Source-linked AI summary
Charging Electric Vehicles in the Smart City: A Survey of Economy-driven Approaches
Wenjing Shuai, Patrick Maillé, Alexander Pelov
TL;DR
EV charging management must account for users’ differing economic interests and the challenge of analyzing self-interested actors in complex settings. This paper surveys and compares economy-driven schemes; it finds that, although some mechanisms perform well in simulations, analytical results remain limited, motivating further work on design principles.
Problem
User economic interests and preferences vary across surveyed models, while self-interested actors make dynamic energy-allocation problems difficult to analyze.
Source: 1) Dealing with uncertainties about… · p. 9, §VI · p. 15Method
The paper surveys charging-management schemes built with optimization and/or game-theory tools, spanning simple energy sharing to settings with uncertainty, mobility, station locations, and grid services.
Source: §VII · p. 15Results
Some proposed mechanisms perform well in simulation scenarios, but analytical results are limited as settings and economic constraints grow more complex.
Source: §VII · p. 16Takeaways & Limitations
Further research is needed to better understand the key principles for designing charging-management schemes.
Source: §VII · p. 16Analytical results are limited by complex settings involving many actors, distribution-network and battery constraints, and nonaligned economic objectives.
Source: §VII · p. 16
Abstract
from arXiv · showhide
Electric Vehicles (EVs), as their penetration increases, are not only challenging the sustainability of the power grid, but also stimulating and promoting its upgrading. Indeed, EVs can actively reinforce the development of the Smart Grid if their charging processes are properly coordinated through two-way communications, possibly benefiting all types of actors. Because grid systems involve a large number of actors with nonaligned objectives, we focus on the economic and incentive aspects, where each actor behaves in its own interest. We indeed believe that the market structure will directly impact the actors' behaviors, and as a result the total benefits that the presence of EVs can earn the society, hence the need for a careful design. This survey provides an overview of economic models considering unidirectional energy flows, but also bidirectional energy flows, i.e., with EVs temporarily providing energy to the grid. We describe and compare the main approaches, summarize the requirements on the supporting communication systems, and propose a classification to highlight the most important results and lacks.
I. INTRODUCTION
EV charging can burden the grid, but flexible charging and batteries can also support the Smart Grid. The paper surveys economic charging schemes for self-interested actors and tracks the communication systems they require.
I. INTRODUCTION
- At 25% EV penetration, almost 30% of network facilities would need upgrading, compared with 5% if charging is shifted to less crowded periods.Uncontrolled charging also threatens distribution networks and can raise electricity prices and CO2 emissions,,,.Source: §I · p. 1
- EVs offer flexible demand and batteries that can support the grid, but owners may reschedule charging only when offered sufficient incentives.Incentives can include fixed rewards, energy auctions, or time-varying prices.Source: §I · pp. 1–2
- Because grid actors have nonaligned objectives, the paper frames EV charging management through market mechanisms and game theory,.Source: §I · p. 2
- The survey classifies economy-driven charging schemes, compares their assumptions and results, and tracks communication requirements such as exchange volume and frequency.It covers unidirectional and bidirectional energy flows, as well as communication aspects and research limitations.Source: §I · p. 2
A. Facilities for Electric Vehicle Charging
The paper narrows its EV scope to passenger cars recharged through a physical plug, then relates EV flexibility and storage to Smart Grid services and market actors.
A. Facilities for Electric Vehicle Charging
- The survey uses “Electric Vehicle” for a passenger car with a battery refilled from external electricity sources, and generally assumes physical plug charging.It excludes roadbed infrastructure, refueling, and battery-swapping solutions because they overlap with conventional refueling or remain experimental.Source: A. Facilities for Electric Vehicle… · pp. 2–3
- Charging facilities differ in power limits, battery capacity, and AC/DC conversion efficiency; SAE J1772 defines two AC and three DC charging levels.The referenced table estimates full-recharge duration for a 25 kWh usable battery starting at 20% state of charge.Source: A. Facilities for Electric Vehicle… · p. 2, Table I · p. 2
B. Electric Vehicles – An enabler of the Smart Grid and a participant in Electricity markets
- EVs can modulate consumption in response to prices or direct control and, in some cases, inject electricity back as distributed energy sources.These capabilities can support renewable integration, self-healing, and ancillary services.Source: B. Electric Vehicles – An enabler of… · p. 3
C. Dealing with self-interested actors
EV charging schemes must account for actors who pursue different objectives and may strategically report private information. The section introduces incentive compatibility, game-theoretic equilibrium, and welfare vocabulary for analyzing those interactions.
