Sei sulla pagina 1di 6

1

Power Quality Issues into a Danish Low-Voltage Grid with Electric Vehicles
Francesco Marra, Student Member, IEEE, Morten Mller Jensen, Rodrigo Garcia-Valle, Member, IEEE, Chresten Trholt and Esben Larsen

Abstract An increased interest on electric vehicles (EVs) and plug-in hybrid electric vehicles (PHEVs) is dealing with their introduction into low voltage (LV) distribution grids. Lately, analysis on power quality issues has received attention when considering EVs as additional load. The charging of EVs is expected without a centralized coordination in the first years, therefore a study on voltage variations and loading of LV grids is needed to estimate the immediate impact by EVs. In this work, the modeling and simulation of a Danish residential LV grid is conducted considering the two scenarios of dumb charging and overnight charging. The effect of different shares of EVs on voltage variations and transformer loading is assessed and analyzed. Results have shown that if single-phase, 16 A charging is employed, the actual LV grid would withstand the contingency of up to 30 - 40% of EVs charging overnight.

Index Terms electric vehicles, low voltage grid, power quality, uncoordinated charging, voltage variations

The available statistical data of driving patterns for ordinary cars are being analyzed in order to forecast the charging period of both PHEVs and EVs [5]. At the same time, considering the Danish targets of 50% penetration of wind power [6], it is being assessed the potential of EVsbased distributed storage to compensate for the intermittency of such renewable resource. This may give an important opportunity to future EVs in playing the role of ancillary services, contributing in the overall grid stability [7]. It is evident that the above scenarios relate to the concept of coordinated charging of EVs, based on smart metering and centralized systems, which makes possible a remote controlled charging of the vehicles [4]. By the way, at present, coordinated charging is not available yet, so the vehicle user has a full-control of the charging process, leading to a situation known as uncoordinated charging [3].

I. INTRODUCTION ow voltage grids in towns and villages are historically characterized by a dominant presence of residential load demand [1]. With the breakthrough on the market of EVs and PHEVs the consumption of residential grids will consequently increase adding new concerns for grid management. The integration of EVs into the electric power system is expected to take place into the low-voltage (LV) grid at first, though several concepts of public infrastructures are constantly in progress, such as public fast charging or battery swapping stations [2]. With EVs in residential grids, the forecasting of LV feeders loading will become more complex, due to increased unpredictability of the EV drivers. The basic household load profile, Fig. 1, will be influenced by the additional load procured by the EV charger. Therefore, the charging power will be a concern for the grid, in terms of voltage drops, voltage unbalance, harmonics distortion and transformer overloading [3]. Several research projects are addressing the feasibility of a centralized communication infrastructure which, by the use smart metering, would be able to manage the charging of EVs, taking into account the aforementioned grid constraints [4].
Mr. Marra, Mr. Garcia-Valle, Mr. Trholt and Mr. Larsen are with the Centre for Electric Technology, Department of Electrical Engineering, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark (e-mail: fm@elektro.dtu.dk). Mr. Morten M. Jensen is with the Danish Energy Association, Rosenrns all 9, 1970 Frederiksberg, Denmark.

Fig. 1. Household load profile [8]

In this work, a study on power quality issues related to a typical Danish LV grid with EVs is conducted, under the case of uncoordinated charging. With uncoordinated charging it is here intended the charging of EVs without a high level centralized control. Furthermore, the study considers that EV users can have the option of time-delayed charging. The effect on the household load profiles, transformer loading and voltage variations are investigated through a simulation model implemented in Matlab/Simulink. In Section II the basic LV grid operational requirements are investigated considering the standard Danish household load profiles, section III introduces the EV loading in the grid due to battery charging profiles. Simulation models and results are reported in section IV and V respectively. A discussion follows in Section VI, which addresses technical issues for the effective integration of EVs, based on the findings of the paper.

