Fourth International Conference on Industrial and Information Systems, ICIIS 2009, 28 - 31 December 2009, Sri Lanka

An Autonomous Distributed Vehicle-to-Grid

Control of Grid-connected Electric Vehicle
Yutaka Ota, Haruhito Taniguchi, Tatsuhito Nakajima, Kithsiri M.Liyanage, and Akihiko Yokoyama
The University of Tokyo, yota@upn.t.u-tokyo.ac.jp

Abstract- Penetrating large amount of renewable energy Because of advanced technology about Li-ion battery and
resources into power system, battery energy storage system hybrid electric vehicle, large amount of next-generation
perform important role for smoothing their natural variability,
ensuring grid-wide frequency stability, and suppressing voltage vehicles including PHEV and EV are indicated in Japan
rise caused by reverse power flow. The ubiquitous power grid (Table I). Therefore, there is a large potential of V2G.
concept has been proposed as a Japanese smart grid, in where In this paper, a V2G control method for grid-connected
total battery capacity could be optimized by coordinating PHEV and EV is proposed. V2G power when EV is plugged-
controllable distributed generators, loads on demand side, for into 200/100V home outlet has simple droop characteristics
example, heat pump system with heat storage, and plug-in
hybrid vehicle or battery electric vehicle with onboard battery. against the power system frequency at plug-in terminal.
In this paper, we propose a vehicle-to-grid control of grid- Therefore, PHEV and EV behave as a governor free. The
connected plug-in hybrid electric vehicle and electric vehicle as a effect for load frequency control of power system is able to
kind of the demand response in the ubiquitous power grid. be estimated through simple information exchange between
Proposed control is based on simple droop characteristics vehicles and control center of power system. Proposed V2G
against the power system frequency at plug-in terminal, and
considers the risks about use for vehicle and battery condition. control also considers the battery condition based on a battery
Autonomous distributed Smart Storage for governor free state-of-charge (SOC) estimation [6] and a charge scheduling
control, spinning reserve, local area voltage control, and other for next drive.
smart grid applications is realized by packaging the proposed An autonomous distributed “Smart Storage” [7] with fast
control to automotive power electronics circuit and electric response and smart cooperation to the power system is
control unit.
constructed by packaging the proposed V2G control to
I. INTRODUCTION automotive power electronics circuit and electric control unit
(ECU). Smart storage concept on ubiquitous power grid is
In Japan, photovoltaic and wind power generation will be described, and the proposed V2G control is validated by
introduced over 30% of the peak demand (179[TW] in 2007 applying actual frequency measurements in this paper.
summer) until 2030 shown in Table I [1]. Large-scale battery
energy storage system is necessary for ensuring grid-wide
II. SMART STORAGE ON UBIQUITOUS POWER GRID
frequency stability. However, total battery capacity could be
optimized by demand response or demand side management Fig. 1 shows a smart storage concept on ubiquitous power
of the controllable distributed generators and loads. This is grid. Power electronics circuit for motor drive is diverted to
one of the smart grid features, and authors has already the grid-connection without any additional circuits [8],[9].
investigated about an ubiquitous power grid concept And interactive V2G control is realized when PHEV and EV
coordinating the blade pitch angle control of wind power are plugged into 200/100V home outlet.
generation [2] or heat pump system [3] together with battery Supply-demand imbalance of the power system is observed
energy storage system. Total coordinating control scheme of from frequency deviation at plug-in terminal [10],[11],[12].
distributed generator, controllable load, and battery energy V2G power is set as droop characteristics against the
storage is also proposed [4]. frequency deviation shown in Fig. 2. V2G gain (Kmax) should
As a kind of demand response, vehicle-to-grid (V2G) of be adjusted considering a tradeoff between the effect of V2G
grid-connected plug-in hybrid electric vehicle (PHEV) and and the fluctuation range of SOC according to additional
electric vehicle (EV) is investigated and demonstrated [5]. charge-discharge cycles for V2G. Maximum V2G power

