PaperlayerView exact revision on arXiv

Source-linked AI summary

Smart Charging for Electric Vehicles: A Survey From the Algorithmic Perspective

Qinglong Wang, Xue Liu, Jian Du, Fanxin Kong

arXiv:1607.07298v1eess.SY

TL;DR

The paper addresses how to coordinate interactions among EVs, aggregators, and the smart grid to obtain benefits while handling charging and system uncertainties. It surveys algorithmic formulations across smart-grid-, aggregator-, and customer-oriented charging, then identifies modeling challenges and future directions. The survey concludes that these perspectives organize existing approaches and that further work on EV batteries and communication networks could accelerate smart-grid–EV interaction.

  • Problem

    Smart-grid–EV interaction remains in its infancy, while coordinated charging must address grid reliability, aggregator profits, customer benefits, and uncertainties across the three parties.

  • Method

    The survey categorizes and analyzes algorithmic research as smart-grid-oriented, aggregator-oriented, and customer-oriented charging, including direct or indirect control and uncertainty formulations.

  • Results

    The survey finds that load-flattening, power-loss-minimization, and load-factor-increase problems are approximately equivalent, while aggregator-oriented control includes direct and indirect approaches.

  • Takeaways & Limitations

    Future research on the distinctive properties of EV rechargeable batteries and communication networks can help accelerate smart interaction between EVs and the smart grid.

Abstract

from arXiv · show

Smart interactions among the smart grid, aggregators and EVs can bring various benefits to all parties involved, e.g., improved reliability and safety for the smart gird, increased profits for the aggregators, as well as enhanced self benefit for EV customers. This survey focus on viewing this smart interactions from an algorithmic perspective. In particular, important dominating factors for coordinated charging from three different perspectives are studied, in terms of smart grid oriented, aggregator oriented and customer oriented smart charging. Firstly, for smart grid oriented EV charging, we summarize various formulations proposed for load flattening, frequency regulation and voltage regulation, then explore the nature and substantial similarity among them. Secondly, for aggregator oriented EV charging, we categorize the algorithmic approaches proposed by research works sharing this perspective as direct and indirect coordinated control, and investigate these approaches in detail. Thirdly, for customer oriented EV charging, based on a commonly shared objective of reducing charging cost, we generalize different formulations proposed by studied research works. Moreover, various uncertainty issues, e.g., EV fleet uncertainty, electricity price uncertainty, regulation demand uncertainty, etc., have been discussed according to the three perspectives classified. At last, we discuss challenging issues that are commonly confronted during modeling the smart interactions, and outline some future research topics in this exciting area.

I. INTRODUCTION

Rising EV adoption creates substantial charging-load challenges for the power grid, while coordinated smart interactions can support grid integration and benefits for grid, aggregators, and customers. The survey reviews these interactions algorithmically, emphasizing aggregator-mediated coordination, EV characteristics, communication infrastructure, and open research needs.

  • I. INTRODUCTION: An AC level 2 EV can draw almost twenty times the load of a typical North American home, and uncoordinated aggregation can worsen grid imbalance, losses, and voltage deviation.Coordinated control can help integrate EVs, renewable generation, and distributed generation through intelligent control of bidirectional power flow.
  • I. INTRODUCTION: Bidirectional power flow and communication form the interaction foundation, with aggregators mediating power exchange and communication networks collecting vehicle data and disseminating control information.Roadside units support real-time information transfer through vehicle-to-vehicle and vehicle-to-RSU communication.
  • I. INTRODUCTION: The survey differs from prior reviews by focusing on intensive smart-grid–EV interaction through aggregators rather than primarily on infrastructure impacts or renewable-energy integration.It frames potential benefits as improved grid reliability and safety, aggregator profits, and customer self benefit, while noting that widespread practice remains limited.
  • I. INTRODUCTION: The survey organizes research into smart-grid-oriented, aggregator-oriented, and customer-oriented EV smart charging, followed by open issues and future research directions.Its background covers EVs, batteries, the smart grid, and communication networks before examining the three charging perspectives.
  • I. INTRODUCTION: EV charging has spatial and temporal characteristics because vehicles carry stored electricity across locations and initiate or stop charging randomly over time.Accurate modeling supports grid planning and charging-service scheduling, but extensive trip-data processing can make complex EV models computationally infeasible.
  • I. INTRODUCTION: Available studies model temporal charging behavior statistically, while synthesized datasets combine real taxi trajectories with PHEV characteristics to represent spatio-temporal charging loads.One cited model uses Gaussian arrival times conditioned on chi-square departure times; another constructs a dataset from North American PHEV features and taxi trajectories.