978-1-4673-0378-1/11/$26.00 2011 IEEE

2 II. LOW-VOLTAGE GRIDS In a typical distribution grid, the purpose of the LV grid is to provide connection from the Medium-Voltage (MV) grid to the supply points for the individual customers. Typically, a Danish LV grid consists of a 10/0.4 kV stepdown transformer to step down the voltage from the MV level to a distribution voltage level of 400 V, line-to-line. A common Danish residential LV grid is radial type, wired by three phase cables with a return neutral conductor and protection devices, such as AC breakers, while voltage regulation is likely to be passive. European regulations on supply power quality, as Standard EN 50160, have been written with a statistically coherent emphasis, stating that during each period of one week, 95% of the 10-minute average main values of the supply voltage in LV networks shall be within the range of Un +/-10%, where Un is the nominal voltage [9]. The Danish recommendations DEFU on LV grids have recently been amended to the same requirements as stated in Std. EN 50160. A. Design Criteria for Low Voltage Grids When planning distribution networks, a maximum percentage of 4-5% voltage drops at MV level are taken into account [10]. This tolerance is chosen for a series of reasons: to allow for the overall uncertainty in the load demand, dead-band for voltage regulation in the substation and the step size of the transformers taps in the secondary substations. Furthermore it is required that the voltage level is ensured in case of coupling in the MV grid due to maintenance or failures. A maximum voltage drop of 5% is the criteria used when planning a new Danish LV grid [10]. For many years, when designing LV grids where the load is unknown, a standardized formulation is commonly used, known as Boggis method [11]. This formulation is based on the concept of After Diversity Maximum Demand (ADMD). The ADMD is the maximum demand, per customer, as the number of customers connected to the network approaches infinity. This is usually derived from the maximum yearly nodal demand on the transmission network divided by the number of customers served by that network. The ADMD is multiplied by a diversity factor (DF) that increases the demand per customer as the size of the group decreases. The formula was implemented in many forms but the maximum demand (MD) generally is given by (1), the ADMD is defined by (2) and the diversity factor (DF) is given by (3).
MD N DF ADMD

maximum demand of the aggregated customers to the coincident maximum of the aggregated customers ADMD Pn N k After Diversity Maximum Demand per customer demand of customer n at the time of system maximum demand number of customers empirically determined constant

The method is generally used when the number of customers N is greater then 25. Anyway a series of limitations applies to the mentioned method: it is not possible to estimate voltage quality at any point in time; the maximum demand of a group of customers can be calculated within a long period of time, usually one year; the demand is calculated as aggregated sum, therefore it is not possible to consider the effects of phase loading unbalance and neutral impedance. While the second point is not really a limit, the first and third are reasons for searching alternative methods. Especially with EVs as additional loads, it is important to ensure the required voltage level at any point in time in the grid. Furthermore, single-phase loading due to EV chargers have an impact in the grid, which cannot be addressed by the Boggis method. To overcome the limitations of the method a Danish LV grid model was implemented and simulations were conducted with different share of EVs. B. Assumption for LV grid study As demand will increase without grid reinforcement, the risk for voltage drops should also be allocated in the MV grid. In this work, a level of 2% voltage drop is allocated to the MV grid. Accordingly, a voltage drop limit of 5% is considered for the LV grid study based on [10]. Assumptions:
'VMV 'VLV 2% 5%

(1) (2) (3)

ADMD

1 N Pn N of N n 1 lim

Another assumption is made on the household load profiles to use in the simulation. Based on the available data, a typical Danish LV grid load profile is considered, as depicted in Fig. 2. The consumption is measured for a LV grid supplying households without electrical heating. The real measurements are taken during four different periods of the year. In particular: - a Wednesday in January (dark blue) - a Sunday in January (light blue) - a Wednesday in July (violet) - a Sunday in July (red) Based on the real measurements, the electrical consumption is considerably higher during winter time, with peaks between 17:00 and 21:00 in the evening. In contrast, there is a pretty flat consumption during summer time.

