CoChat
Sign Up
Cited by 174Electric Vehicles and InfrastructureSmart Grid Energy ManagementElectric Power System Optimization

The DOI matches its Crossref record; 34 of the 34 references it cites were found in a registry or an index; none of the 45 sources its claims cite is checked for support yet.

Adaptive Charging Networks: A Framework for Smart Electric Vehicle Charging

Open access
Lee, Zachary J., Lee, George, Lee, Ted, Jin, Cheng, Lee, Rand, Low, Zhi, Chang, Daniel, Ortega, Christine, Low, Steven H.·2021·IEEE Transactions on Smart Grid
DOI 10.1109/tsg.2021.3074437
Open this paper in CoChat — free accountOpen PDF

Abstract

We describe the architecture and algorithms of the Adaptive Charging Network (ACN), which was first deployed on the Caltech campus in early 2016 and is currently operating at over 100 other sites in the United States. The architecture enables real-time monitoring and control and supports electric vehicle (EV) charging at scale. The ACN adopts a flexible Adaptive Scheduling Algorithm based on convex optimization and model predictive control and allows for significant over-subscription of electrical infrastructure. We describe some of the practical challenges in real-world charging systems, including unbalanced three-phase infrastructure, non-ideal battery charging behavior, and quantized control signals. We demonstrate how the Adaptive Scheduling Algorithm handles these challenges, and compare its performance against baseline algorithms from the deadline scheduling literature using real workloads recorded from the Caltech ACN and accurate system models. We find that in these realistic settings, our scheduling algorithm can improve operator profit by 3.4 times over uncontrolled charging and consistently outperforms baseline algorithms when delivering energy in highly congested systems.

Overview

AI snapshot

Written from the paper’s full text.

TL;DR

The paper describes the Adaptive Charging Network (ACN), a deployed smart EV charging architecture, and its Adaptive Scheduling Algorithm (ASA), which uses convex optimization and model predictive control to manage large-scale charging under real infrastructure constraints. Using Caltech ACN data and simulations, ASA improves energy delivery in congested systems and can increase operator profit by 3.4 times over uncontrolled charging.

Key findings

  • The ACN was first deployed at Caltech in early 2016 and operates at over 100 other U.S. sites; the Caltech ACN grew from 54 custom level-2 EVSEs to 136 commercial level-2 EVSEs plus six 50 kW DC fast chargers, delivering over 1,103 MWh as of July 7, 2020.
  • In profit-maximization simulations using September 2018 data and SCE TOU EV-4 tariffs, ASA achieved $2,835/month (98.1% of offline optimal) in an idealized setting and $2,600/month (90% of offline optimal) with non-ideal batteries and EVSEs, compared with $763/month for uncontrolled charging—a 3.4× increase.
  • For energy delivery under constrained infrastructure, ASA-QC performed within 0.4% of optimal in scenario II and outperformed baselines by as much as 14.1% compared with EDF at 30 kW capacity; non-ideal EVSEs reduced ASA-QC performance, while non-ideal batteries had relatively small effect.
  • Providing a peak hint to ASA-PM increased met energy demands to 97.8% versus 95.6% without hints; ASA-PM was within 3.6% of optimal profit, and with the hint within 1.9%.

Methods

  • The paper presents the ACN architecture and ASA, an online model predictive control and convex optimization framework with configurable utility functions and constraints for unbalanced three-phase infrastructure, EVSE pilot-signal quantization, and non-ideal battery behavior.
  • Data come from the Caltech ACN from May 1 to October 1, 2018, when only 54 EVSEs on transformer t1 were active; this period included 10,415 charging sessions and 92.78 MWh delivered. Simulations use ACN-Sim with 5-minute time slots, a 12-hour optimization horizon, and five scenarios in Table II.
  • The analysis compares ASA variants against least laxity first, earliest deadline first, round-robin, uncontrolled charging, and an offline optimal benchmark under varying transformer capacities (20–150 kW) and SCE TOU EV-4 tariff and demand-charge conditions.

