On the interaction between Autonomous Mobility-on-Demand systems and the power network: models and coordination algorithms

We study the interaction between a fleet of electric, self-driving vehicles\nservicing on-demand transportation requests (referred to as Autonomous\nMobility-on-Demand, or AMoD, system) and the electric power network. We propose\na model that captures the coupling between the two systems stemming from the\nvehicles' charging requirements and captures time-varying customer demand and\npower generation costs, road congestion, battery depreciation, and power\ntransmission and distribution constraints. We then leverage the model to\njointly optimize the operation of both systems. We devise an algorithmic\nprocedure to losslessly reduce the problem size by bundling customer requests,\nallowing it to be efficiently solved by off-the-shelf linear programming\nsolvers. Next, we show that the socially optimal solution to the joint problem\ncan be enforced as a general equilibrium, and we provide a dual decomposition\nalgorithm that allows self-interested agents to compute the market clearing\nprices without sharing private information. We assess the performance of the\nmode by studying a hypothetical AMoD system in Dallas-Fort Worth and its impact\non the Texas power network. Lack of coordination between the AMoD system and\nthe power network can cause a 4.4% increase in the price of electricity in\nDallas-Fort Worth; conversely, coordination between the AMoD system and the\npower network could reduce electricity expenditure compared to the case where\nno cars are present (despite the increased demand for electricity) and yield\nsavings of up $147M/year. Finally, we provide a receding-horizon implementation\nand assess its performance with agent-based simulations. Collectively, the\nresults of this paper provide a first-of-a-kind characterization of the\ninteraction between electric-powered AMoD systems and the power network, and\nshed additional light on the economic and societal value of AMoD.\n