B. Properties of Rechargeable Batteries

Rechargeable-battery properties distinguish EV charging from conventional refueling and constrain coordinated-charging models. Key considerations include controllable power, nonlinear SOC gains, battery-aging costs, and uncertainty in aging estimates.

  • 1) Charging Power Controllability: Li-ion batteries require delicate voltage and current control, while current AC level 1, AC level 2, and DC fast-charging standards do not provide continuously controllable charging power.Without more advanced battery techniques, charging power may vary only across a discrete set of nearly constant values.
  • 2) Battery Charging Rate: Charging time and obtained SOC are nonlinear, especially during the typically last 1/3 of charging, whereas discharge-related charge loss is linearly dependent on time.The effective charging current drops as cell open-circuit voltage rises during the high-SOC portion of charging.
  • 3) Battery Aging: Charging-cost models commonly include electricity expenditure, but battery aging is a separate and harder-to-model cost that can materially affect EV-customer payback.The survey identifies battery aging as an important factor for coordinated-charging design because Li-ion batteries represent a significant share of EV prices.
  • 3) Battery Aging: Battery life is commonly defined by the minimum of calendar life and cycle life, with temperature affecting calendar life and usage pattern, current, and cycle depth affecting cycle life.These factors contribute to capacity fade and increased internal resistance.
  • 3) Battery Aging: Frequent charging and discharging for auxiliary services can shorten battery life, while reducing cycle depth can extend it because aging per cycle depends nonlinearly on cycle depth.Studies incorporate cyclic-aging cost or the fraction of depleted battery life when optimizing V2G energy trading and coordinated charging.
  • 3) Battery Aging: Battery-aging estimates may be inaccurate because real operating conditions rarely match laboratory conditions, and different Li-ion batteries and operating scenarios have different aging characteristics.Relevant scenarios include daily driving cycles, charging initiation times, and charging strategies.

C. EVs Interacting with Smart Grid

The smart grid interacts with EVs through bidirectional energy and communication flows mediated by aggregators, while coordination architectures, V2G, and communication infrastructure shape practical deployment.

  • The smart grid enables bidirectional energy flows through coordinated control, EV aggregation, V2G, and two-way communication.
  • Centralized coordination directly controls all EVs but requires accurate status data and has poor scalability, making it less practical.
  • Hierarchical coordination uses aggregators as intermediaries for energy trading without directly controlling participating EV charging rates.
  • V2G can provide voltage regulation without severe battery degradation, but EV customers remain reluctant to transfer control to third parties.
  • Effective electricity dispatch requires accurate power and phase measurements plus real-time, high-precision, low-latency communication networks.
  • Cellular and WiFi support mobile information exchange, but vehicle mobility can cause inaccurate location measurements, collisions, and transmission failures.
  • VANETs support multi-hop V2V and V2R communication through vehicles and roadside units, improving the efficiency of vehicular information collection.

2) Communication Protocols and Standards:

EV smart charging depends on communication standards that connect smart grids, aggregators, and customers while meeting reliability, latency, privacy, and security requirements.

