Impacts of Plug-in Vehicles and Distributed Storage

on Electric Power Delivery Networks

Peter B. Evans, New Power Technologies Soorya Kuloor, PhD, Optimal Technologies (US),
Los Altos Hills, CA USA Inc.
peterevans@newpowertech.com Raleigh, NC USA
sooryak@otii.com

Benjamin Kroposki, PhD, P.E., National Renewable Energy Laboratory

Golden, CO USA
benjamin.kroposki@nrel.gov

Abstract This paper discusses studies funded by the National Conducting detailed analysis of renewable energy
Renewable Energy Laboratory (NREL) via SunPower, Inc. and system performance and grid effects through electrical
the California Energy Commission (CEC), performed by New transmission and distribution (T&D) system modeling and
Power Technologies and Optimal Technologies that showed that simulation.
high-penetrations of distribution-connected storage devices or
plug-in vehicles can have adverse grid impacts due to their Electric power storage connected to the power delivery
charging loads. Randomly-located or unmanaged additions, such network has the potential to improve the utilization of preferred
as plug-in vehicles, can also have greater impacts at lower generation sources and the performance of the grid itself by
penetrations when compared to managed additions such as supplying incremental capacity and energy on demand, e.g.,
utility-sponsored storage. The studies also found that potential during periods of peak demand or system stress. Other
adverse impacts from such charging loads are highly localized, investigators have demonstrated the use of capacity derived
and once identified are readily managed. The studies also show from demand response as a grid resource such as spinning
the use of a high-definition Energynet power system simulation reserve; storage offers similar potential, but with far greater
and AEMPFAST power system optimization software for practical operational range. Vehicle-to-grid (V2G) refers to the
identifying and managing the potential impacts of distribution- dual-use of plug-in battery electric or hybrid vehicles as storage
connected storage. capacity available to the grid for such purposes. The appeal of
V2G is the reduction in the effective cost to the power delivery
Keywords: Plug-in Vehicles; Battery Electric Vehicles; Plug-in
Hybrid Vehicles; V2G; Electrical Storage; Load Flow Analysis;
network of the storage asset.
Optimal Control; Power Distribution; Power System Simulation; Electric storage also places a corresponding charging load
Power Transmission; Distributed Storage, Distributed Generation on the network. Even where there are adequate electric supply
resources to serve charging loads off-peak, delivering the
I. INTRODUCTION power to these loads may introduce localized problems with in
the delivery network. Further, plug-in vehicles place a charging
The Department of Energy (DOE) Renewable System
load on the network whether or not they are used as V2G
Interconnection (RSI) Study addressed barriers to the
storage resources. Lastly, rapid charging, a clear customer
expansion of renewable energy technologies associated with
value for plug-in vehicles, translates to higher charging loads.
the possible impacts on the electricity grid of high penetrations
of these technologies. The distributed generation element of the The focus of this paper is the potential impacts of such
RSI Study is to address technologies that connect to the grid at charging loads on the power delivery network. For a given
the distribution level, including solar photovoltaic (PV), power delivery network, we sought to determine the
distributed wind, vehicle-to-grid (V2G), and distribution- penetration level at which the charging load of plug-in vehicles
connected storage. Among the issues related to renewable might begin to adversely impact the performance of the power
energy grid integration to be addressed under the RSI Study delivery network. We also sought to assess the role of the size
are: of individual storage additions on their impact on network
performance.
Understanding and developing solutions to utility
concerns about large penetration of renewable generation A fundamental distinction of plug-in vehicles and V2G
resources [including storage] connected to the electric power relative to electrical storage generally is that their location
system, and within the power delivery network is far less likely to be
planned by the power delivery network operator. So if power