C. Dealing with self-interested actors
- EV owners, aggregators, charging stations, and system operators have distinct interests, so allocation and pricing rules must anticipate strategic behavior.Source: C. Dealing with self-interested actors · p. 4
- Incentive compatibility means that actors gain no advantage by reporting untruthful values when decisions use signals such as willingness-to-pay.Source: C. Dealing with self-interested actors · p. 4
- A Nash equilibrium is an outcome where no actor can improve its payoff through a unilateral move, and many surveyed mechanisms rely on this stability concept.Such equilibria can sometimes be reached or approached with limited information exchanges or by learning through trial decisions.Source: C. Dealing with self-interested actors · p. 4
- Social welfare sums user and supplier welfare as users’ valuations minus system costs, excluding money transfers that remain within society.Source: C. Dealing with self-interested actors · p. 4
III. UNIDIRECTIONAL CHARGING MECHANISMS
Unidirectional charging schemes allocate scarce grid energy among EVs under user, supply, and network constraints. The survey compares economic models for sharing energy in a single interval, including their objectives and allocation outcomes.
III. UNIDIRECTIONAL CHARGING MECHANISMS
- The section considers a snapshot with insufficient available energy, allocating the scarce resource among EVs; topology-based models add network capacity constraints to the allocation problem.The approaches also differ in whether supply is fixed or chosen by the aggregator, and whether the objective is user welfare, social welfare, fairness, or revenue.Source: §III · p. 4, 1) Sharing energy efficiently among… · pp. 5–6
1) Sharing energy efficiently among users:
- [25] allocates constrained hub supply using users’ reported willingness to pay, but assumes truthful reporting; instead lets the aggregator choose supply to maximize social welfare.[27] uses iterative price and load updates that converge to an allocation.Source: 1) Sharing energy efficiently among… · p. 5
- [28] models the aggregator as a revenue-maximizing leader setting prices and EVs as followers choosing demands, while uses a VCG-style auction to incentivize truthful bids and maximize user welfare.[29] limits the number of bids per EV as a trade-off between efficiency and communication and computational complexity.Source: 1) Sharing energy efficiently among… · p. 5
- For tree networks, [30] uses link congestion prices to achieve proportional fairness, though users may initially demand more than link capacities allow before convergence.Its users are assumed to have logarithmic utility and lack awareness of link capacity limits.Source: 1) Sharing energy efficiently among… · p. 6
- In the toy example, welfare- and revenue-oriented approaches can yield the same efficient allocations, but VCG-based schemes charge users less than the market price to support truthful valuation revelation.The comparison includes,,, and.Source: 1) Sharing energy efficiently among… · p. 6
2) Electricity sharing over several time slots:
Multi-slot schemes use EV charging flexibility to reshape aggregate load under energy constraints. The survey covers load flattening, arbitrary target profiles, and the equilibrium and incentive conditions behind these approaches.
a) Flat charging curves:
- In, users choose uninterrupted constant-rate charging start times; the game has pure Nash equilibria, and the equilibrium is unique and optimal with infinitely many cars.The charging cost is transferred to users through prices.Source: 2) Electricity sharing over several… · p. 6
- For multiple independent appliances, variable charging rates meeting each client’s demand yield sequential cost-minimizing schedules that converge to a unique equilibrium minimizing total charging cost; its peak-to-average ratio is 0.05% above the lowest achievable.Source: 2) Electricity sharing over several… · p. 7
- [46] schedules arriving EVs sequentially using a forecast 24-hour load profile, but assumes altruistic owners; the authors identify selfish behavior and price design as an open incentive problem.The approach requires bidirectional communication between each EV and the aggregator.Source: 2) Electricity sharing over several… · p. 7
b) Following an arbitrary curve:
- A consumption-dependent price and deviation penalty steer EV charging toward valley hours and yield a Nash equilibrium that is socially optimal with identical deadlines; extends convergence to different deadlines and arbitrary aggregate-load profiles.Following a predefined overall demand curve can account for external un-shiftable loads.Source: 2) Electricity sharing over several… · p. 7
- Other approaches use hourly prices and centralized scheduling to minimize charging costs while respecting distribution constraints and EV energy demands.[36] uses real-time prices and, use predicted prices and congestion prices set by the ISO.Source: 2) Electricity sharing over several… · p. 7
3) Summary of Static unidirectional recharging:
Static unidirectional models differ in economic objective and constraints, while common assumptions leave demand flexibility and convergence as open concerns.