  • 1) Vehicular Ad-hoc Networks: VANETs can collect EV information through multi-hop DSRC communication from vehicles to roadside units and onward to the smart grid.
  • 1) Vehicular Ad-hoc Networks: Communication delay can compromise smart-grid stability and cause unexpected driving costs despite VANETs’ gains in accuracy, range, and reliability.
  • 2) Communication Protocols and Standards: ZigBee, ZigBee Smart Energy Profile, and RF mesh systems are among the communication technologies studied or adopted for smart-grid networks.
  • 2) Communication Protocols and Standards: Smart-grid-to-aggregator communication requires high reliability, availability, and quality of service to maintain grid stability and safety.
  • 2) Communication Protocols and Standards: Aggregator-to-customer communication requires access control and protection for charging records that can reveal customers’ private mobility patterns.
  • 2) Communication Protocols and Standards: Uncontrolled EV charging can increase transformer temperatures, power losses, voltage and frequency deviations, and other threats to grid stability and reliability.
  • 2) Communication Protocols and Standards: Coordinated valley filling shifts large demands toward load-profile valleys, balancing active power and helping maintain grid frequency stability.
  • 2) Communication Protocols and Standards: Grid-oriented charging research addresses load flattening, frequency regulation, and voltage regulation.

A. Load Flattening

Load-flattening research formulates EV charging coordination as optimization under charging-demand constraints, distinguishing direct and indirect objectives while addressing information and controllability assumptions.

  • A. Load Flattening: Load-flattening strategies are divided into direct control of EV charging loads and indirect approaches focused on charging-related costs.
  • 1) Direct Load Flattening: The general formulation assumes continuously controllable charging power, known EV fleet information, and defined initial, requested, minimum, and maximum energy or power quantities.
  • 1) Direct Load Flattening: Optimization formulations commonly minimize aggregated loads, load variance, or EV charging energy costs while satisfying charging demands over a time horizon.
  • 1) Direct Load Flattening: Offline optimization can be used when future base loads, EV counts, and homogeneous charging intervals are perfectly known in advance.
  • 1) Direct Load Flattening: Because the aggregated-load objective is convex, decentralized gradient projection can reach a global optimum, although multiple individual-load combinations may realize it.
  • 1) Direct Load Flattening: Restricting each EV to fixed charging power after initiation makes the decision set discrete and prevents interruption once charging begins.
  • 1) Direct Load Flattening: Minimizing load variance is equivalent to minimizing aggregated load in cited work and designs profiles near the average load.
  • 1) Direct Load Flattening: Offline formulations require future loads, charging demands, and arrival or departure times, so performance depends closely on prediction accuracy.

2) Cost Minimization:

Aggregator-oriented cost minimization models combine generation and consumption costs or directly minimize EV charging cost under energy and deadline constraints. These formulations can be solved using linear programming or MPC-like methods, and global coordination outperforms local coordination for load flattening.

  • Cost objectives: Aggregators model charging objectives using generation cost, consumption cost, or combined generation and carbon-emission tax costs.A linear electricity-price model produces a quadratic consumption-cost function, while piecewise-linear generation costs can be converted for standard linear programming.
  • Charging-cost formulation: EV charging-cost minimization uses instant energy and deadline constraints to ensure feasible battery operation and required final charge levels.The first constraint limits instantaneous energy by maximal battery capacity, and the second guarantees predefined charging levels by the horizon’s end.
  • Solution approach: He et al. solve the charging-cost formulation with an MPC-like approach that avoids requiring long-term prediction of future information.
  • Coordination scale: Global and local load-flattening approaches both effectively flatten loads, but global coordination performs better.Global aggregation coordinates charging across a broader scope, whereas local facilities manage loads within their own residential areas.

3) A General Formulation for Load Flattening Problems:

The survey presents a general optimization view of load flattening and extends it to bidirectional power exchange, frequency regulation, and economic coordination. Different objectives and information conditions produce adaptations ranging from offline optimization to distributed or pricing-based control.

