This paper deals with the problem of allocating berth positions for vessels in export bulk port terminals considering tidal constraints and was first formulated by Ernst et al., (2017). This study investigates the dynamic and continuous berth allocation problem (BAP) with respect to tidal constraints (BAP_TC), and seeks to minimize the total service time of berthed vessels. Since the BAP problem is NP-hard the BAP_TC is also NP-hard. A reduced variable neighborhood search (RVNS) based approach is developed to solve the problem. For parameters tuning a machine learning algorithm is developed and used. Problem instances are benchmarked with CPLEX and the numerical experiments proved that the proposed algorithm is capable of generating high-quality solutions in rather short time. Both small and large-scale instances in the literature are tested to evaluate the metaheuristic effectiveness using other solution approaches from the literature. The computational experiment proves that the proposed algorithm provides state of the art results.