3) Summary of Static unidirectional recharging:
- Static approaches include aggregator-led Stackelberg pricing, topology-constrained sharing through congestion prices, and auctions that dispatch energy subject to network limits.The surveyed objectives include user and social welfare, proportional fairness, and efficient allocation among users with private utilities.Source: 3) Summary of Static unidirectional… · pp. 7–8
- Multi-slot schemes generally treat EV demand from the grid as fixed, although EVs may have alternative energy sources and thus flexible demand in both timing and volume.The fixed-demand assumption simplifies the aggregator’s load-curve choice and makes the optimal load curve unique and computable.Source: 3) Summary of Static unidirectional… · p. 8
- In atomic charging games, stable-rate battery preferences hinder convergence to the global optimum, which is only achievable with infinitely many EVs; penalties can ensure convergence but affect utilities.The passage identifies this as an inherent difficulty in distributed systems.Source: 3) Summary of Static unidirectional… · p. 8
B. Dynamic models
Dynamic charging decisions arise when actors commit before future prices, users, or demand information is known. The survey contrasts repeated static decisions with adaptive methods and discusses the trade-offs of truthfulness and user-preference evidence.
1) Dealing with uncertainties about future events:
- Repeatedly applying static algorithms can be optimal when time slots are independent, but generally requires dynamic algorithms when decisions interact across slots.The survey gives an example involving users with different plug-in durations and energy valuations.Source: 1) Dealing with uncertainties about… · p. 8
- Relaunching a myopic allocation when new information arrives can outperform per-slot static allocation without updates, but does not guarantee the best decision using available future-event probabilities.In the example, later arrivals or changes in available energy trigger further allocation updates.Source: 1) Dealing with uncertainties about… · pp. 8–9
B. Dynamic models
- Dynamic settings involve decisions over time when future information is unavailable, including uncertainty about which clients will arrive and what decisions will be feasible.Static approaches assume the information needed to determine an allocation is already available.Source: 1) Dealing with uncertainties about… · pp. 8–9
1) Dealing with uncertainties about future events:
- [63] guarantees truthful reports in two cases—myopic sellers or buyers interested in one time slot—while relaxing truthfulness may permit higher efficiency through strategic behavior.The survey also calls for surveying current users’ economic interests and potential users’ expectations to develop and validate preference models.Source: 1) Dealing with uncertainties about… · p. 9
2) Mobility-based charging management models:
Mobility-aware charging schemes account for where EVs travel and let drivers or stations make reservation and pricing decisions. The reviewed approaches span queue reduction, cost-aware routing, station competition, and two-sided energy allocation, while analytical results for models combining economics and mobility remain limited.
Charging reservation
- Mobility-based schemes include shortest-path station recommendations that avoid detours and can be revised as EVs travel.Candidate stations lie on the shortest route to the destination; reservations are updated in later rounds.Source: 2) Mobility-based charging management… · p. 9
- Simulations show that a distributed station-reservation algorithm performs close to a proven lower bound on EV waiting time.EVs periodically request reservations from reachable stations, which estimate waiting times; drivers can cancel and reschedule as their journeys progress.Source: 2) Mobility-based charging management… · p. 9
- Drivers may trade longer travel and waiting times for lower energy costs through joint routing and charging decisions.For taxis, a separate model maximizes long-term average profit under state-of-charge dynamics and periodically varying service income and electricity prices.Source: 2) Mobility-based charging management… · p. 9
Station competition
- Price competition among stations yields higher station utility than the Bertrand equilibrium and lets users save up to 11.5% versus choosing the nearest station.Stations may over-provision energy based on requests sent by users to multiple stations, though the authors say shortages are unlikely.Source: 2) Mobility-based charging management… · p. 10
- A two-sided auction assigns each EV a station-slot pair by maximizing the positive difference between user willingness to pay and station-reported cost.EVs report willingness-to-pay matrices, stations report unit costs, and a central controller selects the pair.Source: 2) Mobility-based charging management… · p. 10
Mobility-based models
- Mobility-aware models have few analytical results because of their complexity, though numerical findings suggest the approach could yield significant improvements.Historical travel surveys can provide statistical insights, and the survey treats gasoline-car results as transferable to EVs.Source: 2) Mobility-based charging management… · p. 10
3) The special case of (unidirectional) regulation service and wind-balancing:
The survey treats regulation and wind-balancing as distinct uses of flexible charging, then reviews bidirectional trading in which EVs can also return electricity to the grid. The approaches depend on timely response, adequate EV participation, and economic conditions that make participation worthwhile.
