IEEE Electrification - September 2021 - 32

ACN Research Portal
Beyond serving as a model for smart-charging systems,
the ACN has led to the creation of the ACN Research Portal
(Figure 4). This portal has three parts:
x ACN-Data: a collection of fine-grained charging data
collected from the Caltech ACN and similar sites
x ACN-Sim: an open source simulator that uses ACNData
and realistic models derived from actual ACNs to
provide researchers with an environment to evaluate
their algorithms and test assumptions
x ACN-Live: a framework for safely field-testing algorithms
directly on the Caltech ACN.
ACN-Data
The ACNs at Caltech, NASA's Jet Propulsion Laboratory
(JPL), and elsewhere have produced a vast amount of data
that have enabled new research lines within our lab. However,
most researchers do not have access to data from
systems like the ACN, precluding them from applying
data-driven methods, such as machine learning and tracedriven
simulations, in their work. Instead, most existing
works focus on assumed data distributions or data collected
from internal combustion engine vehicles.
To meet this need in the community, we have collected
and published ACN-Data, which includes data from 207
level 2 EVSEs and six dc fast chargers (DCFCs) from seven
Simulation
Scenarios
ACN-Data
Algorithm
Validation
ACN-Sim
ACN-Live
Constraints
ACN
Figure 4. An overview of the ACN Research Portal.
clusters: five at Caltech (including one at a Laser Interferometer
Gravitational-Wave Observatory facility in Louisiana),
one at JPL, and one at an office building in northern
California. These clusters cover common use cases,
including campus/public-use and access-controlled workplace
charging. The smallest of these clusters has only
four level 2 chargers, while the largest has 78 level 2 chargers
and two DCFCs in a single parking structure.
The data set includes more than 80,000 sessions. Each
session includes the arrival time, departure time, energy
delivered, and user data from a mobile app, including the
vehicle information, estimated departure time, and estimated
energy request. The data set also includes time
series data, including the charging current, power, and
voltage, with 4-s resolution.
ACN-Sim
While the ACN has allowed us to identify many interesting
challenges in real-world EV charging systems, we recognize
that most researchers will not have access to such
physical testbeds. However, it is still important that they
have a realistic environment in which to evaluate their
algorithms. To this end, we developed ACN-Sim, a modular,
data-driven simulation environment for testing
scheduling algorithms for EV charging systems (Figure 5).
x Models: ACN-Sim includes realistic models of the
many components of a real EV charging system,
such as electrical infrastructure, charging stations,
and EV BMSs. These models incorporate our lessons
from real systems. For example, the charging network
constraints include three-phase unbalanced
models, while batteries use the constant-current,
constant-voltage charging scheme discussed previously.
The simulator is also designed to be highly
modular, meaning that users can replace each component
to model different types of hardware or levels
of fidelity.
Algorithms
Interface
Simulator
Charging Network
EVSE
Constraints
EV
Battery
Figure 5. The modular architecture of ACN-Sim.
32
IEEE Electrification Magazine / SEPTEMBER 2021
Event Queue
Events
Signals
ACN-Data
x Events: The simulator is event driven. To generate
events, users can get real event sequences from ACNData,
generate them from statistical models, or manually
create events to investigate edge cases. To make
accessing ACN-Data simpler for users, ACN-Sim provides
direct integration with the ACN-Data application
programming interface (API). This allows the user to
specify a site and date range, and ACN-Sim will gather
the actual workload from that ACN and generate the
appropriate plug-in and unplug events. ACN-Sim also
provides utilities for learning statistical models, such
as Gaussian mixture models (GMMs), directly from
data using tools from scikit-learn.
x Signals: The signals submodule allows ACN-Sim to
integrate with external signal sources, which can be
an important part of EV charging systems such as
utility tariffs, solar-generation curves, and external
loads. These signals are available to the scheduling
algorithm through the simulator's interface.
ACN-Sim
Data
Control Signals
IEEE Electrification - September 2021

Table of Contents for the Digital Edition of IEEE Electrification - September 2021
