IEEE Electrification - September 2021 - 34

Importance of Unbalanced Three-Phase Models
Unbalance can be a significant concern in large-scale
charging systems. However, to date, most algorithms proposed
in the literature implicitly assume single- or balanced
three-phase operation. As we have seen, unbalanced
three-phase constraints are necessary to ensure safety.
These unbalanced models can significantly impact the
performance of an algorithm. To see this, we can use the
ACN Research Portal to evaluate the percentage of user
energy demands met when using balanced and unbalanced
models.
We compare six algorithms over a range of possible
transformer capacities based on the actual charging workload
of the ACN at Caltech in September 2018. For this
experiment, we use our ASA with an objective that promotes
charging as quickly as possible, which we denote
ASA-QC.
From Figure 6, we can see that in the balanced case,
EDF, LLF, and MPC all perform near optimally, exceeding
the performance of RR and FCFS by up to 8.6%. However,
in the unbalanced case, we see that, while ASA-QC can
match the offline optimal as before, EDF and LLF both
underperform. In the highly constrained regime, RR outperforms
EDF and LLF despite having less information
about the workload. We attribute these results to the
importance of phase balancing in three-phase systems,
which has been historically underappreciated in the managed
charging literature.
To understand why ASA performs so much better than
the baselines, we must consider what information each
algorithm uses. RR uses how many EVs are present and
performs the worst. EDF uses only information about
departure time, while LLF also uses the EV's energy
demand. Only ASA-QC actively optimizes over infrastructure
constraints, allowing it to better balance phases
(increasing throughput) and prioritize EVs, including the
current and anticipated congestion. A key feature of the
ASA framework is its ability to account for all available
information cleanly.
100
80
60
40
20
Figure 6 can also be used to evaluate the infrastructure
needs of a site. For example, we can see that, if a host
wants to deliver >99% of charging demand using ASA-QC,
a 70-kW transformer would be sufficient, assuming an
unbalanced three-phase system. Alternatively, if an existing
transformer can only support 40 kW of additional
demand, a host could expect to meet approximately 85%
of energy demands without an upgrade.
Interfacing With the Grid
Charging systems do not operate in a vacuum. In almost
all cases, they draw energy from the power grid. Because
of their enormous power and energy requirements, largescale
EV charging systems can significantly impact the
power grid.
The ACN Research Portal allows us to expand studies
like this to consider how more advanced smart-charging
approaches can help alleviate strain on the distribution
system, especially for large charging systems like workplaces.
To enable studies like this with ACN-Sim, we have
integrated it with several grid simulation packages, including
MATPOWER, PandaPower, and OpenDSS. In each case,
we can use ACN-Sim and ACN-Data to provide a realistic
load profile, which can then feed into the grid simulation
package that evaluates power flows and alerts us to any
voltage or overloading issues.
For this case study, we use OpenDSS to model a 240node
distribution system in the midwestern United States
with hourly smart meter data. Our goal is to evaluate the
effect of installing a 52-EVSE charging network at one
node in this distribution system. We use actual data from
the ACN at JPL on 6 September 2018. There are four cases:
a baseline with no EV charging, uncontrolled charging,
ASA with a load-flattening objective, and ASA with load
flattening and onsite solar.
The results of this experiment are shown in Figure 7.
RR
FCFS
EDF
LLF
ASA-QC
Offline Optimal
20 40 60 80 100
Transformer Capacity (kW)
(a)
20 40 60 80 100
Transformer Capacity (kW)
(b)
We can see that uncontrolled charging at this scale would
overload the distribution transformer and lead to unacceptably
low voltages in the network. However, using ASA
with load flattening, we can stay
below the transformer capacity and
above the voltage limits. Moreover,
we see that if 225 kWac of solar
were installed at the site, we
achieve the same circuit-wide minimum
voltage as the baseline case.
This indicates that onsite solar generation
and smart EV charging
could enable widespread workplace
EV charging without adverse
grid impacts.
Figure 6. The algorithm performance with constrained infrastructure with the (a) balanced and (b)
unbalanced models.
34
IEEE Electrification Magazine / SEPTEMBER 2021
Data-Driven Modeling
In addition to enabling trace-driven
simulation, we can also use the
ACN Research Portal to develop
Demand Met (%)

IEEE Electrification - September 2021

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