DF
Where MD DF

k 1 N

is the maximum demand of the aggregated customers is the diversity factor, ratio of non-coincident

3 A meaningful part of the load profile belongs to overnight hours. In this period indeed, it is clearly observed a very low consumption at any season, with duration of about 5 - 6 hours which could fit for charging an EV battery [12].
Consumption on a LV feeder supplying detached houses without electric heating

3rd Wednesday of January 0,35

3rd Wednesday of July

3rd Sunday of January

3rd Sunday of July

0,3

0,25

0,2

0,15

0,1

0,05

among these use cases, while the limitation for the power level will mainly derive from grid constraints rather than the battery [15]. It is expected that the single-phase, 16 A connection, will be largely employed as home-charging case. By the way, the current fuses, at the household connection, further limit the charging power, so EV manufacturers are setting the chargers to lower power rates, i.e. single-phase, 10 A, which entails a charging power of about 2.3 kW. The charging power profile of a Li-ion battery [16] is used in this work for the modeling. The charging algorithm is characterized by two operational regions: constant current (CC), until the voltage upper limit is reached; constant voltage (CV), until the current reaches the minimum according to the manufacturer. In Fig. 3, the charging profiles of the Li-ion battery are depicted.

Consumption [kWh per 15 mins]

Fig. 2. Typical Danish LV feeder load profiles for households without electric heating

00 :0 0 00 :4 5 01 :3 0 02 :1 5 03 :0 0 03 :4 5 04 :3 0 05 :1 5 06 :0 0 06 :4 5 07 :3 0 08 :1 5 09 :0 0 09 :4 5 10 :3 0 11 :1 5 12 :0 0 12 :4 5 13 :3 0 14 :1 5 15 :0 0 15 :4 5 16 :3 0 17 :1 5 18 :0 0 18 :4 5 19 :3 0 20 :1 5 21 :0 0 21 :4 5 22 :3 0 23 :1 5

SOC

The winter Wednesday load profile is used in the LV grid simulation model. The choice is made considering a weekday in winter as a more conservative scenario. III. ELECTRIC VEHICLES LOADING Recent publications by market consulting companies foresee more than 70% of EVs in 2015 to be powered by lithium-ion (Li-ion) batteries [13]. Compared to other battery technologies, Li-ion batteries offer a greater energyto-weight ratio, no memory effect and low-self discharge when not in use [14]. These are some of the main aspects for the establishment of Li-ion batteries as leading solution for automotive applications. Based on these reasons, EVs in this work are considered Li-ion technology based and therefore EVs loading is performed by Li-ion battery chargers. From a grid perspective, with the increasing share of EVs, a largely residential feeder circuit must be ready to have a significant proportion of its customers adding a fixed load of few kW due to EV charging. Without a centralized coordination of the EV charging, the charger would operate continuously for several hours starting anytime during the day, as well as it would entail night time hours, if a timer device is employed by the user. The charging power rates are commonly considered as the ones shown in Table I.
TABLE I CHARGING POWER RATES Fig. 3. Typical charging profiles for an EV Li-ion battery [16]

The charging power profile seen from the grid gets a similar shape likewise the red curve in Fig. 3, which represents the DC current. According to Li-ion battery manufacturers, a 0.5CA current rate is recommended for charging the battery, where 1C is the nominal capacity of the battery, while A stands for Ampere. A practical example is given considering an EV with a 28 kWh battery pack, 350V, 80 Ah cell. Based on Fig. 3, a charging current of 40 A can be used by the charger on the DC side, which corresponds to the 0.5CA rate for the battery. The charging power used by the charger is calculated as follows:
I ch arg ing _ DC Vch arg ing _ DC Pch arg ing 0.5CA 0.5 80 A 438V 17.5 kW 40 A

(4) (5) (6)

1.25 Vpack

I ch arg ing _ DC Vch arg ing _ DC

AC current
10 A 16 A 32 A 16 A 32 A

AC voltage
230 V 230 V 230 V 400 V 400 V

Grid connection
single phase single phase single phase three-phase three-phase

Power
2.3 kW 3.7 kW 7.4 kW 11 kW 22 kW

The five listed cases are characterized by the different current rates, usually 16A or 32A, as well as based on the grid connection type, single-phase or three-phase. It is expected that all on-board EV chargers will employ one

The formula given by (4) is the charging current on the battery side. Equation (5) is the voltage applied by the charger to the battery. An empirical factor of 1.25 is generally considered in the design of the charging voltage of Li-ion battery chargers. Pcharging in (6) is the charging power used by the charger, seen from the grid. Neglecting the charger efficiency and considering single-phase charging, the current on the AC side would result in the range of 76 A. Such power rate cannot be employed for EV charging in a home-charging case, since it would represent a serious impact on the overall LV grid. In this work, the charging power level of 3.7 kW, Table I, which corresponds to the home-charging case 230 V 16 A, is considered.

