Computing and Analyzing Mixed Equilibrium Network Flows with Gasoline and Electric Vehicles

This article addresses a new network equilibrium problem with mode and route choices for the emerging need of modeling regional transportation networks that accommodate both gasoline and electric vehicles. The two transportation modes (or vehicle types) distinguish from each other in terms of driving distance limit and travel cost composition. In view of the advantages (e.g., low fuel expenses and vehicle emissions) and disadvantages (e.g., limited driving range and long charging time) pertaining to driving electric vehicles, it is anticipated that a large number of households/motorists may prefer to own both gasoline and electric vehicles (although, of course, many households/motorists still only own gasoline vehicles (GVs) and some households may choose to own electric vehicles only) in the transition period from the petroleum era to the electricity era. The purpose of this article is to offer a traffic equilibrium modeling tool for networks that serve households/motorists who can choose between gasoline and electric vehicles. Specifically, we present a convex optimization model for characterizing such mixed equilibrium traffic networks with both gasoline and electric vehicles, which are expected to exist for a long period in the future. Two competing solution algorithms, a linear approximation algorithm of the Jacobi type and a quadratic approximation algorithm taking the form of the Gauss–Seidel decomposition, are implemented and evaluated. Experimental results clearly show that, from the model behavior perspective, the pro∗To whom correspondence should be addressed. E-mail: chi.xie@ sjtu.edu.cn. duced network flow patterns replicate the anticipated combined mode–route choice results, that is, the higher the distance limit or the gasoline price is, the more travelers choose battery electric vehicles (BEVs) when both BEVs and GVs are available to them; and, from the solution efficiency perspective, the quadratic approximation algorithm exhibits linear convergence and can reach higher solution precision in shorter time.