Route Assignment in Transportation Networks. A Сomparison of User Equilibrium and System Optimum

One of the major problems of big cities is traffic congestion. For the purpose of avoiding congestion travelers use traffic-aware Advanced Traveler Information Systems (ATIS), which rely on historical or current traffic data to compute fastest routes. However, the existing solutions are prone to errors associated with the poor prediction of future traffic states and may even produce traffic jams due to uncoordinated decision mechanism which directs all drivers with identical origin-destination pairs departing at the same time through the same route. In this study the recently proposed Predictive Traffic Assignment (PTA) algorithm which exploits previous planned routes to predict their impact on future traffic states is further studied. The results obtained by simulations show that if the fraction of PTA users is sufficiently large it can outperform existing solutions in congested networks, while being unable to beat contemporary navigation systems when market penetration is less than 50%. More importantly, the feasibility of PTA deployment was tested with the conclusion that while PTA introduction to transportation system is beneficial in terms of total travel time it lacks profitability, due to the low equilibrium fraction of its users, to motivate private sector to deploy it.