978-1-4244-2601-0/09/$25.00 2009 IEEE 838

delivery networks are sensitive to charging loads at particular 651.809 MVAR, and off-peak totaled 492.45 MW and 222.40
locations, it is important for network operators have the means MVAR. Importantly for a storage study, the Hobby systems
to proactively identify those locations as well as the penetration off-peak load demands over 850 MW less from existing
levels at which these loads become a concern. generation resources than the systems on-peak load.
Customer-side distributed storage and plug-in vehicles are Within the Hobby system are 102 voltage-regulating
both likely to be connected at customer service voltages and devices (transformer taps and voltage regulators), 839 reactive
potentially at electrically-remote locations within the power power sources (capacitors), 30 embedded generation resources,
delivery network. Meaningful impacts of the charging loads of some of which are also sources of reactive power, and 4,684
such devices could be highly localized, thus their assessment individually-addressable demand response resources, so the
requires the ability to see the power delivery system at the simulated system provides a great deal of operational control.
circuit element level as well as system wide. For this work we We considered all of these resources as individually
used a high-definition Energynet model of the subject controllable to obtain the highest grid performance and to
system, comprised of over 100,000 nodes and integrating accommodate the addition of storage devices. In each case,
regional and local transmission, sub-transmission with all these resources were dispatched using AEMPFAST with an
distribution substations, line segments, and elements optimization objective of minimizing both real and reactive
characterized. This model was developed using ordinarily- power losses while minimizing voltage deviation from nominal
available data sources provided by the utility and is supported at each point in the Hobby portion of the network model. As
by an enhanced version of Positive Sequence Load Flow voltage is also a factor in losses it tends to have a strong
(PSLF) provided for our use by GE Energy. We also used influence with this objective. Also, with this objective overload
Optimal Technologies AEMPFAST power system relief is a factor, but only indirectly as it relates to losses.
optimization software which provides an assessment of the real
(active) and reactive power resource deficiency relative to For budgetary reasons this storage analysis was built from
optimal at each individual node, which, in turn, can be used studies already completed in which we populated the subject
to identify those locations where additions of storage capacity system with an Optimal DER Portfolio consisting of
(and charging load) will move the system closer to a specified capacitors and demand response resources (and distributed
optimum condition. generation in other studies). Each of these resources was
ideally sized and sited to contribute to the optimization
The following summarizes our experiences and findings in objective stated above, subject to external limits such as the
using these tools to study the impacts of the charging load of amount of demand response or distributed generation a
distributed storage and plug-in vehicles on the subject power particular type and size of customer would ordinarily host, and
delivery network, including the impact of size, penetration, and a circuit-level distributed generation limit that would avoid
whether the placement of these devices is unmanaged (e.g., unintended post-contingency islanded operation. We also
with plug-in vehicles) or managed (e.g. with utility-directed incorporated in the model all of the systems existing demand
storage deployments). response as discrete resources. So under on-peak conditions the
model incorporated 104 capacitor additions and under on-peak
II. THE HOBBY HIGH-DEFINITION SYSTEM MODEL conditions reflected the dispatch of existing and added demand
response comprising about 2.9% of load.
The subject system for this study was Southern California
Edisons Hobby system in Southern California as it stood in One consequence of the use of this initial model is the
mid-November, 2005. As part of a larger project funded by the remaining benefit to system performance of the addition of
CEC we developed a high-definition Energynet model of the incremental capacity under on-peak conditions, even at ideal
Hobby system that provides system-wide scope with locations, is attenuated. Accordingly, the potential grid benefits
distribution element-level granularity; this model enables the of storage and V2G as an incremental capacity resource on-
direct observation of the impacts of any simulated change, peak are understated in these results.
addition, or removal on each individual element in the network
and on the network as a whole. The Hobby system consists of III. INITIAL STORAGE CHARGING LOAD STUDIES
58 local transmission and sub-transmission substations, and
215 radial distribution circuits with voltages ranging from 33 As an extension of the study to identify the Optimal DER
kV to 2.4 kV. This system was modeled in full elemental Portfolio of distributed energy resource additions, we
detail, comprising nearly 100,000 buses, and integrated into a investigated the addition of distributed storage representing 2%
power flow model of the western United States provided by the of the subject systems peak load, nominally network operator-
Western Electricity Coordinating Council (WECC). The Hobby managed and sited and sized to provide capacity valuable to the
system has a peak load of about 1,300 MW and serves network under on-peak conditions while minimizing the impact
approximately 275,000 customers at about 46,000 individual on the network off-peak,. We postulated a storage device
capable of discharging for at least an hour at its rated capacity,
distribution transformers, each characterized by customer class.
and that would charge at 1.25x its discharge rating.
The model includes loads derived from actual data recorded Specifically, we simulated the addition of storage capacity
under a variety of operating conditions; for this study the increments one by one up to the 2%-of-peak-load limit, in each
modeled loads were from on-peak and off-peak hours of a day case at the individual node within the system where the active
that saw one of the highest load peaks recorded for the year. power resource deficiency on-peak netted against the active
The served-load modeled on-peak totaled 1,343.42 MW and power resource deficiency (or added to the active power

