Human-piloted, remotely piloted, and autonomous aircraft are likely to share the air space in the near future. Furthermore, the density of air traffic is likely to increase as urban air mobility and air package delivery operations become more prevalent. Hence an approach to collision avoidance that is both scalable and permits integration of human pilots is needed. This paper discusses a collision avoidance approach based on velocity potentials: it is scalable to large numbers of aircraft and is highly likely to be easily interpretable for human pilots, passengers, and bystanders. This paper describes the velocity potential approach for three dimensional collision avoidance and uses Monte Carlo simulations to examine (1) the effect of number of aircraft in the airspace; (2) the effect of number of aircraft that are actively tracked for collision avoidance; (3) the elementary potentials used for collision avoidance.