TABLE I
ACTUAL ACHIEVEMENT AND FUTURE MAXIMUM INTRODUCING CASE OF
RENEWABLE ENERGIES AND NEXT-GENERATION VEHICLES IN JAPAN.

actual achievement maximum introducing maximum introducing

until 2005 case until 2020 case until 2030
Photovoltaic power 1.42 GW 14.32 GW 53.21 GW
Wind power 1.08 GW 4.91 GW 6.61 GW
Next-generation vehicle 0% of total vehicles 20% of total vehicles 40% of total vehicles

978-1-4244-4837-1/09/$25.00 ©2009 IEEE 414

 Hydro Ubiquitous Power Grid
Nuclear
Thermal Control Center

Small
Power Grid
Large-scale
Power Grid
Control Center

Wind Park Control Center Battery Energy

Storage System

Frequency Wind
ECU / Smart Meter Small Power Grid
Measurement V2G Photovoltaic
SOC Estimation

DC-DC Inverter
Battery Converter

Motor Distributed
Load Gas-engine
Heat Pump Load Generator
Plug-in Hybrid Electric Vehicle Gas-turbine

Electric Vehicle
Heat Storage Reversible H 2 Local Network
Smart Storage Fuel Cell

Fig. 1. Smart storage of PHEV and EV on ubiquitous power grid.

⎧ ⎛ SOC − SOC low(high) ⎞

n
⎫
⎪ ⎪ (1)
(Pmax) is decided by specifications of 200/100V home outlet. K V2G = K max ⎨1 − ⎜ ⎟ ⎬
⎜ SOC ⎟
⎪⎩ ⎝ max(min) − SOC low(high) ⎠ ⎪⎭
As battery SOC is near to full (empty), high-power charge
(discharge) is difficult for preventing over-charge (over- SOCmin, SOClow, SOChigh, SOCmax, and n are parameters, and
discharge). During long-term V2G charge-discharge cycles, it Fig. 3 show sample case of V2G suppression (SOCmin=10[%],
is concerned that battery SOC gradually shift from its SOClow=20[%], SOChigh=80[%], SOCmax=90[%] , n=2). When
nominal value. Considering these battery features, V2G gain battery SOC is away from 50[%], V2G power is suppressed
is suppressed as following equation on the premise that to return SOC 50[%]. It is obvious that V2G suppression
accurate SOC estimation is realized. works as a kind of SOC balancing.

P max V1G Charge

K max
V2G/V1G
Charge

V2G/V1G Gain [kW/0.1Hz]

V2G/V1G Power [kW]

V2G Discharge
V1G K max
0
Δf max
Discharge

V2G Charge
V2G
SOClow SOChigh

0 10 20 30 40 50 60 70 80 90 100
Frequency Deviation [Hz]
SOCmin Battery SOC [%] SOCmax
Fig. 2. Droop characteristics against frequency deviation. Fig. 3. Suppression of V2G/V1G considering battery condition.

415
 Battery SOC must to be full or desired capacity until the Switching V2G to V1G mode at around 71[h] based on
next drive. If user set time and mileage of next drive in estimation of necessary time of V1G charge, desired capacity
advance, one-way charge, which is called as V1G, with droop (91.6[%]) is accomplished until the next drive (78[h]).
characteristics against frequency deviation is executed as a There is saturation of V2G power when maximum absolute
kind of smart charging [13]. Because this V1G mode doesn’t value of frequency deviation is more than 0.05[Hz]. This
intend to balance SOC, V1G power is not suppressed saturation is evaluated by changing maximum V2G power
according to battery SOC, but CV (Constant Voltage) charge from 0.5 to 10 [kW]. Frequency of saturation is found to be
is executed when SOC is near to full. relatively small when maximum V2G power is 5[kW] (Fig.
Assuming that power system frequency is dominated by 7). And root mean square of V2G power is almost the same
normal distribution (standard deviation (σ)), the expectation as that when maximum V2G power is 10[kW] (Fig. 8).
value of V1G power (PV1G) is estimated as follows. Fig. 6 shows the results about Western 60Hz systems.
σ (2) Although battery SOC reach full or empty without V2G
PV1G = K max
2π suppression, SOC is kept around 50% with V2G suppression.
On the other hands, necessary energy from present SOC to
destination SOC (WV1G) is estimated by using battery model
used for SOC estimation. Finally, the necessary time of V1G TABLE II
charge (TV1G) is calculated as follows. SPECIFICATIONS OF BATTERY CELL (PACK).

