out of 100 instances in the literature; and reaches the best-known value of the remaining 66 instances. While no methods in the literature have undertaken instances with more than 70 facilities, the proposed algorithm can efficiently handle much larger instances containing between 250 and 300 facilities. • An Adaptive Iterated Local Search (AILS) algorithm is proposed for the PROP. • AILS auto-regulates intensification, diversification and intensity of perturbation. • A technique was used to speed up the calculation of neighborhood move gains. • AILS improves on previous best solutions for 34 out of 100 tested instances. • The AILS algorithm can efficiently handle problems with up to 300 facilities.