This paper addresses an assembly line balancing problem in which the length of the workpieces is larger than the width of the workstations. The problem differs from traditional variants of assembly line balancing in the sense that only a portion of the workpiece, or portions of two consecutive workpieces, can be reached from any workstation. Consequently, at any stationary stage of the cycle, each workstation can only process a portion of the tasks, namely, those which are inside the area of a workpiece that is reachable from the workstation. The objective is to find a (cyclic) movement scheme of the workpieces along the line and a task assignment to stationary stages of the production process, while minimizing the cycle time. We propose three hybrid approaches of metaheuristics and mathematical programming - one based on simulated annealing and the other two based on tabu search, relying on different neighborhood definitions. The two former approaches make use of a classical neighborhood, obtained by applying local changes to a current solution. The latter approach, in contrast, draws ideas from the corridor method to define a corridor around the current solution, via the imposition of exogenous constraints on the solution space of the problem. An extensive computational experiment is carried out to test the performance of the proposed approaches, improving the best results published to date.