Sie sind auf Seite 1von 5

Smart Grid Congestion Caused by Plug-in Electric

Vehicle Charging
Rachel Jarvis and Paul Moses
Department of Electrical and Computer Engineering
University of Oklahoma
Norman, USA
{jarvis & pmoses}@ou.edu

Abstract—Plug-in electric vehicles (PEVs) will place a sizeable loads are already high. Elements of local distribution circuits
load on future residential distribution systems. As the popularity like feeder cables, distribution transformers and substation
of PEVs increases, the network congestion of uncoordinated PEV equipment may experience levels of stress higher than ever
charging will likely also increase. Until smart charging becomes
available, the impact of uncoordinated charging on residential before due to the increase in PEV charging.
distribution systems must be considered. This paper studies This paper studies a range of network congestion situations
the effects of PEV charger activities of randomly populated, from randomly located PEV charging at random times using
randomly arriving PEVs within several low voltage residential a 24-hour load flow analysis executed with 5-minute time
networks. The impacts on network voltages, feeder congestion intervals and directs analysis towards the impact on distri-
overloads and distribution transformer loading are investigated.
bution transformer performance. Distribution transformers are
Index Terms—congestion, plug-in electric vehicles, smart grids crucial elements of the grid, so it is necessary to investigate
PEV chargers effects on them. This examination is carried out
with a 1200 node distribution system composed of several
I. I NTRODUCTION low voltage networks connected through distribution trans-
As the capacity of batteries for electric vehicles increases formers to a 23 kV feeder. This paper studies several PEV
while the cost of them decreases, more consumers are buying penetrations, random PEV locations, and random charging
electric vehicles. In 2016, the market for electric vehicles times. The outcome of this work can provide awareness of
grew by 40% [1]. Although some public charging options potential future PEV loading and the subsequent impact on
have been developed, there are six times as many electric capacity constraints and bottlenecks in critical links such as
vehicles as there are public charging stations [1]. This means transformers.
most charging happens in private locations, especially at the
II. M ODELING A PPROACH
customers residence, which is the focus of this paper.
Currently, there is no system in place to regulate when this The effects of PEVs on power distribution systems have
charging occurs, though some suggestions have been made been studied widely over the past decade. Many studies
[2]–[4]. However, there are several obstacles to overcome employ load-flow based analyses, but consider very simplified
before it becomes viable to implement a widespread charging driving pattern statistics. For instance, in a previous study in
scheduling system. The major issue is the lack of an economi- [9], random temporal PEV charger behavior on a distribution
cal communications infrastructure for utility demand response feeder was modeled over different charging time periods,
control for plug-in electric vehicle (PEV) chargers. For the however, the location of PEVs was fixed a priori for different
time being, these vehicles are charging in an uncoordinated penetration levels. Furthermore, a Gaussian probability distri-
manner as their owners plug them in. It is unknown when bution was used to model the random temporal charging times
a charging coordination solution will be implemented, so the of individual PEVs which is not an appropriate probabilistic
uncoordinated loading patterns’ impact on today’s distribution model for PEV time of arrival statistics.
systems must be analyzed. In the proposed work, stochastic behavior of PEV charging
The load of PEV chargers is considerable, and consequently, arrival locations and times are considered. For the arrival time,
utilities may be facing challenges of serious network stresses a more realistic Poisson distribution which factors the rate
and overloads, diminished power quality and dependability of of PEV arrivals at different times of the day is considered.
the distribution system [5]–[8]. It is likely that PEV owners The variable λ is used to represent the total number of
will plug in their vehicles to charge upon arriving home from PEVs divided by the number of 5-minute timesteps within the
work. As a result, the majority of PEVs will begin charging designated charging time span. The probability distribution is
in a small window of time in the evening, when residential calculated as follows:

e−λ λx
978-1-5386-9284-4/19/$31.00 ©2019 IEEE P (x) = , (1)
x!
where P (x) is the probability of a PEV arriving within time C. Plug-In Electric Vehicles Charger Load Profile
interval x. Based on the best-selling models of fully-electric vehicles in
After randomly selecting which nodes will have PEV charg- 2017, the storage capacity of these batteries exceeds the energy
ing occurring, (1) is used to determine how many PEVs will be consumed in the average daily distance driven in the United
arriving at their respective charging stations at each timestep. States. Even though there is a large variation in the size of
Next, the order in which the PEVs arrive is randomized. batteries, the energy consumed per kilometer driven is similar
Finally, the loads created by the PEV charging are added to a for all the models. Based on a 90% charging efficiency, 9.9 kW
typical daily residential load curve, and the load flow analysis is required to replenish the energy consumed in a day’s drive,
is computed for each time step. The random temporal events of calculated with a weighted average of the top five best-selling
PEV arrivals are discretized over 5 minute timesteps over a 24 vehicles energy required to drive 47 km. A linear discharge
hour period. The PEV charger loads are modelled as constant rate is assumed [10].
real power loads in the load-flow solution. It is assumed that
the battery charger circuitry draws ac power at unity power D. Ratings of PEV Chargers
factor which is achievable with sophisticated power electronic A level 2 charger with an output of 6.6 kW is implemented
converter schemes. In this way, more realistic power flows are in this analysis. This output is achieved with a dedicated, 240
considered leading to improved voltage profile determination V household circuit. This moderate charging speed becomes a
and loading stresses on distribution feeder elements such as primary load of the user’s household. Faster level 2 chargers
cables and transformers. and level 3 chargers are not included in this study because
III. S MART G RID T EST S YSTEM F OR R ANDOM they are not cost-effective for a single user and require special
U NCOORDINATED PEV C HARGING wiring installation [10].
A. System Under Study E. PEV Penetrations and Charging Time Zones
In these simulation studies, the results are based on a Simulations are executed for PEV penetrations of 15%,
modified version of an IEEE test system [9]. The original 30%, 50%, and 70%. This random, uncoordinated charging
system has been modified by adding low voltage sections to exhibits potential loading levels in the near and long-term
the medium voltage network. Each low voltage section consists future. For this study, PEV penetration is the ratio of nodes
of 53 buses and has been repeated 22 times by attaching these with PEV charging activity to the total number of low-voltage
laterals to selected buses of the medium voltage system. The residential nodes. The PEVs are placed randomly throughout
line resistance and reactance data have been obtained from [9]. the network, and a Poisson distribution is used to randomly
B. Load Profile for Residential Households determine arrival times. Charging activity is restricted within
each of the following three time distributions:
In this study we assume that each residential house follows
• 1700h-2200h: This time span is representative of PEV
a fixed load curve. This is a typical loading pattern based on
true loading data of a distribution transformer which shows owners charging their vehicles as soon as or shortly after
peak usage of electricity during the late afternoon as people they arrive home in the evening. This feasible distribution
come home from work. For this study, this load curve is overlaps with a large portion of the residential on-peak
taken as a baseline without any PEV charging. In subsequent period.
• 1700h-0200h: If incentives were offered to PEV owners
simulations, the impact of uncoordinated electric vehicles on
the overall substation demand as well as individual distribution such as lower electricity rates during partially off-peak
transformer loadings are investigated. hours, some consumers may be convinced to delay their
charging until later at night. Many of today’s chargers
offer the ability to schedule charging at a later time, so
this would be simple for users to implement.
• 1700h-0800h: Without additional charger programming,
the achievement of this wide and mostly off-peak charg-
ing distribution is highly unrealistic.
F. Distribution System Transformers
The low voltage sections in these simulation studies are
supplied by 23kV/415V distribution transformers (DT-10 to
DT-31). These distribution transformers are rated at 100kVA
and are based on the impedance model given in [9].
IV. R ESULTS
Simulations are carried out for a variety of PEV penetration
Fig. 2. Typical daily load curve assumed as the load without PEVs for each levels (15%, 30%, 50%, and 70%) within residential networks.
household in this study. The location of PEVs is randomly selected for each trial.
Fig. 1. The 1200 node smart grid distribution system consisting of the IEEE 23 kV 31 bus system connected through 22 distribution transformers (DT-10 to
DT-31) to low voltage 415 V networks (53 nodes each) populated with PEVs.

Fig. 3. Voltage profile at worst node (20-R33) with random, uncoordinated Fig. 5. Voltage profile at worst node (10-R33) with random, uncoordinated
charging (50% PEV penetration) taking place between 1700-0800 hrs charging (50% PEV penetration) taking place between 1700-0100 hrs

Fig. 4. Distribution transformer daily power flow with random uncoordinated Fig. 6. Daily distribution transformer power flow with random uncoordinated
charging (50% PEV penetration) taking place between 1700-0800 hrs charging (50% PEV penetration) taking place between 1700-0100 hrs
TABLE I
D ISTRIBUTION T RANSFORMER P OWER L OADS

Fig. 7. Voltage profile at worst node (18-R33) with random, uncoordinated


PEV charging (50% PEV penetration) taking place between 1700-2200 hrs

a. Daily average of all distribution transformer power loads


b. Distribution transformer at which maximum power magnitude occurs
c. Time at which maximum power magnitude occurs (to the nearest 5
minutes)