839
resource surplus) off-peak was maximized, both as determined kWh with a charging time of about four hours, and that is
in an AEMPFAST analysis. Storage capacity at these locations capable of a maximum-rate, short-duration partial discharge
would yield the greatest net benefit to the network or system during on-peak periods of 25 kW for as little as an hour (or 25
benefit relative to our stated optimization objective of kWh of the devices 65 kWh capacity).
minimizing losses while minimizing voltage deviation from
nominal. These benefits to the power delivery network would As stated above, a key distinction of V2G storage and plug-
be in addition to any bulk system capacity or load management in vehicles is the power delivery network operator arguably has
benefits provided by this storage capacity. little direct influence on the location within the network of the
charging loads of these devices. We simulated the placement of
We found that even relatively small (<100 kW nominal these devices at load-serving transformer sites randomly-
capacity) storage devices, when placed where their capacity selected from the approximately 46,000 in the Hobby system.
yielded the greatest system benefits on peak, could place
burdens on the power delivery system such as voltage At a penetration level of 1,400 V2G additions, the nominal
reductions of 5% or more under off-peak conditions due to total discharging or on-peak capacity of these devices is 35
their charging loads. While these conditions were evident in the MW. Off-peak they represent 22.4 MW of charging load. In
Energynet models distribution element-level granularity, they terms of customer sites, this represents a penetration level of
about 3%, and the nominal on-peak capacity of these additions
were highly localized and might not be visible at the circuit
level or in substation voltage reads. Notably, these impacts represents 2.6% of on-peak load. So even a small number and
occurred even when minimizing off-peak impact was a direct low penetration of V2G units, with limited on-peak availability,
factor in the placement of these storage projects. Also, these yields a meaningful amount of incremental on-peak capacity.
impacts were observed after allowing for the full range of The impact of these additions functioning as incremental
available compensating measures such as capacitor re-dispatch capacity on-peak and as load off-peak is tabulated in Table II.
and transformer and voltage regulator tap settings. Of course In both cases the available system controls have been
these network impacts also emerged well short of meaningful reconfigured using AEMPFAST to minimize losses and
off-peak loading of regional electric supply resources. voltage deviation per the objective. On peak, real power losses
and reactive power consumption are reduced relative to the
We did find that these impacts could be mitigated by system with no storage additions, and the system-wide
restricting the placement of storage devices though that might minimum voltage increases. Off peak, real power losses and
prevent the placement of storage capacity where it would reactive power consumption are increased relative to the
provide the greatest on-peak benefit or by reducing the size system with no storage, and the system-wide minimum voltage
(or charging rate) of storage additions. decreases. The reduced off-peak voltage remains within the
specified limit, however.
These initial findings prompted us to look more closely at
the potential impacts on network conditions of off-peak
charging loads of storage or plug-in vehicles. In particular, we TABLE II. 1,400 V2G ADDITIONS VOLTAGE AND LOSS IMPACTS
sought to assess the charging load impact of smaller but un- Operating Condition Vmin Real Losses Reactive Losses
managed additions, such as customer-driven V2G, and the
size and penetration limits of storage additions managed by the On-Peak + 0.017 PU - 3.523 MW - 9.495 MVAR
network operator. Table I summarizes the initial voltage and Off-Peak - 0.014 PU + 0.813 MW + 2.379 MVAR
loss conditions for these subsequent studies. For these studies
we imposed a low-voltage limit under off-peak conditions of
0.95 per unit (PU), or 95% of nominal voltage, at any point in At a penetration level of 10,500 V2G units, these devices
the network. That limit effectively allowed off-peak voltage represent a significant on-peak resource, nominally totaling
deterioration of 0.028 PU. We judged that a reduction in off- 262.5 MW of incremental capacity even give the operational
peak voltage to this level would be viewed as unacceptable by limits we postulated. Off-peak they represent 168 MW of
the network operator. charging load. In terms of customer sites, this represents a
penetration level of about 22%, and the nominal on-peak
TABLE I. INITIAL VOLTAGE AND LOSS CONDITIONS capacity of these units represents 19.5% of on-peak load. The
impact of these additions functioning as incremental capacity
Operating Condition Vmin a Real Losses Reactive Losses on-peak and load off-peak relative to the system with no
On-Peak 0.9187 PU 62.528 MW 179.129 MVAR additions is tabulated in Table III. Again, in both cases the
available system controls have been reconfigured using
Off-Peak 0.9788 PU 6.614 MW 14.614 MVAR AEMPFAST to minimize the losses and voltage deviation per
(a) System-wide minimum voltage the objective. On peak, real power losses and reactive power
consumption are further reduced, and the system-wide
minimum voltage increases. Off peak, real power losses and
IV. PENETRATION LIMITS OF V2G reactive power consumption are further increased relative to the
system with no storage, and the system-wide minimum voltage
For these studies we postulated a V2G storage addition decreases.
increment of 25kW of capacity on-peak with an off-peak
charging load of 16 kW. This was intended to generally
represent an electric-only plug-in vehicle battery of perhaps 65

