Source-linked AI summary
EV Dispatch Control for Supplementary Frequency Regulation Considering the Expectation of EV Owners
Hui Liu, Junjian Qi, Jianhui Wang, Peijie Li, Canbing Li, Hua Wei
TL;DR
EV participation in SFR must satisfy both control-center regulation commands and owners’ expected battery SOC requirements. The paper proposes hierarchical closed-loop V2G control with uncertain center dispatch, FRC-based allocation, and real-time station correction. Simulations on an interconnected two-area power system show that frequency regulation and expected EV battery SOC can both be achieved.
Problem
EV SFR must simultaneously follow uncertain control-center dispatch and satisfy EV owners’ expected battery SOC levels.
Method
The paper proposes hierarchical closed-loop V2G control using uncertain center dispatch within EV FRC and real-time correction of scheduled V2G power at charging stations.
Results
Simulations on an interconnected two-area power system show that ACE and frequency fluctuations are well suppressed while expected EV battery SOC is guaranteed.
Takeaways & Limitations
EVs can participate in SFR while the proposed control simultaneously maintains frequency-regulation performance and expected battery SOC levels.
Abstract
from arXiv · showhide
Electric Vehicles (EVs) are promising to provide frequency regulation services due to their fast regulating characteristics. However, when EVs participate in Supplementary Frequency Regulation (SFR), it is challenging to simultaneously achieve the dispatch of the control center and the expected State of Charge (SOC) levels of EV batteries. To solve this problem, in this paper we propose a Vehicle-to-Grid (V2G) control strategy, in which an uncertain dispatch is implemented in the control center without detailed EV charging/discharging information. The regulation from the control center is achieved by allocating the regulation task within the frequency regulation capacity (FRC) of EVs. The expected SOC levels of EV batteries are guaranteed by a real-time correction of their scheduled V2G power in EV charging stations. Simulations on an interconnected two-area power system validate the effectiveness of the proposed V2G control in achieving both the frequency regulation and the expected SOC levels of EVs.
S Total regulation-up FRC of all EVs at time k+1
EVs can support SFR because of their fast regulation characteristics, but dispatch must also respect owners’ expected battery SOC levels. The proposed closed-loop V2G strategy addresses both requirements without requiring detailed EV charging/discharging information at the control center.
- Motivation: EVs are promising frequency-regulation resources that can reduce frequency deviation and conventional generators’ regulation reserves.EVs’ increasing deployment supports their use as ancillary-service resources, particularly for frequency regulation.
- SFR Background: SFR determines EV charging/discharging power from the control-center regulation signal, with aggregators needed because individual EV power is only up to 20 kW.Frequency-regulation requirements are in the megawatt range, so the control center must work through an aggregator.
- Research Gap: Prior approaches did not thoroughly address owners’ expected SOC, or required expected battery charging/discharging power to be uploaded to the control center in real time.The latter requirement increases communication costs and is impractical for system operators to incorporate in real power systems.
- Contribution: The proposed hierarchical closed-loop V2G control uses uncertain center dispatch within EV FRC and real-time scheduled-power correction to achieve regulation and expected SOC.The hierarchy includes the control center, EV aggregators, charging stations, and individual EVs.
II. HIERARCHICAL FRAMEWORK OF ELECTRIC VEHICLES PARTICIPATING IN SFR
The paper organizes EV participation in SFR through a four-level hierarchy. Control-center dispatch is allocated through aggregators and charging stations, while individual EV commands and information support real-time SOC management.
- Conventional SFR: SFR suppresses ACE fluctuations and keeps interconnected-system frequency within its tolerance bound by adjusting generating-unit outputs.Under tie-line bias control, ACE is calculated from frequency and tie-line power deviations.
- Conventional SFR: Conventional LFC requires generator characteristics such as ramp speed, power limits, and outputs to calculate FRC and perform dispatch.Generator outputs are adjusted according to the control-center dispatch.
- Hierarchical V2G Framework: The proposed EV framework has four levels: control center, EV aggregator, EV charging station, and individual EV.This hierarchy structures dispatch, aggregation, station-level control, and EV-level execution.
- Hierarchical V2G Framework: The control center calculates ACE and randomly dispatches part of it to EV aggregators within the EVs’ FRC.Aggregators allocate received regulation to charging stations according to uploaded capability.
- Hierarchical V2G Framework: Charging stations allocate regulation to individual EVs, regulate charging power in real time for expected SOC, and calculate each EV’s FRC.An information-management block communicates with individual EVs through interface circuits.