Limitations

  • The paper states that due to limited space it cannot address all possible ACN and ASA use cases; it assumes drivers provide accurate departure-time and energy-request estimates for simplicity; it does not consider V2G; DCFCs are not directly controlled; and load/generation forecasting and future-arrival information are outside scope.
  • The text makes evident that ASA performance is evaluated through simulations using historical Caltech ACN data rather than live field tests of the scheduling algorithm, and the workload data come from a period when charging was free and only 54 EVSEs on one transformer were active.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1109/tsg.2021.3074437

Cites

34 of 34 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 of 45 cited sources checked · 0 supported · 0 flagged

Show all 79 claims

  • Probably not a substantial claim. 72
  • It may state two separate claims as one. 43
  • The quote only partly states this claim. 1
98 of 98

  • The paper relies on this without support.

    Most sites are unable to install more than a few charging ports due to limited infrastructure capacity and fear of high electricity bills.

    “However, with current technology, most sites are unable to install more than a few charging ports due to limited infrastructure capacity and fear of high electricity bills.”

  • The paper relies on this without support.

    Many large-scale charging sites will provide charging during daytime hours.

    “Since many of these sites will provide charging during daytime hours, they can make use of abundant solar energy production and enable EVs to provide grid services throughout the day.”

  • The paper relies on this without support.

    Smart charging will substantially reduce the environmental footprint of transportation.

    “Smart charging will substantially reduce the environmental footprint of transportation while unlocking immense potential for demand-side management.”

  • The paper relies on this without support.

    Providing charging at diverse sites can reduce range anxiety and provide an alternative to personal charging ports for those who cannot install them at their homes.

    “Doing so can reduce range anxiety and provide an alternative to personal charging ports for those who cannot install them at their homes.”

  • The paper relies on this without support.

    Driver demands on the system are stochastic.

    “due to the stochastic nature of driver demands on the system”

  • The paper relies on this without support.

    Most EVSEs on the market today only support a finite set of pilot signal values and require quantization of the control signal.

    “most EVSEs on the market today, including the ClipperCreek, AeroVironment, Webasto, and Tesla EVSEs in the Caltech ACN, only support a finite set of pilot signal values and require quantization of the control signal.”

  • The paper relies on this without support.

    Human behavior can add significant randomness to the system.

    “Human behavior can add significant randomness to the system.”

  • The paper relies on this without support.

    Drivers may arrive, depart, or change their input parameters at any time.

    “Drivers may arrive, depart, or change their input parameters at any time.”

  • The paper relies on this without support.

    Most previous work on charging algorithms has focused on single-phase or balanced three-phase systems, making them inapplicable to charging systems like the ACN.

    “While many algorithms have been proposed to handle charging with an aggregate power limit or even a hierarchy of limits, most previous work has focused on single-phase or balanced three-phase systems, making them inapplicable to charging systems like the ACN.”

  • The authors state this as an assumption.

    Each charging EV has a unity power factor.

    “We assume in this model that each charging EV has a unity power factor”

  • The authors state this as an assumption.

    The voltage phase angles are separated by ±120°.

    “which we assume are separated by ±120 •”

  • The paper relies on this without support.

    For level-2 EVSEs, each EV's onboard charger and battery management system (BMS) controls its charging rate.

    “For level-2 EVSEs, each EV's onboard charger and battery management system (BMS) controls its charging rate.”

  • The paper relies on this without support.

    Requiring the charging rate to be non-zero from when a car plugs in until its charging demand is met helps prevent contactor wear in the EVSEs.

    “This constraint helps prevent contactor wear in the EVSEs and improves user experience since most vehicles will notify their owner when charging stops.”

  • The paper relies on this without support.

    Most vehicles will notify their owner when charging stops.

    “since most vehicles will notify their owner when charging stops.”

  • The paper relies on this without support.

    Current charging hardware does not provide state of charge information.

    “Since this information is not available with current charging hardware, we use a model-free approach to estimate battery behavior in real-time and use closed-loop control to account for modeling errors.”

  • The authors state this as an assumption.

    Nominal voltage is assumed when expressing r_i(k) in amps and e_i(k) and ê_i(k) in δ × amps.

