The discharge of toxic and harmful gases like N_{2}O, CH_{4 }and CO_{2 }traps the heat of the sun which is radiated back to space from earth, resulting in global warming that leads to melting of snow caps, expanding ocean area and increase in sea level. TheInter Governmental Panel (IPC) on climate change forecasts that the rise of temperature from 2.5 to 10 degree Fahrenheit over the next century causes serious problems to human life

To control the limits of excess pollution, electrification is the most important solution. Electric vehicle uses electricity as a fuel and contribute to clean and green environment with zero carbon emissions. EVs have more superior qualities than a conventional vehicle. EVs can use electricity from public places, charging stations and also from renewable energy sources like sun and wind. Along with green environment, EVs provide ancillary services to the grid

Although a lot of research works are going worldwide to bring the optimal coordination over charging and discharging of EV fleet without disturbing the load profile to minimize the mismatch between load and power generation along with renewable energy penetration but this problem is still challenging when the loads are frequently changing continuously. The continuous and unpredictable change in load profile integrated with grid calls a smart infrastructure for optimal scheduling of EV charging and discharging to deal with the peaks and valleys in the load profile of the electric system. In

Three strategies for charging/discharging coordination of EV are proposed in

Coordination of Plugged in EVs (PEV) charging and discharging is presented in

A decentralized optimal scheduling problem for PEV charging/discharging is solved by using economic model predictive control technique in

The major limitation is lying in the scheduling problem of EVs are basically because of V2G operation which requires a primitive control of the charging/discharging behaviors of EVs to tackle the following issues- 1) Stochastic and variable regulation demand 2) Potential conflicts between EVs charging and suitable provision of the regulation service 3) Computational complexity and security issues incurred by the scheduling process of EVs. In this work, an optimal charging/discharging scheduling scheme is proposed to resolve all of these key issues and also to reshape the load profile to improve the overall regulation performance integrated with grid.

The main objective of the proposed centralized scheduling approach is to break the required charging time for 24 hours into several small charging intervals within the parking time, (Rezaei, 2014). In each scheduling slot, Δt, if enabled, a CS can charge the EV connected to it only for the duration of the scheduling slot providing a charging packet

In this scheme, one day scheduling of EV for charging and discharging is considered with a total of 24 hours with 24 intervals. The interval set is denoted by L. Whereas, the length of each interval is taken for one hour which is denoted by l such that l=1 hour. On the other hand, V2G capable EV group can performs both charging and discharging in one day scheduling is denoted by E. At the interval n, the status of EV is also represented by

The plug in time of EV e into the charging spot is represented by

For centralized scheduling scheme of EV charging and discharging, following assumptions are considered based on reference

The proposed centralized scheme is aimed to flatten the peaks and valleys of load profile by charging EV during light loads and discharging EV during peak loads to optimize charging/discharging power

The four equations stated above represent the constraints for centralized optimal EV scheduling. Constraints in eq. (1) represents the total load included with EV load and non-EV load. Constraints in eq. (2) represents instant energy level of the EV battery which must not be zero or not greater than EV battery capacity

Decentralized scheduling scheme is formulated optimally to schedule electric vehicle (EV) charging. The proposed scheme exploits the elasticity of EV loads to fill the peak in electric load profiles. So here we first formulated EV charging scheduling problem as an optimal control problem, whose objective is to impose a generalized notion of peak filling, and study properties of optimal charging profiles. Then we proposed a decentralized scheme to iteratively solve the optimal control problem. In each iteration, EVs are updated with their charging profiles according to the control signal forecasted by the utility. The scheme converges to optimal charging profiles irrespective of the specification of EVs

Decentralized scheduling is based on the groups. According to the charging locations of EV fleet, they are divided into groups like a parking lot, residential garage. In this work, a group of 100 EVs is chosen for analyzing the proposed scheme. Each group is headed by the local controller. All local controllers of each group are controlled by a central controller. Local controllers (LC) communicate with a central controller in the utility company. Using similar day analysis method central controller forecasts the load for the scheduling day and these forecasted load data is sent to all local controllers at the beginning of the day that comes under the central controller and also collects actual charging load for each EV.