840
 TABLE III. 10,500 V2G ADDITIONS VOLTAGE AND LOSS IMPACTS provide a more granular view of the evidently highly-localized
Operating Condition Vmin Real Losses Reactive Losses
impacts of such charging loads on grid voltage.
On-Peak + 0.0228 PU - 21.023 MW - 61.357 MVAR In each study we began with the initial cases and conditions
described in Table I. As with the studies described in Section
Off-Peak - 0.0513 PU + 7.051 MW + 17.833 MVAR III, we postulated a storage increment or device with the
capability to discharge for at least an hour at its nominal
capacity rating, and that would charge for at least an hour at
However, at a penetration level of 10,500 V2G units, the 1.25x its nominal rating. Again, these are postulated as
system-wide minimum voltage under off-peak conditions is not managed placements we sited these devices using a dual
maintained above the 0.95 PU threshold due to the charging AEMPFAST optimization that identified each successive
load, even with optimization of available voltage support storage site as the one at which the net on-peak network benefit
controls. less the off-peak network dis-benefit is maximized. As before,
We also evaluated a penetration level of 35,112 V2G the optimization objective was the simultaneous minimization
additions. In terms of customer sites this represents about 76% of real and reactive power losses and voltage deviation from
penetration. At this penetration level the V2G devices represent nominal. We also allowed multiple increments at each location.
a nominal total of 877.8 MW discharging or on-peak capacity. This allowed us to determine to some extent the ideal size of
Under on-peak conditions the capacity represented by these the storage project at each location that would provide the
units raises Vmin by 0.011 PU, reduces real losses by 24.979 system benefits we sought through the stated optimization
MW, and reduces reactive losses by 114.45 MVAR, in each objective. As with the prior studies we limited the impact of
case relative to the system with no storage additions. charging loads through a 0.95 PU limit for the lowest voltage at
any point system-wide under off-peak condition.
These 35,112 V2G units would represent 561.8 MW of
charging load off-peak. This is still less than the 850 MW A. Impact of Size
differential in initial system demand between on-peak and off-
peak loads. However, the important finding is that we found no The initial studies described in Section III showed that the
feasible power flow solution under off-peak conditions with charging loads of larger storage increments had greater system
this added charging load. The system collapses and is unable to impacts under off-peak conditions. We thus attempted to
handle the charging load of these units. identify a safe storage increment under managed placement.
We define this as the largest individual storage increment that
So we may conclude from the foregoing that even at low could be added to the subject system at locations chosen to
penetration levels of 1-2% and small (< 20 kW) sizes, maximize the net on-peak and off-peak network performance
randomly-placed charging loads associated with distributed benefit that would not also individually cause an off-peak
storage or plug in vehicles has a measurable effect on system voltage reduction that would violate the 0.95 PU off-peak low-
voltage and losses, and at penetration levels around 20% such voltage limit. For this study we specified the total nominal
loads could have a unacceptable impact on system voltage. capacity of storage additions at a modest amount, e.g., 1-3% of
Also, it is possible, though at possibly unreasonable penetration peak load. In practice this simulation might represent a network
levels, for distributed charging loads to collapse a power operators distributed storage strategy, and an overall storage
delivery system even if supply resources remain. budget of 1-3% of peak load might be indicative of the level of
It is important to note again that the on-peak impacts or embedded storage a utility might incorporate in a power
benefits of storage capacity presented here are probably delivery system for operational purposes such as regulation.
understated. With the initial system model populated with large Our focus in this study was on the system impact of individual
amounts of ideally-placed distributed resources, as described storage additions rather than the cumulative impact on the
above, the potential benefits from incremental capacity on-peak system of high storage penetrations.
are attenuated. We found that the Hobby system could accommodate
storage additions in individual increments of up to 70 kW
while maintaining an off-peak minimum voltage of 0.95 PU or
higher at every point. In experimenting with larger minimum
V. CHARGING LOAD IMPACTS OF MANAGED DISTRIBUTED increments, we found that the system could accommodate
STORAGE ADDITIONS storage additions in increments as large as 100 kW only if the
We performed a separate set of studies of the impact of acceptable off-peak low voltage limit was allowed to fall to
network storage additions in which the placement of each 0.90 PU or lower, and that the system could accommodate
addition within the power delivery network is actively additions in increments as large as 90 kW with an off-peak
managed by the network operator, and in this case targeted to voltage at .93 PU or lower. In all cases we re-dispatched all
minimize the off-peak burden to the network as well as available controls according to the same optimization objective
maximize the on-peak network benefit. These studies shed light to compensate for the charging load impact. So for this system,
not only on the value of managed placement of storage, but minimum storage increments of 90 or 100 kW are too large to
also on the roles of unit size and penetration in the grid impacts permit the system to maintain the 0.95 PU specified minimum
of off-peak charging loads generally, whether distributed voltage at every point even with targeted placement according
storage or plug-in vehicles. Most importantly, these studies to our criteria, and thus would not meet our criterion for a
safe maximum increment.

841
 More importantly, we found that the voltage impact of the
charging loads of these storage additions is not a general
system-level phenomenon, but is actually highly localized. We
found that there are four or five specific locations within the
modeled network that are the most sensitive to the addition of
real power load off-peak in terms of voltage reduction. These
few locations effectively define the largest single storage
increment that can be added to the system while maintaining a
pre-specified minimum off-peak system-wide voltage.
This finding implies a reasonable conclusion that there
isnt really one maximum storage and charging load size that
can be placed in a system without concern for causing a voltage
violation off-peak. Each location has a different voltage
tolerance for the addition of real load under off-peak
conditions. So framing the question the way we did forces the Figure 1. Tree and Grape Substations. Loon circuit lies between Tree
identification of a small number of controlling buses. substation (lower right) and Grape substation (upper left).
We also found that there is very small number of locations
in the subject system with off-peak real power resource surplus
conditions in other words, locations where the addition of
storage charging load would actually improve network
performance relative to the objective. Specifically, these are
locations with positive P Indices greater than about 0.20 as
measured by AEMPFAST. In this system these locations are
very rare, representing only about 0.4% of the buses in the
subject system, and these buses are located on only four
circuits. We also found that the addition of one storage project
(and its charging load) on one of these circuits appears to
reverse the off-peak real power resource surplus condition on
the rest of that circuits buses and on other circuits as well. The
storage locations on these few circuits are of course
consistently the highest-ranked sites within our screen netting
on-peak system benefit and off-peak dis-benefit as storage at
these sites yields benefits under both conditions; they are also
the most sensitive in terms of off-peak voltage impact.
The following example will illustrate how the addition of a Figure 2. Loon Circuit. Loon circuit extends from Tree substation to Grape
storage device in an otherwise beneficial location may be size- substation.
limited. It also shows the localized nature of storage impacts
and thus the importance of evaluating storage impacts on the
circuit element level as well as at the system level.
Loon circuit is one of four 33 kV circuits in the Hobby
system served from Tree 115/33 kV substation. Loon extends
from Tree substation toward Grape substation. Figure 1 shows
Tree and Grape substations, and Figure 2 shows Loon circuit
extending between them from Tree substation. Loon in turn
serves three lower-voltage substations, Puffin, Crane, and
Penguin. Penguin substation, in turn, serves Cormorant circuit.
Figure 3 shows Penguin substation and Cormorant circuit.