  • 3) A General Formulation for Load Flattening Problems:: Load variance, power losses, and load factor form a closely related optimization triangle, with load-variance and load-factor equivalence independent of topology.Because these problems are commonly convex, load-flattening algorithms can approximately address related objectives with lower computational cost than direct power-loss minimization.
  • 3) A General Formulation for Load Flattening Problems:: A general load-flattening formulation can vary its objective across aggregated loads, aggregated energy, and aggregated charging cost.
  • 3) A General Formulation for Load Flattening Problems:: The formulation supports offline computation when forecasts are sufficiently accurate and adaptive approaches when exogenous information is uncertain or unavailable.
  • 3) A General Formulation for Load Flattening Problems:: Bidirectional power flow enables V2G services such as voltage regulation, frequency regulation, and spinning reserve in addition to grid-to-EV charging.Allowing minimum charging power to be negative converts load flattening into peak shaving by treating batteries as distributed generators.
  • B. Frequency Regulation: Frequency regulation aggregates small EV power contributions because individual EVs otherwise do not meet the MW-scale minimum power threshold.PEVs are described as suitable for regulation because they can respond rapidly to frequency changes.
  • B. Frequency Regulation: Frequency regulation and voltage regulation can be interpreted as balancing active power and reactive power, respectively.A joint coordination algorithm balances both supply-demand dimensions, while price-based control minimizes active-power consumption cost for individual EVs.
  • B. Frequency Regulation: Frequency-regulation models distinguish capacity-based regulation payments from energy-based charging expenses and penalize unmet charging demand at the deadline.The formulation represents regulation price as linearly dependent on EV state of charge and uses a control sequence for charging operation timing.
  • B. Frequency Regulation: Game-theoretic coordination can minimize backup-battery usage while satisfying regulation capacity, and pricing can yield a Nash equilibrium maximizing participating EV profits.Consensus filtering is also proposed to obtain consistent frequency-deviation signals when centralized measurement acquisition is impractical.

C. Voltage Regulation

Voltage regulation is formulated as reactive-power compensation while respecting network-flow, voltage, charging, and apparent-power constraints. Surveyed approaches include direct or price-based PQ control, hierarchical strategies, and Stackelberg games for EVs co-located with wind generation.

  • C. Voltage Regulation: Voltage regulation maintains reactive-power balance, and EV on-board inverters can provide reactive-power compensation alongside traditional capacitors.
  • C. Voltage Regulation: Direct and price-based voltage-regulation approaches couple active power P and reactive power Q through a maximal apparent-power constraint.The apparent power is defined as A = V × I, the product of grid voltage and charger current.
  • C. Voltage Regulation: A hierarchical strategy prioritizes reactive-power adjustment and gives greater weight to nodes nearer the target node.The strategy can trigger PQ control by reducing active charging power from its maximum level to generate reactive power.
  • C. Voltage Regulation: Another PQ-control formulation minimizes active-power cost together with the maximum voltage deviation from the substation reference voltage.The voltage function is f(V) = max_i |V_i − V_0| across the distribution-network buses.
  • C. Voltage Regulation: Distribution-network formulations use DistFlow transition functions for current, active-power flow, reactive-power flow, and voltage, while enforcing EV charging and apparent-power limits.The constraints ensure EVs reach desired energy levels by their deadlines without violating maximal allowable apparent power.
  • C. Voltage Regulation: For EV charging stations paired with wind generation, a Stackelberg game tunes pricing and reactive-power demand before EVs reach a second-stage Nash equilibrium.The resulting optimum has all EVs provide equal reactive power at an extreme or at the averaged compensation difference.

D. Smart Grid Oriented Uncertainty

Smart-grid coordination faces load and regulation uncertainty, both inherited from uncertainty in EV mobility and fleet behavior. Existing work commonly relies on future-load forecasts, while regulation uncertainty is handled by categorizing EVs according to their capabilities or conditions.

  • D. Smart Grid Oriented Uncertainty: The smart-grid operator’s main uncertainties are load uncertainty and regulation uncertainty, both originating from EV-fleet mobility uncertainty.
  • D. Smart Grid Oriented Uncertainty: Load flattening commonly assumes forecasts of future load learned from historical data.Such information is needed when minimizing load variance or load deviation.

IV. AGGREGATOR ORIENTED EV SMART CHARGING

Aggregator-oriented charging examines how aggregators coordinate EV customers while balancing grid responsibilities with customer charging service and aggregator objectives. The survey covers direct and indirect control, utility-based allocation, real-time scheduling, and distributed solution methods.