III. DISPATCH CONTROL OF EVS FOR PERFORMING SFR
The dispatch controller must accommodate uncertain regulation commands while preserving EV owners’ expected SOC levels. It therefore dispatches ACE within EV FRC and relies on station-level V2G adjustment to manage battery energy.
- Dispatch Objectives: The proposed V2G control addresses both control-center regulation dispatch and EV owners’ expected battery energy levels.These are identified as the two central concerns for EV participation in SFR.
- Regulation Dispatch: ACE dispatch to EVs must remain within their FRC, while uncertainty arises from markets, EV randomness, and load fluctuations.ACE reflects generation-load mismatch, whereas FRC represents available EV regulation capability.
- Expected SOC: EV owners’ SOC expectations fall into three types: increasing battery energy, decreasing battery energy, or maintaining battery energy.The required V2G behavior differs according to the owner’s driving and energy-use preferences.
- Uncertain Dispatch: The control-center dispatch is represented as a function of ACE and is allocated to EVs within their available regulation capacity.The dispatch function may depend on market conditions, renewable penetration, generator reserves, and EV FRC, making it uncertain.
- Uncertain Dispatch: The uncertain dispatch depends on ACE, aggregate regulation-up and regulation-down FRC, and a random ratio R, modeled as normally distributed between 0 and 1.The normal-distribution assumption is motivated by typically similar up/down dispatches and a preferred participation proportion.
C. Dispatch Control in EV Aggregators
At the aggregator level, EV regulation capability is aggregated across charging stations and used to distribute the control-center task. Any remaining regulation task is assigned to conventional generating units.
- Total FRC: An EV aggregator’s total FRC is calculated from the FRCs uploaded by its EV charging stations.The aggregator sums charging-station regulation-up and regulation-down capabilities.
- Station Dispatch: Each charging station receives a regulation task proportional to its uploaded FRC.This allocation uses the station-level capabilities reported to the aggregator.
- Residual Dispatch: Generating units undertake the remainder of the regulation task after the EV contribution is allocated.The dispatch therefore shares the regulation requirement between EV aggregators and conventional generation.
D. V2G Strategies in EV Charging Stations
The charging-station strategy combines scheduled V2G power with regulation dispatch while correcting scheduled power in real time to maintain each EV’s expected SOC. Regulation tasks are allocated within EV frequency regulation capacity (FRC).
- FRC calculation: The FRC of each EV is calculated from its real-time V2G power and varies as scheduled charging and regulation change that power.Maximum V2G power is device-dependent, whereas real-time V2G power changes with regulation and scheduled charging.
- SOC deviation: Unequal regulation-up and regulation-down FRC can produce unequal dispatch energy and drive the battery SOC away from its expected level.When V2G power is positive or negative, regulation-up and regulation-down FRC change in opposite directions.
- Scheduled V2G power: When regulation energy is balanced, constant scheduled V2G power can guarantee expected SOC; otherwise, retaining it causes SOC deviation.The proposed correction is needed because uncertain dispatch generally makes regulation-related energy change nonzero.
- Real-time correction: A real-time correction adjusts scheduled V2G power to compensate for battery-energy changes caused by uncertain control-center dispatch.The correction uses the EV battery SOC and remaining plug-in duration and operates in a real-time closed loop.
- Regulation allocation: The regulation task is assigned proportionally within each EV’s uploaded FRC so that every EV can achieve its assigned regulation.The strategy commonly uses a constant charging or discharging efficiency because variable efficiency is difficult to obtain.
E. Discussion on FRC and Scheduled V2G Power
Scheduled V2G power changes an EV’s available regulation capacity and determines whether it can participate in SFR. Extreme scheduled-power values remove one regulation direction or exclude the EV from SFR.
- Scheduled V2G power and FRC: Increasing scheduled V2G power increases regulation-up FRC and decreases regulation-down FRC; decreasing it has the opposite effect.The FRC is determined by the EV’s real-time V2G power when regulation is not considered.
- SFR participation limits: At scheduled V2G power equal to or above maximal V2G power, an EV does not participate in SFR because regulation-down capability is unavailable.Regulation-up would further reduce battery energy, so charging is the only option in this condition.
- SFR participation limits: At scheduled V2G power equal to or below negative maximal V2G power, the EV likewise does not participate in SFR.The participation rule applies symmetrically at the negative power boundary.