Figure 3. Penguin Substation and Cormorant Circuit. Loon circuit serves

three lower-voltage substations, Puffin, Crane, and Penguin. Penguin
substation, in turn, serves Cormorant circuit

842
 Figure 4 shows the active power resource deficit or surplus Loon and Cormorant Circuits
(P Index) under both on-peak and off peak conditions before 1.05
Off Peak Voltage Profile

any storage additions, along a single path from Tree substation

at the left axis, through Loon circuit, Penguin substation and
Cormorant circuit. Cormorant circuit has a P resource deficit, 1.025

or negative P Index, on-peak. It is also one of the areas with a

resource surplus, or positive P Index, off-peak. This means the 1

incremental on-peak capacity and the incremental off-peak load

of a storage device would improve network performance 0.975
Before Storage Additions

relative to the optimization objective under both on-peak and Four 70 kW Storage Adds Area-wide

off-peak conditions. Accordingly, Cormorant is one of the most

Penguin P.T. Substation

highly-ranked storage locations in the entire system under our 0.95

screening approach. Loon Circuit

(Penguin Path)
Loon Circuit
(1316E Path)

0.925

Loon and Cormorant Circuits 0.9

P Deficiency/Surplus Before Storage Additions
1
Figure 5. Loon-to-Cormorant Path Off-peak Voltage Profile before and after
Four 70 kW Storage Additions. The charging loads of 70 kW storage
additions drive Cormorant circuit voltage from above nominal to 0.95 PU.
0.5

This indicates that even though device 1361E on Cormorant

0
is an advantageous location for a storage device for its system
impacts under both on-peak and off-peak conditions, if the
charging load is larger than the 87.5 kW associated with a
Penguin P.T. Substation

Of f -peak P Index
-0.5
Peak P Index

Loon Circuit
(Penguin Path)
Cormorant Circuit
(1361E Path)
nominal 70 kW storage device, the off-peak voltage will fall
-1
below the acceptable range at that location. More importantly,
this indicates the 70 kW safe maximum storage increment for
this system might be established by a single point in a network
-1.5
comprised of over 100,000 nodes.

-2
The addition of 500 such storage increments of 70 kW
(nominal) each represents 35 MW of incremental on peak
Figure 4. Loon-to-Cormorant Path Initial Off-peak and On-peak Active
Power Resource Deficit/Surplus. Cormorant circuit has attractive storage sites
capacity, or about 2.6% of peak load. It also represents 43.75
having both an active power resource deficit (negative P Index) on-peak and MW of charging load off peak. The condition of the system
an active power resource surplus (positive P Index) off-peak. with the addition of this amount of storage at optimal locations
relative to the system with no storage additions is tabulated in
The specific location on Cormorant where a storage Table IV. Under off-peak conditions, real power losses and
addition would yield the greatest net on-peak and off-peak reactive power consumption both increase, as expected. There
system benefit is at device 1361E, which is at the end of path is also a clear impact on voltagethe system-wide minimum
shown in Figure 4. Figure 5 shows the off-peak voltage profile voltage decreases to just at the 0.95 PU low-voltage limit. This
in per-unit terms along the same Loon-to-Cormorant path is the influence of the 70 kW size of the additions acting on one
before any storage additions, and also after the addition of four or more of the sensitive locations as described above.
70 kW storage devices at chosen beneficial locations within the
Under on-peak conditions there is effectively no impact on
Hobby system, including a single 70 kW storage device at
voltage. We believe this is due to the large amount of ideally-
1361E on Cormorant. Before any storage additions the off-peak
place capacity associated with the Optimal DER Portfolio
voltage at the end of Cormorant is slightly elevated, consistent
resources already incorporated in the model in these studies.
with a real power resource surplus. After the four storage
On-peak losses decrease, and in this case more than with the 35
additions the off-peak voltage at the end of Cormorant drops
MW of randomly-placed V2G storage addition shown above
due to the charging load of the storage devices, including the
and in Table II. The greater loss reduction for an equal amount
one at that location. In fact, the voltage at that point touches the
of incremental capacity is due to the targeted placement of the
.95 PU low-voltage limit. Note as well that the Cormorant
70 kW storage additions in this case.
circuit voltage change at the substation is modest, thus the
impact of these charging loads on Cormorant circuit voltage is
only visible at the sub-circuit level. TABLE IV. 500 70 KW STORAGE ADDITIONS VOLTAGE AND LOSS
IMPACTS