3) The special case of (unidirectional) regulation service and wind-balancing:
- EVs can provide regulation by increasing charging during oversupply and reducing it during shortages, but meaningful service requires sufficient EV penetration.Regulation targets frequency deviations rather than trading off EV energy valuation against generator production costs.Source: 3) The special case of… · p. 10
- Two-sided regulation is more economically attractive when capacity payments are offered; without them, parameter uncertainty makes accurate prediction important.The compared uncertainties include day-ahead wholesale prices, regulation prices, and vehicle mobility.Source: 3) The special case of… · p. 10
- Regulation models face seconds-scale response, unpriced battery effects from rapid power changes, and hourly capacity commitments that may exceed EV plug-in durations.The survey finds no representative unidirectional regulation models that incorporate these EV costs; uncertain availability also complicates payment sharing.Source: 3) The special case of… · pp. 10–11
- Wind-balancing uses stochastic optimization at each time slot to choose EV charging rates that minimize current and expected production mismatch.Wind farms declare next-day production using forecasts of generation and EV availability.Source: 3) The special case of… · p. 11
IV. BIDIRECTIONAL ENERGY TRADING
- Bidirectional trading lets EVs buy electricity from the grid and sell it back through Vehicle-to-Grid technology, using batteries as grid storage.The grid-scale contribution requires sufficient EV penetration, and owners need a share of the benefit to make storage participation attractive.Source: §IV · p. 11
A. Individual arbitrage
- Individual arbitrage buys electricity at low prices and sells it when prices are higher, with mobility and conversion losses affecting revenue.One proposed algorithm assumes advance knowledge of market-clearing prices, which the survey calls a very strong assumption.Source: A. Individual arbitrage · p. 11
- A household charging scheme using estimated mobility and energy demand performs close to one with perfect information, while congestion adjustments minimize added EV cost.The scheme uses Ontario time-of-use pricing; Fig. 5 identifies its price periods.Source: A. Individual arbitrage · p. 11, Figure 5 · p. 11
B. V2G for regulation services
The survey reviews V2G regulation and renewable-energy storage schemes, emphasizing how incentives and aggregation address EV owners’ constraints and grid needs. It also describes approaches that use EV batteries to reduce renewable-generation variability.
B. V2G for regulation services
- Regulation services can be more profitable for EVs than selling energy as peak power, though profitability varies by market and may disappear when payments fail to cover battery degradation.The cited case study found V2G regulation profitable in most of four US markets; the German market yielded significantly higher PHEV profits than the Swedish market.Source: B. V2G for regulation services · p. 12
- Aggregators pool EVs to meet regulation-market reliability and capacity requirements that individual vehicles cannot satisfy.The cited estimates put individual-EV reliability at 83.6%–91.7% over one hour, versus 98.89% for natural-gas turbines; market minimum contract capacities range from 0.1 MW to 10 MW.Source: B. V2G for regulation services · p. 12
- Using real PJM price signals, the strategy estimates that an EV providing regulation for 15 hours per day can earn about $100 per month at current Regulation Market Clearing Prices.A conservativeness scaling parameter balances revenue against penalties for unmet power requests, and the authors suggest Shapley-value revenue sharing [78].Source: B. V2G for regulation services · p. 12
- Regulation allocation schemes trade off user welfare, state-of-charge balance, and payoff equality; a dynamic Lyapunov algorithm is reported asymptotically optimal and substantially better than a greedy per-slot method.The water-filling approach drives SOC variance toward zero but produces greater variance in user payoffs than social-welfare maximization.Source: B. V2G for regulation services · p. 12
- EV regulation must accommodate travel plans: one user-level method replans from current state of charge as bounded regulation fluctuations occur, relying on regulation requests averaging near zero.The initial plan maximizes utility over the plugged-in period assuming no regulation; later replanning adjusts it as SOC changes.Source: B. V2G for regulation services · pp. 12–13
- Aggregator pricing schemes can elicit voluntary EV participation, but some assume identical users; a heterogeneous-user model computes an optimal price under a regular willingness-to-pay distribution.When that distribution is unknown, the aggregator can learn a price through interactions with EVs.Source: B. V2G for regulation services · p. 13
C. V2G as storage for renewable energy
- A virtual power plant model rewards EVs with free electricity proportional to the storage they provide and adjusts hourly storage decisions as wind-generation predictions improve.The approach was tested in a realistic case study using wind-generation data, market prices, and EV characteristics.Source: C. V2G as storage for renewable energy · p. 13, Figure 6 · p. 13
- V2G storage can buffer renewable output variation, and one comparison found wind-farm penalty reductions exceeded EVs’ lost regulation revenue, leaving room for both sides to benefit.A separate estimate suggests renewable electricity could supply 90%–99.9% of the time at costs comparable to today’s when generation and storage technologies, including EV fleets, are optimized.Source: C. V2G as storage for renewable energy · p. 13
V. COMMUNICATION ASPECTS
The paper treats communication requirements as part of charging-scheme implementation, documenting information exchanges across actor configurations and the timescales at which decisions and updates occur.