W 2π W V1G (3) Nominal voltage (Vnom) [V] 3.7 (325.6)

T V1G = V1G =
PV1G σ K max Nominal capacity (Cnom) [Ah] 50 (50)
When the necessary time is smaller than the period to the start Adjusting parameter (α) 15
time for next drive, V2G mode is transferred to V1G mode. Charge-discharge efficiency (η) 1.0
Information exchange between power system control Internal resistance (Rint) [Ohm] 0.004 (0.352)
center and each electric vehicle is not required for the Voltage of CV charge [V] 4.1 (360.8)
proposed autonomous control. However, the parameters for Voltage of discharge cutoff [V] 2.7(237.6)
power system frequency control (amount of spinning reserve
and system constant) should be determined considering the
effect of Smart Storage. Therefore, Smart Storages are 4.3 CCV
aggregated through periodical information collection. 4.2 CV(4.1V) Charge
4.1
4 CCV
III. VERIFICATION THROUGH FREQUENCY MEASUREMENTS 3.9 CC(50A) Charge
3.8 OCV
Battery voltage [V]

Proposed V2G control is applied to the actual frequency 3.7

measurements by Campus PMU/WAMS [14] on two major 3.6
3.5 CCV
power systems in Japan, Eastern 50Hz systems and Western 3.4 CC(50A) Discharge
60Hz systems. Data sampling is 10[s], and the period of 3.3
analysis is 84[h]. 3.2
3.1
Simplified battery model is assumed as PHEV or EV 3
lithium ion battery (Appendix A). Specifications of battery 2.9
cell and pack, which is consisted by 88 cells, are summarized 2.8
2.7
in Table II. And Fig. 4 shows OCV and CCV as CC 2.6
(Constant Current (50A)) charge-discharge. CV charge is 0 10 20 30 40 50 60 70 80 90 100
executed when cell CCV is higher than 4.1[V]. On the other Battery SOC [%] 91.6%
hand, discharge is cutoff when cell CCV is lower than 2.7[V]. Fig. 4. OCV and CCV of assumed lithium ion battery.
The parameters of proposed V2G control are summarized
in Table III. Maximum V2G power (Pmax) is 5[kW] based on
the assumption of 200V/25A home outlet. V2G gain (Kmax) is TABLE III
10[kW/0.1Hz]. The parameters of V2G suppression are the PARAMETERS OF V2G CONTROL.
same conditions in Fig. 3. Initial SOC is set as 20%, and Maximum V2G power (Pmax) [kW] 5
desired capacity is 91.6[%] at 78[h] for next drive. Necessary V2G gain (Kmax) [kW/0.1Hz] 10
time of V1G charge (TV1G) is calculated from the minimum Parameters of V2G suppression :
value of standard deviation (σ) during past twelve hours. n, SOCmin, SOClow, SOChigh, SOC max [%] 2, 10, 20, 80, 90
Fig. 5 shows the results about Eastern 50Hz systems. Initial SOC [%] 20
Normal distribution of frequency is confirmed from standard Time for next drive [h] 78
deviation and moving average. V2G is realized at plug-in Destination SOC for next drive [%] 91.6
standby timing, keeping SOC balance with V2G suppression.