Operating Condition Vmin Real Losses Reactive Losses

On-Peak - 0.006 PU - 4.537 MW - 12.829 MVAR

Off-Peak - 0.029 PU + 1.92 MW + 4.702 MVAR

843
B. Impact of Penetration locations within the distribution portion of the subject system
We also sought to identify the greatest penetration of are 420 kW. The largest projects overall were on 26 of the 115
storage projects where, if added to the subject system in safe kV local transmission buses and represented 4.2 MW each.
sizes and again at optimal locations chosen to maximize the These projects represent 877.8 MW of incremental capacity on-
net on-peak and off-peak network performance benefittheir peak and charging load of 1,097.25 MW off-peak.
cumulative impact was such that the system was still able to The impact of these 5,564 projects relative to the system
maintain the 0.95 PU low-voltage limit. In practice this might with no storage additions is tabulated in Table VI. Under off-
represent a network operators distributed storage deployment peak conditions real power losses and reactive power
going beyond the 1-3% of peak load range described above. consumption increase dramatically relative to the initial
Here one of the findings from the previously-described conditions. The system-wide minimum voltage remains at the
study comes in to play. We found that apart from the few very limit. Under on-peak conditions real power losses and reactive
sensitive locations previously identified, this system is capable power consumption are reduced and the system-wide minimum
of absorbing a large amount of storage charging load while voltage is slightly increased.
maintaining voltage off-peak within the limits we set, as long
as the storage increments are optimally-placed. TABLE VI. 12,540 70 KW STORAGE ADDITIONS (5,564 PROJECTS)
VOLTAGE AND LOSS IMPACTS
We added a total of 3,000 storage increments of 70 kW
(nominal). These additions represent incremental on-peak Operating Condition Vmin Real Losses Reactive Losses
capacity of 210 MW, or 15.6% of peak load, and charging load On-Peak - 0.013 PU - 33.689 MW - 130.371 MVAR
of 262.5 MW. These increments are again sited at optimal
locations to maximize net on-peak and off-peak network Off-Peak - 0.029 PU + 43.659 MW + 152.72 MVAR
benefits according to the screening method described above.
We permitted multiple increments of 70 kW at a single location
if that was shown to yield network benefits. Thus we arrived at These off-peak conditions represent very significant
a set of storage projects at 2,514 different locations within loading of the system due to the charging of the added storage
the network and having a range of sizes. 2,281 of the projects devices; the total load is actually greater than the total on-peak
(about 91%) are individual-increment 70 kW projects, and the load in the initial pre-storage conditions. The 1,097.25 MW
largest projects are nominally 420 kW (6 individual charging load is also substantially more than the 561.8 MW
increments). The sites of these projects are widely charging load associated with the addition of 35,112 V2G units
distributed, located on 174 different circuits and in one described above which caused voltage collapse. Yet in this case
substation. not only can the system support the load, but the low-voltage
limit is maintained at every location. This shows that the
The impact of these 2,514 storage projects relative to the system can absorb much more charging load if the placement is
system with no storage additions is tabulated in Table V. Under directed by the informed network operator.
off-peak conditions, real power losses and reactive power
consumption are increased. The system-wide minimum voltage The incremental capacity available on-peak from these
remains at the limit even with the additional charging load, 5,564 projects is identical to the 877.8 MW of on-peak capacity
though AEMPFAST results indicate that there is an increasing provided by the 35,112 V2G units described above. However,
number of buses near this limit. Under on-peak conditions real in this case the real power loss reduction of 33.689 MW far
power losses and reactive power consumption are reduced. The exceeds the real power loss reduction of 24.979 MW associated
system-wide minimum voltage is again largely unchanged. with the V2G units. Again, the difference is due to the targeted
placement of the storage projects in this case.
TABLE V. 3,000 70 KW STORAGE ADDITIONS (2,514 PROJECTS) The makeup of these 5,564 storage projects also supports
VOLTAGE AND LOSS IMPACTS some conclusions. These results do confirm that while
Operating Condition Vmin Real Losses Reactive Losses
maximum safe addition increment for this system remains at
70 kW (nominal), there are sites that can accept and benefit
On-Peak - 0.010 PU - 21.720 MW - 54.439 MVAR from larger storage additions. However, recall that each storage
Off-Peak - 0.029 PU + 10.4 MW + 29.12 MVAR
increment was placed according its net on-peak and off peak
network benefit relative to the optimization objective. That the
majority of the projects in these results are single-increment
We continued to increase the storage penetration to the 70 kW nominal projects indicates that few sites would derive
point where there were no remaining sites for a 70kW storage additional network benefits from larger projects. Further, the
addition that would not result in a violation of the .95 PU low larger substation-sited and transmission-level storage projects
voltage limit off-peak. This is effectively the penetration limit did not emerge until very high penetration levels were reached,
for projects of this size. At this point we had added a total of indicating that such projects yield relatively less in terms of net
12,540 units of 70 kW (nominal) of storage at 5,564 different system benefits than the smaller, more distributed projects that
project locations within the network. Again, these projects prevail at lower penetration levels.
are highly distributed; there are storage projects on nearly all of
the subject systems circuits. The majority, 3,904 sites (70%),
are single-increment 70 kW projects. The largest projects at