A. Information exchanges
- Information-exchange requirements are organized for two-actor and three-layer schemes, with the three-layer exchanges varying in sequence across schemes.The two-actor table covers EVs and energy sellers; the three-layer table covers configurations such as EV–aggregator–ISO.Source: A. Information exchanges · pp. 13–14, Table VI · p. 14
- Communication burden ranges from lighter to heavier schemes; convergence-based mechanisms require repeated, low-latency exchanges before decisions can be made.The classification also notes that information retrieval, including environmental data and travel records, requires communication.Source: A. Information exchanges · pp. 13–14
B. Time granularity
- Charging-management algorithms operate from seconds to day-ahead timescales, with update frequency tied to the events or new information driving decisions.The time-scale classification distinguishes rapid regulation allocation, event-driven responses, roughly hourly updates, and day-ahead planning.Source: B. Time granularity · p. 14
VI. CLASSIFICATION OF APPROACHES AND RESEARCH CHALLENGES
The classification distinguishes static from dynamic models and separates centralized optimization from approaches that account for interactions among rational actors. It identifies open challenges in dynamic incentive compatibility and convergence.
VI. CLASSIFICATION OF APPROACHES AND RESEARCH CHALLENGES
- Static models optimize over an isolated interval using current state information, while dynamic models update actions as system information changes over time.The paper presents the static setting as a special case in which the state is constant, though it still depends on actions.Source: §VI · p. 14, Figure 7 · p. 14
- Optimization-based approaches treat a central controller’s decisions as unaffected by other actors, whereas game-theoretic approaches let actors’ selfish behavior affect allocations.The classification also distinguishes revelation schemes, which exchange information actors may strategically withhold, from iterative tâtonnement schemes that converge allocations and often prices.Source: §VI · pp. 14–15
- Research priorities include more computationally efficient or near-optimal static optimization and more accurate prediction and robustness to residual uncertainty in dynamic models.The paper also argues for developing game-theoretic approaches for dynamic settings despite the difficulty of designing incentive-compatible schemes.Source: §VI · p. 15
- Dynamic incentive-compatible schemes remain an open challenge because actors’ knowledge, beliefs, and actions evolve, while analytical proofs apply only to specific utility functions.The paper notes that analyzing equilibria becomes especially complex when actions reveal private information.Source: §VI · p. 15
- Tâtonnement-based mechanisms have a convergence-speed limitation, especially in dynamic settings where prices and allocations may not settle before conditions change.The literature barely addresses this issue, and convergence is established only for some specific utility functions that need validation.Source: §VI · p. 15
VII. CONCLUSION
The survey reviews economy-driven EV charging schemes and their potential grid and owner benefits. It finds limited analytical results amid complex settings, and identifies emerging applications and modeling boundaries.
VII. CONCLUSION
- Coordinated charging and discharging can mitigate grid impacts and support frequency regulation and renewable-energy storage, while offering owners lower energy costs or profits.The conclusion frames EVs and the Smart Grid as mutually reinforcing.Source: §VII · p. 15
- The survey covers economy-driven models from energy sharing to settings with uncertainty, mobility, station locations, and grid services, but reports relatively few analytical results.The paper attributes this scarcity to complex actors, network and battery constraints, and nonaligned economic objectives.Source: §VII · pp. 15–16, Table IX · p. 16
- The authors identify microgrids and EV fleets as promising research directions, including the prospect of driverless taxi fleets with new charging and service-management possibilities.They note that roadbed infrastructure enabling charging in motion would change assumptions about battery capacity and plug-in time, and that economic models for such scenarios remain undefined.Source: §VII · p. 16