Utilizing Plug-in Electric Vehicles for Peak Shaving and Valley

Filling in Non-residential Buildings with Solar Photovoltaic Systems

Konstantinos N. Genikomsakis

, Benjamin Bocquier

, Sergio Lopez

and Christos S. Ioakimidis

DeustoTech Energy, University of Deusto, Avenida de las Universidades 24, Bilbao, Spain

Icam Nantes, 35 Avenue du Champ de Manoeuvre, Carquefou, France

Department of Industrial Technologies, University of Deusto, Avenida de las Universidades 24, Bilbao, Spain

ERA Chair Holder ‘Net-Zero Energy Efficiency on City Districts’, Research Institute for Energy, University of Mons,

56 Rue de l'Épargne, Mons, Belgium

Keywords: Forecasting, Non-residential Building, Peak Shaving, Photovoltaic, Plug-in Electric Vehicle, Solar Power,

Valley Filling.

Abstract: This paper examines the concept of utilizing plug-in electric vehicles (PEVs) and solar photovoltaic (PV)

systems in large non-residential buildings for peak shaving and valley filling the power consumption profile,

given that the energy cost of commercial electricity customers typically depends on both actual

consumption and peak power demand within the billing period. Specifically, it describes a hybrid approach

that combines an artificial neural network (ANN) for solar irradiance forecasting with a MATLAB/Simulink

model to simulate the power output of solar PV systems, as well as the development of a mathematical

model to control the charging/discharging process of the PEVs. The results obtained from simulating the

case of the power consumption of a university building, along with experimental parking occupancy data

from a university parking lot, demonstrate the applicability and effectiveness of the proposed approach.

1 INTRODUCTION

Despite the fact that the high use of private cars,

combined with low vehicle occupancy, is known to

have negative impact on the quality of life and cause

serious environmental and social problems (Katzev,

2003; Wolfler Calvo et al., 2004), passenger cars are

still the dominant mode of transportation on land

across the EU, representing some 79.9% of all

passenger kilometres in 2012 (European

Commission, 2014). On the one hand, with the

renewed interest in electro-mobility, the

reintroduction of electric vehicles (EVs) in the

market has the potential to significantly reduce the

environmental impact of transportation activities. On

the other hand, passenger cars are largely

underutilized with respect to their primary purpose.

Kempton and Tomić (2005) estimate that personal

vehicles are typically used for less than 5% of the

time for transportation, hence plug-in electric

vehicles (PEVs) are potential candidates for

secondary applications in electricity markets while

parked, such as provision of peak power, spinning

reserves, and regulation to the grid through the

vehicle-to-grid (V2G) concept. Towards the

realization of the vision of the future smart grid

(Güngör et al., 2011), PEVs play a key role as an

energy storage resource to the grid and intelligently

interact with the electric utilities (Cheng et al.,

2014).

In this direction, the potential of the PEVs to

serve as a dispatchable load and provide power

system services, such as valley filling and peak load

reduction, has received particular attention in the

scientific literature (Brooks et al., 2010; Tie and Tan

2013). Indicative examples include an intelligent

charging scheme for PEVs to significantly reduce

the power system cost (Valentine et al., 2011), a

protocol for coordinating the charging of PEVs with

the electric grid (Zhang et al., 2014), as well as a

strategy based on a V2G control algorithm for peak

shaving and valley filling of grid power (Wang and

Wang, 2013). In this regard, the large scale

deployment of PEVs can leverage the penetration of

renewable energy sources (RES) into the power

system (Gerbelová et al., 2013; Mwasilu et al.,

Genikomsakis, K., Bocquier, B., Lopez, S. and Ioakimidis, C.

Utilizing Plug-in Electric Vehicles for Peak Shaving and Valley Filling in Non-residential Buildings with Solar Photovoltaic Systems.

In Proceedings of the 5th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS 2016), pages 179-188

ISBN: 978-989-758-184-7

179

2014).

At building level, in particular in residential or

small commercial ones, the integration of RES and

PEVs with capability to exchange electricity bi-

directionally (i.e., charge and discharge) has been

shown to allow for more efficient energy

management decisions (Genikomsakis et al., 2013;

Ioakimidis et al., 2014b). In this respect, forecasting

of the RES output ahead of time (e.g., with a time

horizon of 24 h), either in the form of solar power

(Ioakimidis et al., 2013), wind power (Ioakimidis et

al., 2014a; Ioakimidis et al., 2015) or both, is

beneficial for the home energy management systems

of the so-called electricity “prosumers”, which are

capable of not only consuming, but also producing

and storing electricity (Grijalva and Tariq, 2011).