844
 VI. IMPLICATIONS FOR A SMART GRID might be more successful if the dispatch parameters and
The results in Section IV suggest that on a partial-capacity limitations can be tailored for each individual device and/or
dispatch basis, plug-in vehicle batteries in a V2G deployment customer. We have, somewhat arbitrarily, specified the V2G
can represent a meaningful amount of incremental distributed capacity that might be available from a plug-in vehicle battery
capacity for a power delivery system as penetration grows. The of a given size for purposes of these studies; clearly this
results in Sections IV and V show loss reductions and some relation is one with many, many factors.
voltage benefits from incremental capacity under on-peak Device-level management of distributed storage or V2G
conditions; further, as stated elsewhere, the results presented resources could be supported through utility Distribution
here probably understate the potential system benefits from this Management Systems or Advanced Metering Infrastructure,
incremental capacity under on-peak conditions. provided a) these systems are designed to accommodate such
In terms of charging load system impacts, the results functions, and b) the storage devices themselves are
presented in Sections III, IV, and V indicate that charging loads individually identifiable by and interoperable with these
associated with storage projects and plug-in vehicles are a real systems. Use of this device-level management capability will
concern in terms of their impact on the power delivery system, also require advanced system modeling and analytical tools that
and worthy of close evaluation. The results in Section IV show allow network operators to see conditions in their networks
that system was unable to maintain off-peak minimum voltage at the distribution element level and to gauge the impact of
with V2G charging loads at 20% of customer sites, and each of a very large number of resources locally and system-
collapsed with V2G charging loads representing less than the wide.
capability of available supply resources. The grid impacts of
plug-in vehicles at a given penetration level might be greater
still if their charging loads are concentrated in more VII. CONCLUSION
electrically-remote residential locations.
Distribution-connected electric storage, including plug-in
The results of Section V indicate that larger discrete sizes of vehicles as V2G, can provide benefits in terms of power
charging loads can have larger impacts, and the small size of delivery network performance, including loss reduction and
V2G loads postulated here is likely mitigates impacts. The voltage profile improvement. These benefits are in addition to
results of Section V also indicate that charging load impacts other storage benefits such as load management improved
can be substantially reduced through management of the resource utilization or ancillary services. These studies are not
placement of these loads, this may not be a realistic mitigation comprehensive in terms of their examination of all the potential
approach when it comes to plug-in vehicles. benefits of storage, and the benefits shown here are likely
The results of Section V also indicate that the system understated.
impacts of charging loads are most dramatic at a very small Charging loads of electric storage and plug-in vehicles have
number of discrete locations individual buses within a few two distinct sets of impacts on the electric power system;
individual circuits, at least in this system. Moreover, the rest of impacts on the power delivery system itself must be considered
this particular system demonstrated the capability to absorb in addition to the impacts of incremental demand on electric
very large charging loads. supply resources. Plug-in vehicles present a fundamentally
This suggests that electric power storage and plug-in different challenge in that the network operator has little direct
vehicles relying on off-peak charging even at high-kW rates control over the placement of these devices within the network.
remains viable from a power delivery system standpoint, and Electrical storage or plug-in vehicle charging loads have
that it may not be necessary for the grid operator to dictate the observable impacts on voltage and losses even at low
placement of the charging load of each individual storage penetration levels. These impacts are highly-localized
project or vehicle. It may be sufficient for the grid operator, charging load impacts appear at the sub-circuit or distribution
having identified the delivery systems sensitive locations, to element level. Even if these impacts are visible at the system
manage the charging rates of a few storage devices or vehicles level, assessment of charging load impacts at the system level
only if connected to the grid in those few known sensitive only or even at the circuit level may not provide sufficient
locations and possibly only under some conditions. detail to reveal the nature, extent, or location of meaningful or
Device-level storage charging rate control would also actionable impacts. Our re-optimization of all of these cases
enable device-level storage dispatch and its related benefits. A shows that these impacts cannot be overcome solely through
comparison of the on-peak loss impacts of 12,540 70 kW the redispatch of available system controls and resources.
storage additions in Section V and 35, 112 V2G additions in V2G can be a feasible approach to deploying electric
Section IV clearly shows that an equal amount of incremental storage, yielding both real benefits from incremental capacity
capacity on-peak yields greater results from specific, targeted and providing real charging load issues to manage. A V2G
locations. Dispatch of storage capacity to relieve overloads or profile in which a 65kWh plug-in vehicle battery is capable of
under post-contingency conditions would be even more discharging a portion of its capacity at a rate 25 kW for an hour
location-specific. Thus even a network operator that does not and charges at a rate of 16 kW for several hours represents
specify the network location of plug-in vehicles may seek enough incremental capacity and charging load to visibly
device-level control of V2G capacity associated with those impact the performance of the power delivery network both
vehicles. Also, the use of plug-in vehicles as V2G storage beneficially on-peak and adversely off-peak even at low