  • Hierarchical coordination replaces one global controller with multiple local aggregators that manage charging loads within local areas.
  • Aggregators bridge smart-grid operations and EV-customer charging services while pursuing operational-cost savings or increased profits.
  • Direct Coordinated Control: Direct coordination strategies maximize aggregate customer welfare by allocating instantaneous charging power under distribution-line and transformer capacity limits.Utility models include logarithmic utility, which supports proportional fairness.
  • Direct Coordinated Control: Convex welfare and cost formulations can be decomposed across measurement-and-control nodes and EV chargers, enabling simultaneous updates or fully distributed consensus solutions.The decentralized consensus approach assumes EVs coordinate with neighbors through peer-to-peer communication.
  • Direct Coordinated Control: Real-time charging allocation can be formulated as resource scheduling that decides admission and charging schedules for EV tasks.Approaches include TAGS, EDF- and LLF-based heuristics, and MPC using forecasts of future renewable generation.

B. Indirect Coordinated Control

Indirect coordinated control uses prices or revenues to influence EV-customer participation rather than directly controlling charging. The aggregator must balance customer attraction, grid-loading risk, service revenue, and uncertainty across market and operating conditions.

  • Indirect control sets service prices or revenues to attract EV customers while aggregators earn rewards for committed services or pay penalties for failing them.Examples include motivating customers to sell excess energy during peak hours.
  • Service pricing creates a trade-off between attracting customers and preserving aggregator profits.Prices that are too high may reduce participation, while prices that are too low may increase overload and penalty risk.
  • Effective aggregator coordination must account for EV-fleet, electricity-price, regulation-demand, regulation-price, and distributed-generation uncertainty.

V. CUSTOMER ORIENTED EV SMART CHARGING

Customer-oriented charging prioritizes EV owners’ benefits, especially minimizing charging cost while meeting energy needs across intermediate stops and destinations. The survey also identifies modeling limitations and future work involving realistic batteries, charging-pattern forecasts, scheduling, communications, privacy, and security.

  • Individual Charging Cost Reduction: Customer-oriented smart charging minimizes an EV owner’s charging cost while fulfilling energy requirements at intermediate stops and the destination.
  • Individual Charging Cost Reduction: Customer decisions may require predicted service prices, regulation signals, driving cycles, and demanded energy, with expected-cost minimization used when some information is difficult to predict.
  • Individual Charging Cost Reduction: Stochastic formulations incorporate charging cost, penalties for insufficient energy at intermediate stops, regulation participation, and random exogenous information over a time horizon.Continuous decision spaces and random energy states can require discretization.
  • Individual Charging Cost Reduction: Individual EV charging-cost scheduling remains rarely studied, despite the potential to optimize both charging cost and charging time under nonlinear SOC relationships.
  • Customer-side uncertainty mainly concerns EV mobility, electricity prices, regulation prices, and regulation demand.
  • Future Work: Future work should address nonlinear battery charging, inaccurate charging-pattern assumptions, intelligent routing and scheduling, communication requirements, privacy, and cyber-security threats.The survey notes that linear battery models can increase charging cost and charging time, while communication delay can affect scheduling and grid stability.

VII. CONCLUSION

The survey organizes algorithmic research on EV smart interactions by three standpoints and identifies approaches for grid objectives, aggregator satisfaction, and customer charging costs. It also proposes future research directions grounded in EV batteries and communication networks.

  • The survey categorizes EV smart-charging research into smart grid oriented, aggregator oriented, and customer oriented perspectives.
  • Smart grid oriented EV smart charging: Smart grid oriented studies address load flattening through optimization-based approaches, with related objectives including minimizing power losses and increasing load factor.
  • Aggregator oriented EV smart charging: Aggregator oriented studies examine direct and indirect control approaches to maximize the overall satisfaction of present EV customers.
  • Customer oriented EV smart charging: Customer oriented studies use stochastic optimization to design charging schedules that minimize individual EV customers’ charging costs.
  • Future research directions: The survey identifies research directions based on the crucial and unique properties of EV rechargeable batteries and communication networks.
Loading 1607.07298v1…