This paper considers the case of valley filling

and peak reduction services by integrating PEVs and

RES in large non-residential buildings in the context

of vehicle-to-building (V2B), as a representative

example of a commercial electricity customer where

the energy cost depends on the actual usage

(consumption), as well as on the peak power demand

within the billing period. To this end, the present

work combines an artificial neural network (ANN)

with a MATLAB/Simulink model to forecast the

output of a solar photovoltaic (PV) system and

describes the development of a mathematical model

for peak shaving and valley filling the consumption

profile under the presence of PEVs, using as a case

study the consumption characteristics of a university

building, along with parking occupancy data from a

university parking lot. In this context, the main

contribution of this paper is to provide the integrated

framework for the simulation of the system under

study and the optimization of its operation, as well

as to examine a number of scenarios representing

different system configurations and assess the

potential output by applying the proposed approach

on real-world data from field measurements and

consumption profiles.

2 METHODOLOGY

This section presents the methodology followed to

develop the mathematical model for peak shaving

and valley filling the building’s consumption profile,

starting from the solar irradiance forecasting model,

which provides input to a MATLAB/Simulink

model to simulate the output of solar PV modules.

2.1 Forecasting of Solar Irradiance

As an intermediate step of forecasting the power

output of solar PV systems, the present work

employs the ANN proposed by Ioakimidis et al.

(2013) in order to forecast the solar irradiance for

the next 24 h. The reason for considering this

specific model for forecasting solar irradiance is

three-fold: i) it is suitable both for sunny and cloudy

days, ii) it combines forecasting accuracy with

computational efficiency, and iii) it is validated with

field measurement data. The key characteristic of

this approach is the use of statistical feature

parameters in the multilayer perceptron (MLP). The

structure of the feed-forward ANN employed

includes 2 hidden layers (Figure 1), with p=12 and

q=16 neurons respectively, while the input vector

consists of the following factors: (i) the average

solar irradiance G

Savg

in the last 24 h, (ii) the average

temperature T

avg

in the last 24 h, (iii) the maximum

of the third order difference of solar irradiance

TOD

max

, (iv) the normalized discrete difference

NDD between the surface and extraterrestrial

irradiance, and (v) the day of the year d. The output

layer is a vector of 24 components representing the

forecasted solar irradiance for the next 24 h.

The datasets for training the ANN are

normalized in the interval [-1,1] using equation (1),

where y

min

equals -1, y

max

equals 1, while x

max

and

min

are the maximum and minimum values in each

dataset.



 



min







max min

min

max min

yyxx

(1)

Figure 1: Structure of the ANN-based model for solar

irradiance forecasting.

The forecast of solar irradiance for a period of 6

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

180

days (using a resolution of 24 hourly values) in the

area under study in Bilbao, Spain, is illustrated in

Figure 2.

2.2 Simulation Model of PV Output

For the purposes of this work, the simulation of the

solar PV output is based on the model proposed by

Pandiarajan and Muthu (2011). More specifically, a

PV cell is represented with the equivalent electrical

circuit in Figure 3, thus its output current I is given

in equation (2).



ph d sh

II I I

(2)

where I

is the cell photocurrent, I

is the diode

current and I

is the shunt current. The relation

between the cell parameters, output current I and

output voltage V is expressed in the characteristic

equation (3).





()

exp 1













qV IR V IR

nkT

(3)

where I

is the reverse saturation current, q is the

elementary charge, R

is the series resistance, n is the

diode ideality factor, k is the Boltzmann's constant, T

is the absolute temperature, and R

is the shunt

resistance.

PV cells are combined together to form PV

modules, while the latter are inter-connected in

larger assemblies to form PV panels. In this work,

the model that describes the operation of an

assembly of N

cells in parallel and N

cells in series

is mathematically expressed in equations (4)-(7).

Specifically, equation (4) expresses the module

photocurrent I



( 298)

1000



 

PH SCr i

IIKT

(4)

where I

SCr

is the PV module short-circuit current (at

25 °C and 1000 W/m

), K

is the short-circuit current

temperature co-efficient at I

SCr

, and λ is the PV

module illumination.

