Optimal Multi-Commodity Network Flow of Electric Vehicles with Charge Constraints

Abstract: The focus of this thesis is to find, visualize and analyze the optimal flow of autonomous electric vehicles with charge constraints in urban traffic with respect to energy consumption. The traffic has been formulated as a static multi-commodity network flow problem, for which two different models have been implemented to handle the charge constraints. The first model uses a recursive algorithm to find the optimal solution fulfilling the charge constraints, while the second model discretizes the commodities’ battery to predetermined battery levels. An implementation of both methods is provided through simulations on scenarios of three different sizes. The results show that both methods are capable of representing the traffic flow with charge constraints, with limitations given by the size of the problem. In particular, the recursive model has the advantage of considering the charge as a continuous quantity. On the other hand the discretization of battery levels allows to handle charge constraint setups with higher complexity, that is when longer detours are needed to fulfill the charge constraints.