4 IV. MODELING OF LV GRID WITH EVS To study the power quality issue in the existing LV grid with EVs, two ways were addressed. First, an estimate on the available capacity of LV grids was conducted based on the average loading data of 10/0.4 kV Danish distribution transformers. In a second step, a typical residential LV grid was modelled with the software Matlab/Simulink. In the simulations, the effect on voltage drops was examined and discussed. A. Analysis of LV grids capacity Analysis on LV grid loading was performed, in order to gain an estimate on the usable capacity which can be allocated for EV loading. At the scope, the annual average loading data of 10/0.4 kV Danish distribution transformers was analyzed. The analysis results are depicted in the diagram of Fig. 4. Five classes of transformers are mostly used; their nominal power that ranges between 50 kVA to 630 kVA. Every class of transformers is loaded in average up to half its nominal capacity, during one year. This means that the utilization factor for each transformer is around 0.5; this gives optimal reasons for considering the integration of EVs into LV grids at several charging power levels, according to Table I. The number of EVs which can be allocated in a LV grid can be roughly calculated based on the average annual loading data of all 10/0.4 kV transformers, but this turns out to be a more conservative scenario, if considering overnight charging of the vehicles.
0.6 0.55 0.5 0.45
Utilization factor

R,2
25mm

25mm

S,3

23m 34m 44m


25mm

A,2
25mm

23m
25mm

B,3

I,3
25mm

L,2

68m 13m
25mm

64m 19m
25mm

C,3
25mm

D,2

M,3
25mm

N,2

56m 66m
95mm

39m

E,2
25mm

26m

46m

O,0
25mm

F,2

95mm

95mm

38m

40m

G,3
25mm

77m

100 kVA 10/0.4 kV


25mm

P,5
65m

H,3

Q,4

Fig. 5.

LV residential grid used for the study

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 50 100 200
Rated power (kVA)

Fig. 6. Load profiles of household with EV - overnight charging

400

630

Fig. 4. Average loading of secondary distribution transformers in Denmark

It is to be noted also that bottlenecks of capacity which may occur in the MV level are outside the scope of this work. It is therefore assumed that the power required by LV grids is available at MV level. B. Modeling of a residential Danish LV grid The Danish LV grid considered for the study is depicted in Fig. 5. A 100 kVA transformer serves a predominantly residential area with 44 households and 17 distribution cabinets from which the cables are derived. The letter and the number next to each cabinet indicate the cabinet ID and the number of served houses respectively. The cables lengths and cross-sections are also considered for the modeling and simulation work.

C. Modeling of household consumption The household load model was performed considering an approximation of the winter Wednesday load profile of Fig. 2. The measurements in Fig. 2 enlighten an average overnight consumption of about 280 W per household. A morning peak is visible at 7.00, up to 600 W, while between 17:00-18:00 the evening peak is recorded, up to 1200 W. With respect to the measurements, the household load profile was modelled considering two components: The basic household load profile, without EV The EV load profile

The charging takes place according to the following two scenarios: Dumb charging starting around 17.00 Overnight charging starting around midnight, with time-delay set by the EV user

The charging power profile is set at 3.3 kW for both simulation scenarios, and a PF of 0.9 is considered for the charger. For all cars the charging process entails a time interval of about 6 hours according to [12]. The charging starts allowing a random uncertainty time window of +/- 15 minutes for all EVs. The State-of-Charge (SOC) of the vehicles is neglected in this work, since the focus is on the