Equation (5) expresses the module reverse

saturation current I

exp 1











SCr

NkAT

(5)

where V

is the open-circuit voltage and A is an

ideality factor.

Figure 2: Example output of forecasted solar irradiance

with the ANN-based model.

Figure 3: Equivalent circuit of a PV cell.

Equation (6) expresses the module saturation

current I

()exp















TBkTT

(6)

where E

is the band gap for silicon (= 1.1 eV), T

the reference temperature (= 298 °K), and B is an

ideality factor.

The module output current I

is given in

equation (7):



exp 1



 















PV PV S

PV P PH P

qV I R

ININI

NAkT

(7)

where V

is the output voltage and R

is the series

resistance of the PV module.

The aforementioned equations that describe the

operation of a PV module were modelled in

MATLAB/Simulink, as proposed by Pandiarajan

and Muthu (2011), with basic modifications to adapt

the model according to the specifications and

requirements of the present work. Indicatively,

Figure 4 illustrates the high-level representation of a

PV system model in MATLAB/Simulink, while

Figure 5 shows the output power of a single PV

module with N

= 1 cell and N

= 50 cells.

Utilizing Plug-in Electric Vehicles for Peak Shaving and Valley Filling in Non-residential Buildings with Solar Photovoltaic Systems

181

Figure 4: Model of PV system in MATLAB/Simulink.

Figure 5: Power output of 1 PV module.

2.3 Mathematical Model for Peak

Shaving and Valley Filling of

Power Consumption using PEVs

This subsection considers the charging and

discharging process of PEVs for a period of 1 day,

divided evenly in N intervals, based on the approach

proposed by He et al. (2012). The interval length τ is

chosen equal to 10 min, while the charging or

discharging power in each interval i is assumed to be

constant. Let also m denote a vehicle in the set of

vehicles M charging and discharging over the day.

By convention, it is assumed that the charging or

discharging power x

, for each m



M and i



N, is

negative during discharging and positive during

charging of the PEV battery. Assuming a slow

charging process, the constraint imposed by the

charging station on the maximum charging or

discharging power p

max

is expressed in (8).

max max

,  

pxp mM

(8)

A PEV m may charge and discharge during a

period T

, defined by the time of arrival t

arr

and

time of departure t

dep

to/from the parking lot of the

building. To this end, the presence of the PEV is

denoted by f

as defined in equation (9).

1 if interval is in period

0otherwise









(9)

Let y

be the load of charging/discharging the

PEVs in interval i, as given in equation (10).







imimi

(10)

Moreover, let E

ini

be the initial energy of the

PEV battery, E

cap

be its capacity (assumed equal to

24 kWh), and E

Tm+1

be the battery energy required

for the next trip of the PEV (assumed to be provided

by the user of the vehicle). The final energy of the

PEV battery E

fin

at the end of the charging period

is expressed in (11).

, ,













fin ini

m m mi mi T

ExfEmMiN

(11)

In addition, the lower and upper bounds of the

energy during the charging/discharging process of

the PEV battery are expressed in (12), i.e. the battery

energy cannot be less than 0 and cannot exceed its

capacity (He et al., 2012), where Q(i) denotes the set

of intervals prior to interval i.

()

0 , ,











ini cap

mmkmkm

kQi

xf E m M i N

(12)

The constraint in (13) specifies that the energy

charged to the PEV battery at the end of the period

is at least equal to that discharged from the

battery during T

, while the constraint in (14)

ensures that the energy exchange takes place only

when PEVs are present (parked).









mi mi

fmM

(13)

0 , ,



 

mi mi

fmMiN

(14)

where

is the binary complement of f

Furthermore, it is assumed that Pu

denotes the

power consumption of the building in the interval i,

and C is the constant defined in equation (15).

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

182





max( ) min( ) / 2

CPuPu

(15)

Mathematically, the objective of peak shaving

and valley filling the curve of the building’s power

consumption, while taking into the load y

from the

charging/discharging process of the PEVs and the

power output Ppv

from the PV system in each

interval i, is formulated in (16). For the purposes of

this work, this optimization problem is solved in

MATLAB using the fmincon solver.