416
 0.12 0.12
2σ 2σ

Frequency deviation [Hz]

Frequency deviation [Hz]

0.08 0.08

0.04 0.04

0.00 0.00

-0.04 -0.04

-0.08 -0.08
-2σ Moving average -0.12
-2σ Moving average
-0.12
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

(a) Frequency deviation (a) Frequency deviation

6 6
5 5
4 4

V2G/V1G Power [kW]

V2G/V1G Power [kW]

3 3
2 2
1 1
0 0
-1 -1
-2 -2
-3 -3
-4 -4
-5 -5
-6 -6
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

(b) V2G/V1G power (b) V2G/V1G power

14 18
16
12

V1G charge time [hour]

V1G charge time [hour]

14
10
12
8 10
6 8
6
4
4
2 2
0 0
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

(c) Necessary time of V1G charge (c) Necessary time of V1G charge
100 100
90 90
80 80
With V2G suppression
Battery SOC [%]
Battery SOC [%]

70 With V2G suppression 70

60 60
50 50
40 40
30 30
20 Without V2G 20
Without V2G suppression
10 suppression 10
0 0
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

(d) Battery SOC (d) Battery SOC

370 370
360 360
Battery voltage [V]

Battery voltage [V]

350 350
340 340
330 330
320 320
310 310
300 300
0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

(e) Battery CCV Time [hour] (e) Battery CCV Time [hour]

Fig. 5. Results of proposed V2G control (Eastern 50Hz systems) Fig. 6. Results of proposed V2G control (Western 60Hz systems)
Hour rate of V2G saturation [%]

70 2.5
Western 60Hz systems
RMS of V2G power [kW]

60
2
50
Eastern 50Hz systems
40 1.5
30
1
20
Western 60Hz systems
10 0.5
Eastern 50Hz systems
0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Maximun V2G power [kW] Maximun V2G power [kW]
Fig. 7. Hour rate of V2G saturation. Fig. 8. RMS of V2G power.