PEV penetration, Table II denotes that the voltage is deviating


Fig. 8. Daily distribution transformer power flow with random uncoordinated below regulation limits.
charging (50% PEV penetration) taking place between 1700-2200 hrs. Charging randomly spanning 1700-0100 hrs is more at-
tainable. PEV owners will likely plug in their vehicle upon
Additionally, random charging is investigated within three arriving home for the day, but through incentives, such as
time distributions (1700-2200hrs, 1700-0100hrs, and 1700- lower electricity costs during off-peak hours, owners may
0800hrs). A Newton-Raphson based 24-hour load flow analy- be persuaded to delay charging for a few hours. In this
sis with 5-minute time intervals is used for all calculations for charging span, Table I indicates the maximum distribution
network voltage levels and power demands. All simulations transformer power load ranges from 24.4% to 131% higher
and load flow algorithms are programmed in MATLAB [11]. than the nominal maximum power load at 15% and 70% PEV
penetrations, respectively.
V. D ISCUSSION
The most likely charging situation occurs when PEV owners
It is improbable that random PEV charging would take place immediately plug in and begin charging their vehicles upon
over 1700-0800 hrs without remotely controlling when the arrival home, with charging occurring over 1700- 2200 hrs.
charging of each PEV begins. Another method to potentially For low PEV penetrations of 15% and 30%, the increase from
achieve this time distribution would be to program each no-PEV levels of the maximum distribution transformer power
charger to begin charging at a random time within the desired load is 58.3% and 110%, respectively. This increase is more
window. Table I indicates that such randomization of delays significant because more charging is coinciding with on-peak
would aid in reducing the load on distribution transformers residential load hours.
and reduce the magnitude of load peaks. In this dispersed
charging scenario with 15% PEV penetration, Table I indicates VI. C ONCLUSION
a maximum power load 12.2% higher than the nominal load. This paper outlines the impact of random, uncoordinated
This maximum power load increases to more than double the PEV charging on residential low-voltage networks and the
nominal maximum with a 70% PEV penetration. Even for 15% associated distribution transformers. The results are derived
TABLE II [4] T. Zhao and Z. Ding, ”Distributed Initialization-Free Cost-Optimal
VOLTAGE P ERFORMANCE Charging Control of Plug-In Electric Vehicles for Demand Manage-
ment,” in IEEE Transactions on Industrial Informatics, vol. 13, no. 6,
pp. 2791-2801, Dec. 2017.
[5] K. Moslehi and R. Kumar, ”A reliability perspective of the smart grid,”
IEEE Transactions on Smart Grid, vol. 1, no. 1, pp. 57-64, 2010.
[6] A. R. Metke and R. L. Ekl, ”Security technology for smart grid
networks,” IEEE Transactions on Smart Grid, vol. 1, no. 1, pp. 99-107,
2010.
[7] P. S. Moses, S. Deilami, A. S. Masoum, and M. A. S. Masoum,
”Power quality of smart grids with plug-in electric vehicles considering
battery charging profile,” in IEEE PES Conference on Innovative Smart
Grid Technologies Europe, Chalmers Lindholmen, Gothenburg, Sweden,
2010, pp. 1-7.
[8] A. Ipakchi and F. Albuyeh, ”Grid of the future,” IEEE Power and Energy
Magazine, vol. 7, no. 2, pp. 52-62, 2009.
[9] P. S. Moses, M. A. S. Masoum, and S. Hajforoosh, ”Overloading of
distribution transformers in smart grid due to uncoordinated charging
of plug-In electric vehicles” in 2012 IEEE PES Innovative Smart Grid
Technologies, 2012.
[10] M. Yilmaz and P. T. Krein, ”Review of Battery Charger Topologies,
Charging Power Levels, and Infrastructure for Plug-In Electric and
Hybrid Vehicles,” in IEEE Transactions on Power Electronics, vol. 28,
no. 5, pp. 2151-2169, 2013.
[11] R. D. Zimmerman, C. E. Murillo-Sanchez and R. J. Thomas, ”MAT-
POWER: Steady-State Operations, Planning, and Analysis Tools for
Power Systems Research and Education,” in IEEE Transactions on
Power Systems, vol. 26, no. 1, pp. 12-19, Feb. 2011.

a. Daily average of all node voltages


b. Time at which worst voltage occurs (to the nearest 5 minutes)

from simulations of a 1200 node test system in which PEVs


are randomly located throughout low-voltage networks and are
plugged in randomly within three time intervals. The main
conclusions are:
• This study indicates that even under low penetrations,
transformers experience load surging.
• There is a significant increase in distribution transformer
power load between PEV penetrations 15% and 30%. It
is conceivable that penetrations of this magnitude could
occur in residential networks in the near future.
• Under the lowest considered PEV penetration and the
most spread out charging case, there are occurrences of
voltage deviations greater than what is allowed by the
strict regulations set for low-voltage systems.
• If a coordination strategy is not implemented soon, util-
ities may need to replace distribution transformers more
frequently and upgrade the capacities of the transformers.
• Future studies could include analysis of additional effects,
such as phase imbalance and power quality harmonics.
R EFERENCES
[1] ”International Energy Agency”, Global EV Outlook, April 2017.
[2] C. Li, X. Yu, W. Yu, G. Chen, ”Efficient Computation for Sparse Load
Shifting in Demand Side Management”, IEEE Tans. smart Grids, vol.
8, no. 1, pp. 250-261, 2017.
[3] L. Yao, W. H. Lim, T. S. Tsai, ”A Real-Time Charging Scheme for
Demand Response in Electric Vehicle parking Station”, IEEE Trans.
Smart Grid, vol. 8, no. 1, pp. 52-62, 2017.

Das könnte Ihnen auch gefallen