5 contingency of simultaneous charging of the EVs. In Fig. 6, it is shown the case of a household load profile with an EV charging during night. In this particular case, the EV starts charging at midnight and stops charging at 6.00 in the morning. The total household consumption is derived summing up the two loading curves. Similarly, the dumb charging simulation scenario is generated with a household with EV, who starts charging around 17.00 and stop charging around 23.00. The different EVs penetration levels in the grid are simulated with the random introduction of an EV load per each household. D. Simulation parameters The LV grid model developed used the following parameters: Transformer capacity: 100 kVA Voltage Level: 0.4 kV Household PF: 0.98 MV grid voltage drop: 2% Safety factor cable load: 1.2 Safety factor security flow value: 1.2 Maximum LV grid voltage drop: 10% 10 kV mains impedance ratio X/R: 0.7 two 0.4 kV LV feeders Number of households: 44 overnight load per household without EV: 280W EV charger power: 3.3 kW; PF: 0.9 EV state-of-charge: non considered feeders were measured for both simulation scenarios of dumb charging and overnight charging. Simulation results of the two scenarios are collected in Table II and Table III respectively. In the dumb charging scenario, Table II, the EV charging takes place around 17.00 and this corresponds with the overall LV grid peak. A 20% share of EVs is possible into the LV grid without overloading the transformer. In the overnight charging scenario, Table III, the EV charging takes place around midnight. In this case, almost 40% of EV share is possible without overloading the transformer.
TABLE II SUPPLY VOLTAGE VARIATIONS WITH DIFFERENT SHARE OF EVS DUMB CHARGING SCENARIO Penetration Level [%] 100 80 60 50 40 30 20 10 5 Transf. Loading [%] 208 173 152 139 125 104 90 70 69 Voltage at feeder top [p.u.] 0.925 0.94 0.95 0.955 0.96 0.97 0.975 0.98 0.982 Voltage at feeder terminal [p.u.] 0.887 0.9 0.925 0.935 0.94 0.95 0.957 0.97 0.97

In the simulation it was considered that all 44 households share the same base load profile and they can potentially own one EV. The EVs are randomly connected at household points of connection in the LV grid. Different penetration levels of EVs are simulated into the grid ensuring an equal distribution among the phases. The transformer loading was derived by the ratio between the mean value of the transformer phase-current and the mean value of the nominal transformer phase-current. To calculate the nominal phase-current the following standard formula was considered:
Sn 3 V1n I1n 3 V2 n I 2 n

TABLE III SUPPLY VOLTAGE VARIATIONS WITH DIFFERENT SHARE OF EVS OVERNIGHT CHARGING SCENARIO Penetration Level [%] 100 80 60 50 40 30 20 10 5 Transf. Loading [%] 177 142 120 108 98 77 65 44 33 Voltage at feeder top [p.u.] 0.945 0.955 0.97 0.972 0.978 0.982 0.989 0.99 0.992 Voltage at feeder terminal [p.u.] 0.91 0.922 0.939 0.943 0.955 0.962 0.975 0.98 0.985

(7)

Where An V1n V2n I1n I2n is the transformer nominal power is the transformer line-to-line voltage on the primary circuit is the transformer line-to-line voltage on the secondary circuit is the phase current on the primary circuit is the phase current on the secondary circuit V. SIMULATION RESULTS The simulation work investigated a number of power quality issues involved in the LV grid, with different charging scenarios and EVs penetration levels. The transformer loading and the voltage magnitude across the

On the other hand, voltage drops meet in both scenarios the voltage requirement of Std. 50160. With the 20% EV penetration level of the dumb charging scenario, a 4% voltage drop is measured at the feeders terminals. In the overnight scenario, with a 40% penetration level, a 4.5% voltage drop is measured at the feeders terminals. VI. DISCUSSION A typical LV distribution grid, serving a mostly residential area should comply with the power quality requirements of Std. 50160. Adding to such an area tens or hundreds of EVs charging without a centralized coordination may result in a system overload. Based on the findings of this paper, a LV grid should be locally analyzed to allow a correct assessment on the