Each local controller contacts with other local controllers at the beginning of each interval n to collect EV information, current on going EV set

At the starting of the interval,

Load at each interval is the forecasted load, obtained from averaging load values of similar weather conditions. The forecasted load is denoted as

The four equations stated above represent the constraints for decentralized optimal EV scheduling. Constraints in eq. (5) represents the total load included with EV load at the current sliding window and non EV load at the current interval. Constraints in eq. (6) represents the instant energy level of the EV battery which must not be equal to zero or not greater than EV battery capacity

The merits of this scheme can be handled with a large number of EV population and at each interval local controller updates charging power by gathering EV information. Thus, this scheme responds quickly to the dynamic EV arrivals.

The following settings are considered for the

S. N EV No. EV Arrival time (Specific Timing of Arrival in 24 Hours) EV Departure Time (Specific Timing of Departure in 24 Hours) 1 1 2 14 2 2 2 14 3 3 11 23 4 4 13 24 5 5 18 24

EV model No. of EVs Battery Capacity Range Maximum Charging Power Nissan leaf 2017 200 30kWh 107 miles 12kW

Modeling of EV information such as arrival time, charging period and the initial energy is as follows. Though EV arrival is evenly distributed among all intervals, expected chances for the arrival of EVs is less than 30 number of EVs for each interval. Each EV charging period lies in between 4 to 12 hours and the initial energy of each EV lies in between 0 to 80% of the battery capacity. Number of EVs in EV fleet is set to 200.

Centralized scheduling scheme requires perfect information of EV and load so that the actual load is used in simulation. However, actual load in the future interval is impossible to determine. In decentralized scheduling scheme, forecasted loads are used in the simulation. Hence, this scheme is called practical solution of EV scheduling.

Interval EV load (kW) Total load (kW) Centralized Scheme Decentralized Scheme Centralized Scheme Decentralized Scheme 1 -78.13363108409 -122.735931457548 2276.41092605877 2231.80862568531 2 119.103093079389 -62.5192363242416 2276.41012165082 2094.78779224719 3 269.837650180049 141.022085683165 2276.40980732291 2147.59424282602 4 350.494527418265 327.700753287288 2276.40969884684 2253.61592471586 5 340.275007370227 430.038921325179 2276.41005022737 2366.17396418232 6 304.361456242981 450.206838442673 2276.41061338584 2422.25599558553 7 204.581941383818 346.806650021991 2308.00869852668 2450.23340716485 8 270.971344672741 270.971373913276 2585.05688752988 2585.05691677042 9 238.478880154260 238.479063817813 2773.80275158283 2773.80293524638 10 161.466231187618 161.466054210181 2863.95064547333 2863.95046849590 11 167.591450766293 167.591493418123 3059.86196505201 3059.86200770384 12 91.5884753476354 192.887098696143 3168.74083249049 3270.03945583900 13 4.50044462949983 79.7046293132441 3168.74250177236 3243.94668645610 14 -43.183105714434 -38.1785872272166 3168.74303714271 3173.74755562993 15 -56.176903299617 -43.0880192799744 3168.74316812895 3181.83205214860 16 -51.111912604768 -62.5652469708220 3168.74320168095 3157.28986731489 17 -47.107606516604 -109.275903966870 3168.74313634054 3106.57483889027 18 -32.530073162933 -123.901871047992 3168.74294112278 3077.37114323772 19 43.5165682103781 -89.3744463320328 3168.74243963895 3035.85142509654 20 225.233676171407 196.199860687819 3168.74160474284 3139.70778925925 21 265.123823151481 263.070589196822 3168.74089458005 3166.68766062539 22 333.576812585843 314.585238968978 3168.74034115727 3149.74876754041 23 531.247146319109 684.613859633050 3168.73966060482 3322.10637391876 24 575.999882539769 575.999906264426 2970.85009682548 2970.85012055014

EVs charging load in centralized scheduling scheme, at each interval is shown in

Proposed centralized scheme achieves better results than decentralized scheme as shown in

Optimal scheduling schemes require perfect information of EV and load so that the actual load can be used in simulation. However, actual load in the future interval is impossible to determine. In decentralized scheduling scheme, forecasted loads are used in the simulation. Hence, this scheme is called practical solution of EV scheduling. Two scheduling methods such as centralized and decentralized schemes are proposed in this study. Proposed schemes are evaluated in Matlab simulations for one-day scheduling of EV charging and discharging. From simulation results, it is concluded that both schemes are achieving good results in terms of EV scheduling and load reshaping but the centralized scheme is accurate in results. Whereas, decentralized scheme results are nearly equal to the centralized scheme.The centralized scheme requires perfect information regarding EV and load which is not possible in practice. Overall, the decentralized scheme is based on forecasted data which can give the practical solution.