In order to explore the operation rules of the mixed traffic flow consisting of human-driven vehicles (HDVs) and connected and autonomous vehicles (CAVs), this paper studies the impact of different managed lane setting strategies on the traffic flow of urban expressways. First, on the basis of the coupling relationship between car-following and lane selection probability of different types of vehicles, the evolution mechanism of expressway capacity was quantitatively described under different managed lane setting strategies. Then, the variation pattern of vehicle delays was analyzed using the SUMO simulation platform under mixed traffic flow conditions. Finally, the influence of different managed lane setting strategies on the collision risk of vehicles was investigated by analyzing the vehicle lane changing types and gaps. Results showed that the capacity reached the maximum value when the CAV penetration rate was less than 30% or more than 80% and HDVs were restricted to general purpose lanes. Bus and CAV lanes should be considered when the CAV penetration rate was between 30% and 80%. The average delay on the roadway was the lowest and almost unrelated to the CAV penetration rate when bus and CAV lanes were designed but their right-of-way was not restricted. The risk of vehicle collisions increased when managed lanes were provided only for CAVs or high-occupancy vehicles (HOVs). These results indicate that CAV penetration rate is an important factor for development of a reasonable managed lane setting strategy, and has significant influence on the capacity of the road section with managed lanes. The average delay of road sections and the collision risk between vehicles in the traffic flow are mainly influenced by the managed lane setting strategy.