845
penetration levels of around 2%. As penetration grows the distributed storage units or plug-in vehicles (both charging and
incremental capacity serves to reduce system losses on peak, providing capacity as V2G devices), and storage devices
and the incremental charging load serves to increase losses and designed to be visible to and interoperate with these systems,
reduce the overall system-wide minimum voltage off-peak. We may be of significant value in the wide deployment of electrical
found that due to the charging load, for this power delivery storage. This work demonstrates the integration needs for V2G
system there is a penetration level at which the system cannot and distributed storage as the Smart Grid of the future
maintain a specified 0.95 PU minimum voltage at every point, continues to evolve.
even with compensating capacitor or tap adjustments. There is
a higher penetration level at which the system collapses even
with supply resources remaining.
ACKNOWLEDGMENT
The location of storage within a power delivery network is
a key factor in its impacts on the network. Storage additions This technical effort was performed in support of the
placed at or dispatched from specific locations where their net Department of Energys Renewable System Interconnection
on-peak and off-peak system impacts are shown to be initiative under a subcontract to SunPower, Inc. A portion of
maximized yield greater system benefits on-peak and have this work was performed in support of the California Energy
reduced impacts off-peak for a given nominal capacity. The Commissions Public Interest Energy Research in energy
charging load carrying capability of a network is far greater if systems integration under PIER Contract 500-04-008.
the placement of those loads is managed to maximize net The authors also acknowledge the in-kind support of
system benefits. Southern California Edison for substantial technical review and
The size of each storage increment is a factor. There is a generous availability of system data.
maximum size of storage addition that can be placed in certain
of a systems otherwise attractive storage sites. If this DISCLAIMER
increment is exceeded the system will be unable to maintain its This report was prepared as a result of work sponsored by
voltage within specifications at every location. In these studies the California Energy Commission (Energy Commission). It
this maximum was well over the size of the V2G additions, and does not necessarily represent the views of the Energy
the probable size of plug-in vehicle charging loads likely Commission, its employees, or the State of California. The
mitigates their impact. At the same time, where storage Energy Commission, the State of California, its employees,
additions were sited and sized for their net impact on network contractors, and subcontractors make no warranty, express or
performance, the sites showing benefits from very large storage implied, and assume no legal liability for the information in
additions were fewer and lower-ranking in terms of system this report; nor does any party represent that the use of this
benefits than those showing benefits from smaller, more information will not infringe upon privately owned rights. This
distributed additions. report has not been approved or disapproved by the Energy
The management of storage impacts is a highly-localized Commission nor has the Energy Commission passed upon the
affair. These studies show the voltage impact of charging loads accuracy or adequacy of the information in this report.
can be highly localized in this system the system-wide
voltage impact of charging loads was determined by the impact REFERENCES
of charging loads in a few discrete locations on a few circuits.
If charging loads exceeded a given threshold in these sensitive
[1] Kroposki, B.; Margolis, R.; Kuswa, G., Torres, J.; Bower, W.; Key, T.;
locations the system was unable to maintain its voltage within Ton, D.; Renewable Systems Interconnection: Executive Summary,
specifications. At the same time, if the critical charging load NREL Technical Report (NREL/TP-581-42292); February, 2008.
levels at these locations are not exceeded, the system overall Available at: http://www1.eere.energy.gov/solar/solar_america/rsi.html
shows great capability to absorb charging loads without further [2] Eto, J.; Demand response spinning reserve demonstration; Earnest
adverse voltage impact. Therefore, the identification of these Orlando Lawrence Berkeley National Laboratory, Environmental
sensitive locations, and possibly the management of individual Energy Technologies Division; LBNL-62761; May, 2007.
charging load sizes or rates at those locations, frees up the [3] Kirby, B.; Ancillary services: technical and commercial insights;
prepared for Wrtsil; July, 2007.
ability of the rest of the system to support significant amounts
[4] Evans, P.; Optimal Portfolio Methodology for assessing distributed
of high-kW rate charging loads. Advanced power system energy resources for the Energynet; CEC-500-2005-096, March, 2005.
modeling that allows engineers to see the system in element- Available at http://www.energy.ca.gov/2005publications/CEC-500-
level detail and advanced analytics that allow engineers to 2005-096/CEC-500-2005-096.pdf
accurately assess the resource deficiency or surplus at each [5] Evans, P.; Hamilton, S.; Dossey, T.;High-definition modeling of
individual node in the system can be of great value in electric power delivery networks; Conference Proceedings,
identifying and effectively managing around these sensitive DistribuTECH Conference and Exhibition, Feb. 3-5, 2009, San Diego,
CA; Pennwell Corporation, February, 2009.
locations within a system.
Utility Distribution Management Systems or Advanced
Metering Infrastructure systems designed to manage individual

846