Minimize ( )







ii i

zPuyPpvC

(16)

3 SIMULATION AND RESULTS

As a case study, the present work considers the

power consumption curve of a building at the

campus of University of Deusto, Spain. The aim is

to simulate the impact of controlling the

charging/discharging process of (potentially parked)

PEVs in order to flatten the power consumption

curve, as shown in Figure 6, by taking into account

real-world profiles of parking occupancy from

conventional cars at a university parking lot.

3.1 Simulation Setting and Scenarios

The developed model is employed to examine six

scenarios for different number of PEVs and PV

modules. Table 1 shows the configuration of the six

scenarios under study.

The required data for the parking occupancy of

PEVs were obtained from a field experiment, where

a team of participants recorded the parking spot

(place) and time of (conventional) cars at a specific

parking lot of the university campus, having a

capacity of 65 cars (Genikomsakis et al., 2015).

Indicatively, the occupancy of the 8 (randomly

chosen) parking spots employed for scenarios 1 and

2 is illustrated in Figure 7, where the colored bars

denote the time of presence of the different vehicles

(horizontal axis) at each parking spot (vertical axis).

3.2 Simulation Results

The simulations were carried out in MATLAB on

the basis of the scenarios defined in the previous

subsection. As already pointed out, it is assumed that

the PEVs can discharge their battery to provide

power to the building and reduce its power

consumption, whereas they add extra load to the

power consumption of the building when they

charge their battery. Figures 8 and 9 illustrate the

obtained results of the scenarios 1 and 2

respectively.

Figure 6: Power consumption curve of the building under

study for a typical winter day.

Figure 7: Occupancy of 8 randomly chosen parking spots.

Table 1: Overview of the six simulated scenarios.

Scenario Number of parking

spots

Number of PV

modules

1 8 300

2 8 600

3 35 300

4 35 600

5 65 300

6 65 600

As a first observation, the results of scenarios 1

and 2 indicate that the impact of the available PEVs

at the 8 parking spots on the power consumption

curve of the building is rather low. Nevertheless,

there is a more pronounced effect when considering

both the contribution of PEVs and the PV

production. As expected, the contribution of the PV

system becomes higher during midday hours when

the PV production reaches its peak.

Figure 9 shows that the power consumption

curve of the building is peak shaved between 7:00

am and 12:00 pm. After this period, the charging of

PEVs causes the red curve representing the final

power consumption to cross the green curve

representing the building power consumption minus

Utilizing Plug-in Electric Vehicles for Peak Shaving and Valley Filling in Non-residential Buildings with Solar Photovoltaic Systems

183

the photovoltaic power production. Moreover, the

final power consumption curve does not cross the

initial power consumption curve (in blue color) until

6:30 pm, indicating that the PV production is enough

to cover the charging of the PEVs for this period. At

the end of the day, after 6:30 pm, the final

consumption is slightly higher or equal to the initial

consumption.

Figures 10 and 11 illustrate the results obtained

for scenarios 3 and 4, where the number of available

parking spots for charging/discharging the PEVs

amounts to 35, while the number of simulated PV

modules is 300 and 600 units respectively. The

increase of the available PEVs flattens the power

consumption curve of the building, approaching the

target constant as a result of the improved valley

filling and peak shaving. The majority of the PEVs

discharge at peak power consumption and charge

when the building’s power consumption is at lower

levels.

When the PV production is doubled (Figure 11),

the valley filling effect becomes more intense.

Moreover, the effect of lowering the power

consumption curve of the building takes place

mostly between noon and 3:00 pm (compared to

scenario 3), where the PV production is at its highest

Figure 8: Scenario 1 results.

Figure 9: Scenario 2 results.

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

184

Figure 10: Scenario 3 results.

Figure 11: Scenario 4 results.

level. The roughness of the final consumption curve

is mainly due to the randomness of the

charging/discharging process of the PEVs.

The results of scenarios 5 and 6 are depicted in

Figures 12 and 13. Specifically, the last two

scenarios consider 65 available parking spots for

charging/discharging of PEVs. In both cases, the

results reveal that the number of PEVs is high

enough to regulate by themselves the power

consumption of the building, with the PV production

contributing to a lesser extent to the smoothing of

the final consumption curve. It is obvious that in

scenarios 5 and 6, the final curve of the building

consumption is almost flat, resulting in effective

peak shaving and valley filling. In particular, it

remains almost at the same level from 6:00 am until

8:00 pm by shifting the charging of the PEVs, more

notably after 4:00 pm, where the initial power

consumption curve of the building would decrease

further.

