The Open Vehicle Routing Problem (OVRP) is a version of the well known Capacitated Vehicle Routing Problem (CVRP), in which each route ends at the last served customer. The objective of the OVRP is to deliver a set of customers with known demands on minimum-cost vehicle routes originating and at a depot. Each vehicle has a maximum capacity. We study the following variants of this problem: the number of vehicles is unlimited or not and the demand of each customer is equals to one. This last version originates in the transportation of handicapped people which is the final goal of our study. For addressing these problems, we propose an Iterated Local Search combined with a Tabu Search on multiple neighborhoods. The neighborhoods are generated from Ejection Chains and Path Exchange moves. The Path Exchange operator removes two non consecutive paths and reinserts each path at the other position. It can be used as intra or inter route exchange operator. Our ejection chains are produced from performing infeasible moves and then repairing the incumbent solution. Experimental tests show that our method is competitive with the most advances solutions techniques in the literature and suitable for the case of the demand equal to one.