6 transformer loading due to EVs as additional load. On the other hand, the cables length and their cross sections should be modeled in order to estimate the effects on cable overloading and the entity of voltage variations. On the vehicle side, the three-phase charging power levels of 11 kW and 22 kW, which are not addressed, would have different impacts in the LV grids. The charging power used in the work is still limited to few kW due to grid constraints, rather than to the battery. As shown in Section III, a common 20 kWh EV battery can be charged with a power level of about 10 kW, complying with the manufacturer recommendations. It seems evident that, the market acceptance of EVs by future customers and their favorite charging time will play a key role in the strategies to adopt in future LV grids reinforcement. VII. CONCLUSIONS In this paper, analysis of grid issues such as voltage variations and transformer loading of a typical Danish LV grid incorporating EVs was performed. The requirements on supply voltage variations were addressed with reference to Std. EN 50160. The work considered average loading data of secondary Danish distribution transformers, ranging from 50 kVA to 630 kVA as a pre-study on LV grids. The outcome of the assessment enlightened an average annual utilization factor of about 0.5 for all classes of transformers. This aspect offers optimal chances for the introduction of EVs into LV grids. A simulation model of a Danish LV grid was then built in order to investigate the voltage variations and grid load profile, based on the two scenarios of dumb charging and overnight charging. Different levels of EVs penetration were simulated for the two scenarios, while the EV load was modelled with a fixed power rate of 3.3 kW. From the simulations it came out that, if single-phase charging is employed, rated at 3.3 kW per vehicle, LV grids would allow up to 40% of EVs penetration charging during night and meeting the voltage requirements of Std. EN 50160. At the same time, a 20% share of EVs would be possible in the dumb charging scenario, without the need of grid reinforcement. VIII. REFERENCES
[1] D. H. O. McQueen, P. R. Hyland, and S. J. Watson, Monte Carlo Simulation of Residential Electricity Demand for Forecasting Maximum Demand on Distribution Networks, IEEE Transactions on Power Systems, 2004. L. Pieltain Fernndez, T. Gmez San Romn,R. Cossent, C. Mateo Domingo and P. Fras, Assessment of the Impact of Plug-in Electric Vehicles on Distribution Networks, IEEE Transactions on Power System, 2010. K. Clement-Nyns, E. Haesen and J. Driesen, The Impact of Charging Plug-In Hybrid Electric Vehicles on a Residential Distribution Grid, IEEE Transactions on Power Systems, Vol. 25, No. 1, February 2010. C. Binding, D. Gantenbein, B. Jansen, O. Sundstrom, P. B. Andersen, F. Marra, B. Poulsen and C. Trholt, Electric Vehicle Fleet Integration in the Danish EDISON Project - A Virtual Power Plant on the Island of Bornholm, IEEE Power & Energy Society General Meeting, 2009. B. Adornato, R. Patil, Z. Filipi, Z. Baraket, T. Gordon, Characterizing naturalistic driving patterns for Plug-in Hybrid Electric Vehicle analysis, IEEE Vehicle Power and Propulsion Conference, 2009. Z. Xu, M. Gordon, M. Lind, J. Oestergard, Towards a Danish Power System with 50% wind Smart grids activities in Denmark, IEEE Power & Energy Society General Meeting, 2009. W. Kempton and J. Tomic, Vehicle-to-grid power fundamentals: Calculating capacity and net revenue, J. Power Sources, vol. 144, no. 1, pp. 268279, Jun. 2005. W.L.O. Fritz, and D.C. Callis, Domestic load-profile measurements and analysis across a disparate consumer base, Cape Peninsula University of Technology, Cape Town, South Africa, public article. Standard EN 50160, Voltage Characteristics of electricity supplied by public distribution networks, Cenelec 2010. SEAS-NVE, Design manual for 10 kV and 0.4 kV, Denmark J. Boggis, Diversity, bias and balance, Distrib. Elect., pp. 357362, 1953. Nissan Leaf Electric Car, Charging info, www.nissanusa.com. Frost & Sullivan, public report M5B6-Global Electric Vehicles Lithium-ion Battery Second Life and Recycling Market Analysis, online, 2010. N. A. Chaturvedi, R. Klein, J. Christensen, J. Ahmed and A. Kojic, Algorithms for Advanced Battery Management Systems, IEEE Control Systems Magazine, 2010. F. Marra, C. Trholt, E. Larsen and Q. Wu, Average Behavior of Battery-Electric Vehicles for Distributed Energy Studies, IEEE Innovative Smart Grid Technology Conference, 2010. Thunder Sky, LFP Battery User Manual.

[6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

[2]

[3] [4]

[5]

Potrebbero piacerti anche