A. Simulation System
The simulations use a MATLAB model of an interconnected two-area power system based on real Chinese grid data, with EV and wind integration in area A. Battery energy variation is computed from V2G power over time.
- System model: The simulated system is an interconnected two-area power grid modeled in MATLAB from real power-grid data in China for SFR with EV participation.Area A uses TBC, area B uses FTC, and EVs and wind power are integrated in area A.
- System model: Area A includes 30 of 73 generating units in SFR, while the remaining units follow generation curves.Practical operating strategies for conventional generating units are also modeled in area A.
- Load profile: The load profile is historical data from a real power grid in China and is used to represent system-load variation.The profile is presented as a random load profile for the simulation system.
- Battery model: Battery energy variation is calculated by integrating EV V2G power over time, with SOC obtained from initial energy and rated battery energy.The formulation defines battery-energy variation as the time integral of power.
- Model parameters: The model parameters for the interconnected two-area power grid are summarized in Table I.The table provides the basic parameters used by the simulation system.
B. Simulation Scenarios
The simulations model EV battery states, charging-station participation, and uncertain regulation dispatch in an interconnected two-area system. Results show that both V2G strategies suppress frequency fluctuations, while CS2 also achieves expected EV battery SOC levels.
- Simulation assumptions: EV battery SOC levels are sampled from a normal distribution using Monte Carlo simulation, with EV plug-in duration assumed to be 8:00–17:00.The model represents differing driving behaviors and residual battery energy levels at the parking lot.
- EV aggregation: The aggregator manages 100 EV charging stations, each serving 500 EVs, with TYPE I, TYPE II, and TYPE III behaviors.Scheduled V2G power is recalculated hourly at each charging station to ensure EV charging demands.
- Uncertain dispatch: The uncertain dispatch ratio R is modeled with a normal distribution, using μ=0.5 as an example in which the control center dispatches half.The preferred proportion depends on market uncertainty, EV randomness, and renewable-energy fluctuations.
- Frequency-regulation performance: 0.0175 MW RMS ACE is obtained by both CS1 and CS2, compared with 0.0252 MW without V2G.The corresponding maximum and minimum ACE values are 0.0599 and -0.0633 MW for CS1, and 0.0550 and -0.0606 MW for CS2.
- EV battery performance: CS2 guarantees expected battery SOC levels, whereas CS1 does not, because CS2 corrects scheduled V2G power in real time.The correction compensates for battery-energy changes caused by uncertain regulation.
- EV battery performance: Charging and discharging curves have nearly constant slopes because EV power is low over short intervals and regulation-up and regulation-down capacities are close.This behavior persists despite significant fluctuations in the load profile.
D. Control Performance under Different Mean Values of R
Increasing the mean of the uncertain dispatch ratio R improves ACE and frequency-deviation suppression because more regulation is assigned to faster-regulating EVs. The simulations compare R~N(0.5, 0.01) with R~N(0.8, 0.01).
- Control performance: A greater mean value of R produces better suppression of ACE and frequency deviations.More regulation is distributed to EVs, which have faster regulating characteristics than conventional generators.
- ACE performance: 0.0167 MW RMS ACE is achieved by CS2 with R~N(0.8, 0.01), compared with 0.0252 MW without V2G.CS1 achieves 0.0169 MW RMS ACE under the same dispatch distribution.
E. Robustness to Different Distributions of R
The control performance depends mainly on the mean and variance of R and is robust to the specific distribution when those parameters match. CS2 also remains robust when the variance of initial EV SOC increases substantially.
- Distribution effects: R~N(0.8, 0.01) suppresses ACE and frequency deviations better than R~U[0,1] because it has a greater mean and smaller variance.The comparison is supported by the ACE and frequency-deviation results in Tables VIII and IX.
- Distribution effects: When mean and variance match, R~U[0,1] and R~N(0.5, 1/12) produce very similar results, indicating robustness to the distribution form.The reported dependence is primarily on the mean and variance of R.
- Initial SOC robustness: When initial SOC variance increases from 0.01 to 0.1 with SOC bounded in [0.1, 0.9], CS2 still produces very small RMS deviations from expected SOC.The results support robustness to initial battery SOC across TYPE I, TYPE II, and TYPE III EVs.
- Conclusion: The proposed V2G control suppresses ACE and frequency fluctuations while guaranteeing expected EV battery SOC in simulations of an interconnected two-area power grid.The paper’s conclusion covers both grid regulation and EV-owner SOC objectives.