In Figure 13, which illustrates the simulation

results for the last scenario considering 600 PV

modules, the pattern of the final consumption curve

presents no significant changes (compared to

Utilizing Plug-in Electric Vehicles for Peak Shaving and Valley Filling in Non-residential Buildings with Solar Photovoltaic Systems

185

scenario 5), despite the fact that the increased PV

production reduces further the final power

consumption.

Table 2 summarizes the main findings of this

work. As expected, the highest peak shaving occurs

under scenario 6. The high number of available

PEVs at the 65 parking spots in combination with

the high additional energy contribution of the 600

PV modules, offers enough flexibility to the system

resulting in more than 25% peak shaving compared

to the initial power consumption of the building. In

addition, it is noted that the power gap with the

Figure 12: Scenario 5 results.

Figure 13: Scenario 6 results.

Table 2: Comparative simulation results.

Parking spots 8 8 35 35 65 65

PV modules 300 600 300 600 300 600

Final peak consumption (kW) 420 395 380 370 355 350

Decrease compared to initial peak (%) 10.1 15.4 18.6 20.8 24.0 25.1

Gap with constant C (kW) 152 127 112 102 87 82

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

186

reference flat consumption, i.e. constant C, becomes

smaller as the number of available PEVs and PV

modules increases.

4 CONCLUSIONS

This paper examines the impact of PEVs and PV

production as a means of providing peak shaving

and valley filling services in the context of V2B.

More specifically, it employs the profiles of power

consumption and parking occupancy from a building

and a parking lot at University of Deusto, Spain, in

order to provide the required input to the proposed

model and simulate a number of scenarios for the

envisaged system.

To this end, the present paper initially described

the integration of an ANN-based solar irradiance

forecasting model with a MATLAB/Simulink model

to simulate the output of solar PV modules. Next, a

mathematical model was developed and solved in

MATLAB in order to examine and quantitatively

analyze the impact of connected PEVs and PV

production on the power consumption of the

building.

As confirmed also by the simulation results, the

higher the number of available PEVs and PV

modules, the closer the achievable load curve of the

building comes to the target (flat) curve. On the one

hand, the results demonstrate the feasibility of the

peak shaving and valley filling approach proposed in

this paper, and on the other hand, they highlight the

importance of the number of connected vehicles on

its effectiveness.

As a concluding remark, it is noted that this work

employed a deterministic approach to model the

consumption of the building, the presence of PEVs

at the parking lot and the energy of their battery both

in the initial and final state. Hence, directions for

future work include incorporating the uncertainties

in the arrival and parking duration of the PEVs, the

initial and final energy of their battery, as well as the

consumption profile of the building.

REFERENCES

Brooks, A., Lu, E., Reicher, D., Spirakis, C., Weihl, B.,

2010. Demand dispatch. IEEE Power and Energy

Magazine 8(3), pp. 20-29.

Cheng, X., Hu, X., Yang, L., Husain, I., Inoue, K., Krein,

P., Lefevre, R., Li, Y., Nishi, H., Taiber, J.G., Wang,

F.-Y., Zha, Y., Gao, W., Li, Z., 2014. Electrified

vehicles and the smart grid: The ITS perspective.

IEEE Transactions on Intelligent Transportation

Systems 15(4), pp. 1388-1404.

European Commission, 2014. EU transport in figures:

Statistical Pocketbook 2014, pp. 1-138.

Genikomsakis, K. N., Angulo Gutierrez, I., Thomas, D.,

and Ioakimidis, C.S., 2015. Simulation and Design of

a Fast Charging Battery Station in a Parking Lot of an

e-Carsharing System. In Proceedings of the 4th

International Conference on Renewable Energy

Research and Applications, ICRERA 2015.

Genikomsakis, K. N., Ioakimidis, C. S., Eliasstam, H.,

Weingraber, R., Simic, D., 2013. A non-myopic

approach for a domotic battery management system. In

Proceedings of the 39th Annual Conference of the

IEEE Industrial Electronics Society, IECON 2013, pp.

4522-4527.

