Electric vehicle (EV) adoption in long-distance logistics faces challenges such as range anxiety and uneven distribution of charging stations. Two pivotal questions emerge: How can EVs be efficiently routed in a charging network considering range limits, charging speeds and prices? And, can the existing charging infrastructure sustain the increasing demand for EVs in long-distance logistics? This paper addresses these questions by introducing a novel theoretical and computational framework to study the EV network flow problems. We present an EV network flow model that incorporates range constraints and nonlinear charging rates, and identify conditions under which polynomial-time solutions can be obtained for optimal single EV routing, maximum flow, and minimum-cost flow problems. Our findings provide insights for optimizing EV routing in logistics, ensuring an efficient and sustainable future.

Shared rides, which pool individual riders into a single vehicle, are essential for mitigating congestion and promoting more sustainable urban transportation. However, major ridesharing platforms have long struggled to maintain a healthy and profitable shared rides product. To understand why shared rides have struggled, we analyze procedures commonly used in practice to set static prices for shared rides, and discuss their pitfalls. We then propose a pricing policy that is adaptive to matching outcomes, dubbed match-based pricing, which varies prices depending on whether a rider is dispatched alone or to what extent she is matched with another rider. Analysis on a single origin-destination setting reveals that match-based pricing is both profit-maximizing and altruistic, simultaneously improving cost efficiency (i.e., the fraction of cost saved by shared rides relative to individual rides) and reducing rider payments relative to the optimal static pricing policy. These theoretical results are validated on a large-scale simulation with hundreds of origin-destinations from Chicago ridesharing data. The improvements in efficiency and reductions in payments are especially noticeable when costs are high and demand density is low, enabling healthy operations where they have historically been most challenging.

Increasingly, data-driven and robust optimization approaches are being leveraged to optimize short and long-term planning decisions in energy systems under uncertain supply and demand. In particular, effective decision-making in future decarbonized energy systems necessitates modeling supply-side uncertainties, such as solar and wind availability, at a high spatio-temporal resolution. To this end, risk-averse decision makers require high-fidelity models that accurately characterize tradeoffs between nominal and worst-case operational outcomes and consequently yield decisions that are simultaneously low-cost and robust to future uncertainties. However, existing approaches for two-stage optimization in energy applications trade off realism for tractability in representing both second-stage uncertainties and decision-making, potentially overestimating incurred costs and tipping the balance of planning towards undue risk aversion. Here, we propose a deep learning-based approach to risk-averse two-stage optimization that aims to reduce this conservative bias. Specifically, we propose utilizing a variational autoencoder with a monotonic decoder trained to generate realistic high-dimensional realizations of energy supply and demand that enter as uncertain parameters in the second stage. We then solve the risk-averse two-stage model using a constraint generation approach whose subproblem is solved in each iteration by maximizing second-stage costs - a function of generated supply-demand data - with respect to the autoencoder latent variable. Finally, we demonstrate the efficacy of our approach in a case study that considers a zonal representation of the New England generation and transmission system and compare against benchmark optimization approaches.

In this paper, we study platforms where resources and jobs are spatially distributed, and resources have the flexibility to strategically move to different locations for better payoffs. The price of the service at each location depends on the number of resources present and the market size, which is modeled as a random state. Our focus is on how the platform can utilize information about the underlying state to influence resource repositioning decisions and ultimately increase commission revenues. We establish that in many practically relevant settings a simple monotone partitional information disclosure policy is optimal. This policy reveals state realizations below a threshold and above a second (higher) threshold, and pools all states in between and maps them to a unique signal realization. We also provide algorithmic approaches for obtaining (near-)optimal information structures that are monotone partitional in general settings.