417
 IV. CONCLUSION Storage and Frequency Regulation in the PJM System”, Publications of
MAGICC (Mid-Atlantic Grid Interface Cars Consortium),
In this paper, we proposed an autonomous distributed V2G http://www.magicconsortium.org/_Media/test-v2g-in-pjm-jan09.pdf
control of grid-connected PHEV and EV by coordinating (2009)
[6] Y.Ota, M.Sakamoto, R.Kiriake, T.Kobe, and Y.Hashimoto, “Modeling
battery condition, use for vehicle, and power system state. of Voltage Hysteresis and Relaxation of HEV NiMH Battery”, Proc.
V2G with SOC balancing and V2G/V1G switching is 17th IFAC World Congress (2008)
validated through frequency measurements on Japanese major [7] Y.Ota and Y.Hashimoto, “Smart Storage Concept of Grid-connected
Electric Vehicle for introducing Renewable Energy to Power System”,
power systems. Proc. the 24th International Battery, Hybrid and Fuel Cell Electric
Future work is packaging proposed V2G control, SOC Vehicle Symposium and Exhibition (EVS24) (2009)
estimation, and power system sensing to automotive power [8] W.Kramer, S.Chakrabolty, B.Kroposki, and H.Thomas, “Advanced
Power Electronic Interfaces for Distributed Energy Systems Part 1
electronics circuit and ECU. Design of power system control Systems and Topologies”, Technical Report of National Renewable
scheme considering Smart Storage is on the next stage. Energy Laboratory, NREL/TP-581-42672 (2008)
[9] B.Kramer, S.Chakraborty, B.Kroposki, “A review of plug-in vehicles
and vehicle-to-grid capability”, Proc. IEEE IECON, pp.2278-2283
(2008)
ACKNOWLEDGMENT [10] Z.Zhong, C.Xu, B.J.Billian, L.Zhan, S-J.Steven Tsai, R.W.Conners,
V.A. Centen, A.G.Phadke, and Y.Liu, “Power System Frequency
This research was supported by Specially Promoted Monitoring Network (FNET) Implementation”, IEEE Transactions on
Research Grant from Power Academy of Japan and Power Systems, Vol.20, No.4, pp.1914-1921 (2005)
Ubiquitous Power Grid Endowed Chair of the University of [11] O.Samuelsson, M.Hemmingsson, A.H.Nielsen, K.O.H.Pedersen, and
J.Rasmussen, “Monitoring of Power System Events at Transmission
Tokyo. The authors gratefully acknowledge these supports. and Distribution Level”, IEEE Transactions on Power Systems, Vol.21,
No.2, pp.1007-1008 (2006)
[12] X.Ding, T.Littler, J.Morrow, P.A. Crossley, V.Terzija, M.O’Malley,
“Synchronized Phasor Measurement on the All-Ireland Electrical
APPENDIX A SIMPLIFIED BATTERY MODEL Network”, Proceeding of IEEE PowerTech (2007)
[13] A.Brooks, S.Thesen, “PG&E and Tesla Motors: Vehicle to Grid
Firstly, battery SOC is updated by integrating charge- Demonstration and Evaluation Program”, Proc. the 23rd International
Battery, Hybrid and Fuel Cell Electric Vehicle Symposium and
discharge current (I) from initial SOC. Exhibition (EVS-23) ( 2007)
d SOC [14] Y.Ota, T.Hashiguchi, H.Ukai, M.Sonoda, Y.Miwa, and A.Takeuchi,
= ηI (A-1) “Monitoring of Interconnected Power System Parameters using PMU
dt based WAMS”, Proc. IEEE PowerTech, pp.1718-1722 (2007)
η is charge-discharge efficiency. Then battery OCV (Open
Circuit Voltage) is defined as following Nernst equation.
RT ⎛ SOC ⎞ (A-2)
OCV = V nom + α ln ⎜⎜ ⎟⎟
F ⎝ C nom − SOC ⎠
Vnom and Cnom are nominal voltage and capacity, respectively.
R, F, T is gas constant, faraday constant, and battery
temperature, respectively. α is a adjusting parameter about
voltage change. Battery CCV (Closed Circuit Voltage) is also
calculated considering internal resistance (Rint) as follows.
CCV = OCV + R int I (A-3)

REFERENCES
[1] “Long-term energy supply-demand forecast”, Ministry of Economy,
Trade and Industry, http://www.enecho.meti.go.jp/topics/080523b.pdf
(2008). (in Japanese)
[2] Y.Nishizaki, H.Irie, A.Yokoyama, and Y.Tada, “Blade Pitch Angle
Control and its Capacity Reduction Effect on Battery for Load
Frequency Control in Power System with a Large Capacity of Wind
Power Generation”, IEEJ Transactions on Power and Energy, Vol.129,
No.1,pp.50-56 (2009). (in Japanese)
[3] T.Masuta, A.Yokoyama, K.M.Liyanage, and Y.Tada, “System
Frequency and Tie-line Power Flow Control by Coordination of
Batteries and Heat Pump Based Water Heaters on Customer Side in the
Interconnected Power System with a Large Penetration of Wind Power
Generation”, The Papers of Joint Technical Meeting on Power
Engineering and Power Systems Engineering. IEE of Japan, PE-09-163
PSE-09-171 (2009) (in Japanese)
[4] K.M.Liyanage, A.Yokoyama, Y.Ota, H.Taniguchi, and T.Nakajima, “A
Method to Support Load Frequency Control in Ubiquitous Power
Networks”, Proc. IEEE ICIIS (2009) (to be presented)
[5] W.Kempton, V.Udo, K.Huber, K.Komara, S. Letendre, S.Baker,
D.Brunner, and N.Pearre, “A Test of Vehicle-to-Grid (V2G) for Energy

418