    “For simplicity of notation, we express r i (k) in amps and e i (k) and êi (k) in δ × amps, assuming nominal voltage.”

  • The authors state this as an assumption.

    The maximum time between scheduler calls is set to the length of one period, so only the first charging rate in r* will be applied.

    “For simplicity, we assume that the maximum between scheduler calls (see Algorithm 1) is set to the length of one period, so that only the first charging rate in r * will be applied.”

  • The authors state this as an assumption.

    The system does not consider V2G.

    “Constraints (5a) ensure that the charging rate in each period is non-negative (we do not consider V2G) and less than its upper bound defined by the EV's BMS and the maximum pilot supported by the EVSE.”

  • The paper relies on this without support.

    Future load L(t) and on-site generation G(t) can be predicted for the profit objective.

    “Since L(t) and G(t) are unknown for t > 0 this formulation relies on a prediction of these functions into the future.”

  • The paper relies on this without support.

    Equal sharing among EVs minimizes line losses along the lines feeding each EVSE.

    “This regularizer also promotes equal sharing among the EVs, which is desirable for the operator and drivers and minimizes line losses along the lines which feed each EVSE.”

  • The authors state this as an assumption.

    Drivers provide accurate estimates of their departure time and energy request.

    “For simplicity we assume that drivers provide accurate estimates of their departure time and energy request.”

  • The authors state this as an assumption.

    Ideal batteries follow the pilot signal exactly.

    “Finally, ideal batteries are assumed to follow the pilot signal exactly.”

  • The authors state this as an assumption.

    It is feasible to deliver a minimum charging rate to each EV.

    “This constraint implicitly assumes that it is feasible to deliver a minimum charging rate to each EV, thus charging infrastructure should be designed with this constraint in mind if the operators want to ensure a minimum charging rate to each EV.”
23 of 23

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Flagged first, then by citations

Show all 32 references

Research works printed but not linked: 4

  1. lee, 2021, acnportal-Experiments
  2. 2017, General Service Time-of-Use Electric Vehicle Charging—Demand Metered
  3. 2020, Caltech ACN Dashboard
  4. 2020, Global EV outlook 2020: Entering the decade of electric drive?, International Energy Agency

Cited by

Sorted by citations

Items 101–174 of 174

Related

See all

Metadata

Authors
Lee, Z. J., Lee, G., Lee, T., Jin, C., Lee, R., Low, Z., Chang, D., Ortega, C., & Low, S. H.
Affiliations
California Institute of Technology (US)
ORCID
Venue
IEEE Transactions on Smart Grid
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
2021-09
Volume
12
Issue
5
Pages
4339-4350
ISSN
  • 1949-3053 (ISSN-L)
  • 1949-3061
DOI
10.1109/tsg.2021.3074437
OpenAlex
W3112320897
Semantic Scholar
5de6a0fef702431a52a7d1724499404a9ac66aee
Cited by
174
References
32
Citations by year
2026: 24 · 2025: 43 · 2024: 41 · 2023: 37 · 2022: 23 · 2021: 6
FWCI
7.70
Type
journal article
Language
en
Open access
Licence
Locations
  • IEEE Transactions on Smart Grid (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion, other-oa)
  • arXiv (Cornell University) (submittedVersion)
  • CaltechAUTHORS (California Institute of Technology) (submittedVersion, cc-by)
  • arXiv (Cornell University)
Funders
  • National Science Foundation through the Graduate Research Fellowship Program (1745301)
  • NSF AIR-TT (1602119)
  • NSF Division of Electrical, Communications and Cyber Systems (1932611)
  • NSF Division of Computing and Communication Foundations (1637598)
  • Resnick Sustainability Institute Graduate Fellowship
  • Caltech Rocket Fund
  • Caltech CI2 Grant
  • Emerging Technologies Coordinating Council of Utilities
  • Well Fargo/NREL IN2
Topics
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Electric Vehicles and Infrastructure
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Smart Grid Energy Management
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Electric Power System Optimization
Keywords
electric vehicle charging, charging scheduling, convex optimization, model predictive control, deadline scheduling, real-time monitoring and control

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-10-23, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-23 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review