“…Considering capacity constraints only (i.e., taking D = ∞), Sariklis and Powell [154] propose a two-phase heuristic which first assigns customers to clusters and then builds a Hamiltonian path for each cluster, Tarantilis et al [163] describe a population-based heuristic, while Tarantilis et al [164,165] present threshold accepting metaheuristics. Taking into account both capacity and distance constraints, Brandão [40], Fu et al [90,91] and Derigs and Reuter [69] propose tabu search heuristics, Li et al [115] describe a record-to-record travel heuristic, Pisinger and Ropke [141] present an adaptive large neighborhood search heuristic which follows a destroy-and-repair paradigm, while Fleszar et al [88] propose a variable neighborhood search heuristic.…”