Gerbelová, H., Genikomsakis, K. N., Ioakimidis, C. S.,

2013. Electric vehicles charging under a G2V and

V2G approach: The case of the portuguese power

system. In Proceedings of the 26th International

Conference on Efficiency, Cost, Optimization,

Simulation and Environmental Impact of Energy

Systems, ECOS 2013.

Grijalva, S., Tariq, M. U., 2011. Prosumer-based smart

grid architecture enables a flat, sustainable electricity

industry. IEEE PES Innovative Smart Grid

Technologies Conference Europe, ISGT Europe.

Güngör, V. C., Sahin, D., Kocak, T., Ergüt, S., Buccella,

C., Cecati, C., Hancke, G.P., 2011. Smart grid

technologies: Communication technologies and

standards. IEEE Transactions on Industrial

Informatics 7(4), pp. 529-539.

He, Y., Venkatesh, B., Guan, L., 2012. Optimal

scheduling for charging and discharging of electric

vehicles. IEEE Transactions on Smart Grid 3(3), pp.

1095-1105.

Ioakimidis, C. S., Genikomsakis, K. N., Dallas, P. I.,

Lopez, S., 2015. Short-term wind speed forecasting

model based on ANN with statistical feature

parameters. In Proceedings of the 41st Annual

Conference of the IEEE Industrial Electronics Society,

IECON 2015, pp. 971-976.

Ioakimidis, C. S., Lopez, S., Genikomsakis, K. N.,

Rycerski, P., Simic, D., 2013. Solar production

forecasting based on irradiance forecasting using

artificial neural networks. In Proceedings of the 39th

Annual Conference of the IEEE Industrial Electronics

Society, IECON 2013, pp. 8121-8126.

Ioakimidis, C. S., Oliveira, L. J., Genikomsakis, K. N.,

2014a. Wind power forecasting in a residential

location as part of the energy box management

decision tool. IEEE Transactions on Industrial

Informatics 10(4), pp. 2103-2111.

Ioakimidis, C. S., Oliveira, L. J., Genikomsakis, K. N.,

Dallas, P. I., 2014b. Design, architecture and

implementation of a residential energy box

management tool in a SmartGrid. Energy 75, pp. 167-

181.

Katzev, R., 2003. Car Sharing: A New Approach to Urban

Transportation Problems. Analyses of Social Issues

Utilizing Plug-in Electric Vehicles for Peak Shaving and Valley Filling in Non-residential Buildings with Solar Photovoltaic Systems

187

and Public Policy 3(1), pp. 65-86.

Kempton, W., Tomić, J., 2005. Vehicle-to-grid power

fundamentals: Calculating capacity and net revenue.

Journal of Power Sources 144(1), pp. 268-279.

Mwasilu, F., Justo, J. J., Kim, E.-K., Do, T. D., Jung, J.-

W., 2014. Electric vehicles and smart grid interaction:

A review on vehicle to grid and renewable energy

sources integration. Renewable and Sustainable

Energy Reviews 34, pp. 501-516.

Pandiarajan, N., Muthu, R., 2011. Mathematical modeling

of photovoltaic module with Simulink. In Proceedings

of the 2011 1st International Conference on Electrical

Energy Systems, ICEES 2011, pp. 258-263.

Tie, S. F., Tan, C. W., 2013. A review of energy sources

and energy management system in electric vehicles.

Renewable and Sustainable Energy Reviews 20, pp.

82-102.

Valentine, K., Temple, W. G., Zhang, K. M., 2011.

Intelligent electric vehicle charging: Rethinking the

valley-fill. Journal of Power Sources 196(24), pp.

10717-10726.

Wang, Z., Wang, S., 2013. Grid power peak shaving and

valley filling using vehicle-to-grid systems. IEEE

Transactions on Power Delivery 28(3), pp. 1822-1829.

Wolfler Calvo, R., de Luigi, F., Haastrup, P., Maniezzo,

V., 2004. A distributed geographic information system

for the daily car pooling problem. Computers and

Operations Research 31(13), pp. 2263-2278.

Zhang, L., Jabbari, F., Brown, T., Samuelsen, S., 2014.

Coordinating plug-in electric vehicle charging with

electric grid: Valley filling and target load following.

Journal of Power Sources 267, pp. 584-597.

SMARTGREENS 2016 - 5th International Conference on Smart Cities and Green ICT Systems

188