Nowadays, public bike-sharing systems are broadly adopted and deployed in many major cities, however, as public facilities, bicycles will be prone to damage and need to be replaced frequently, which results in high system maintenance costs. One of the root causes of bicycle damages is the serious load-unbalance of bicycle usage. In this paper, we propose a hybrid bicycle allocation strategy for bicycle lifetime optimization, which can effectively reduce the degree of imbalance of system load. First, we analyze and verify the load-unbalance status of bicycle usage in the existing bike-sharing system. Then, a hybrid bicycle allocation strategy is proposed, which is evaluated on Washington D.C. bike-sharing system. Furthermore, according to a bicycle lifetime model based on Weibull distribution, the proposed bicycle allocation strategy could significantly cut down the percentage of the bicycles need to be replaced in a certain period of time.