Abstract
As delivery technologies have advanced, novel last mile delivery (LMD) channels have been emerged in recent years. Among these channels, parcel lockers stand out, as they reduce the number of failed delivery attempts in addition to offering cost-effective and efficient delivery experience. Location selection is a key concept for establishing a parcel locker network, as it directly influences ability of the demand coverage, required initial investment and compliance to any additional constraints. For this reason, the location allocation problem for parcel lockers should be addressed with a systematic and quantitative methodology. This study focuses on parcel locker location selection for Beşiktaş, a district of İstanbul, Türkiye which is densely populated and presents highly challenging urban conditions. Within this context, parcel locker allocation problem is solved with a modified p-median model, while the demand points are obtained from geographical information systems and the candidate service points are identified through k -means clustering algorithm. The results obtained provide a fundamental basis for establishing a parcel locker system in Beşiktaş. Moreover, traditional capacitated p-median model is enhanced with additional constraints to satisfy the demand coverage, providing a more effective initial solution than the traditional model.