Traffic congestion is a universal challenge that affects urban transportation networks, which inevitably age and deteriorate. Maintenance is an essential method for alleviating road congestion. Most of the previous studies concentrate on node load and capacity analysis. The capacity of an idle node is also an important element that affects traffic congestion, such as road damage or traffic accident at the crossroads. To explore the effect of the capacity of an idle node on road congestion, this paper introduces a traffic transfer principle to improve road maintenance efficiency. The nodes of a traffic network can be ranked based on their failure severity. The failure paths of a traffic network can be identified through the internal connections between nodes. Using the transfer time as the weight of each edge and the service time as the weight of each node, this paper proposes a maintenance model to find the shortest repair path for minimizing the maintenance time. To evaluate the proposed model, four different types of road networks are adopted by comparing the maintenance time. The experimental results show the proposed model outperforms previous models.