Keywords

1 Introduction

The recent advancements in power system and restructuring policy allowed private players to participate in power generation due to which rapid penetration of distributed energy resource (DER) is observed in the distribution network (DN). Distributed generators (DG), electric vehicle (EV) and other interactive loads and some new features emerge in DN. Due to the intermittent nature of renewable sources like Photovoltaic (PV) and Wind, the presence of a battery energy storage system (BESS) becomes crucial. India holds 5th position in renewable energy installed capacity in the world and the government has taken aggressive steps to top the list. In line with this, the wind energy installed capacity is 38GW as of today and the National Institute of Wind Energy (NIWE) proposed that 302GW wind energy can be extracted from the windy states of India. Solar energy installed capacity is around 41GW and the state of Karnataka tops the list in India. Apart from solar and onshore wind, the Ministry of Power (MoP) and NIWE have ventured to take up offshore wind projects (ZERO to 5GW) in the states of Tamilnadu and Gujarat. It is clear from the developments that the Indian government is showing tremendous interest in promoting renewable energy. This gives rise to the rapid penetration of renewable sources in the distribution network which makes it difficult for the distribution system operator (DSO) to monitor and control. The intermittent nature of DGs need large-scale battery storage systemand in this case EV batteries can serve the purpose. EVs act as a distributed source or a connected load. A group of EVs used for power transactions with the grid is known as Vehicle-to-Grid (V2G) [1,2,3,4] which supports shaving peak, filling valley, load levelling and provides support to the grid. A conceptual framework of EV aggregation as a Cyber Physical System (CPS) is shown in Fig. 1 [5,6,7].

Fig. 1
figure 1

EV Aggregation as a Cyber Physical System

Full size image

Some DN features include intermittent and random power sources, changing operating modes of the power grid and islanding morphology which complicate the operation of DN. DNs are complex due to the presence of multiple nodes, short distances, amplitude and angle difference in between the nodes is smaller, dynamically changing and lack of standard documentation. Due to these complexities, dynamic monitoring system with high accuracy and precision for situational awareness is preferred. Most of the utilities worldwide use SCADA to monitor DN. SCADA suffers from the data rate which is around 2–4 samples/sec, whereas PMUs work at higher data rates with high-level accuracy and has exhibited excellent dynamic performance. For real-time monitoring and assessment, PMUs are used in WAMS [8]. EVs have a significant impact on the grid if there is no coordinated/uncoordinated charging and it might lead to grid instability. So, coordinated charging can curb this issue. There are four modes of operation in each of the quadrants. In this work, we propose EVs to function in the fourth quadrant in which active power (P) is consumed and reactive power (Q) is injected into the grid. This strategy helps the grid to meet the time-varying intermittent load demand. In the existing literature, the application of multiagent-based µPMU communication is not reported and still unexplored for coordinated charging in the distribution network. Agent-based µPMU communication is proposed for coordinated EV charging in smart distribution network where EVs operate in the fourth P-Q quadrant.

2 Phasor Measurement Unit (PMU)

The data sensed and processed is time-stamped and sent to PDC located at the control center through the communication network as shown in Fig. 2 [3,4,5]. PMUs have an excellent dynamic performance which could be used to solve challenges in the distribution network. A low-cost micro PMUs (µPMU) are developed and are used in the distribution network for monitoring and surveillance. µPMU are very handy in addressing distribution network issues as compared to conventional PMUs used in transmission network [6,7,8,9,10,11,12,13,14,15]. The information of each PMU is being communicated to the PDC and Energy Management System, (EMS)/WAMS using communication links. Communication links, sensors and actuators are prone to and highly vulnerable to cyber physical attacks.

Fig. 2
figure 2

Phasor Measurement Unit

Full size image

3 EV Aggregation in Smart Distribution Grid

EVs are widely accepted and have great potential to address demand response due to their flexibility and sizeable power rating. V2G supports the grid requirements and will be an effective tool to manage the time-varying load demand [16]. Aggregator agents act as an interface between EV owners and the DSO and facilitate interested EV owners to participate in V2G operation by providing suitable resources such as parking lot, charging slot with free parking space.

The grid and load balance equation is given by

$$P_G + \sum_{n = 1}^N {P_{EV,i,k,disch\arg e} = P_L + } \sum_{n = 1}^N {P_{EV,i,k,ch\arg e} }$$
(1)

where \(P_G\) and \(P_L\) generation plant power and grid load.

\(P_{EV,i,k,disch\arg e}\) and \(P_{EV,i,k,ch\arg e}\) EV discharging and charging rate.

SOC of EV battery is given by

$$SOC_{i,e,t} = SOC_{i,e,t - 1} + \eta_{i,e} \frac{{P_{i,e,t}^{EV} \Delta k}}{{E_{i,e}^{\max } }}$$
(2)
$$\begin{aligned} & SOC_{\min } \le SOC_{EV,i,k} \le SOC_{\max } \\ & SOC_{EV} \le SOC_{EV,\min } ;ch\arg ing \\ & SOC_{EV} \ge SOC_{EV,\max } ;disch\arg ing \\ \end{aligned}$$
(3)

where

\(\eta\) charging efficiency;\(E^{\max }\) EV capacity; \(t\) time.

3.1 Distribution Grid Model

Distribution grid is a very complex structure due to the participation of many entities such as DGs, EVs and other loads. Flexible loads (FL) help the grid in flattening the load curve, and the objective of the grid is to maximize the FL penetration [17], and is given by

$$O = \sum_{i,k} {P_{i,k}^{FL} }$$
(4)

where

$$\begin{aligned} P_{i,k}^{FL} & = \Re (V_{i,k} \overline{I}_{i,k}^{FL} ) \\ Q_{i,k}^{FL} & = \mathfrak{I}(V_{i,k} \overline{I}_{i,k}^{FL} ) \\ \end{aligned}$$
(5)

DSO sends PFL and QFL to the aggregators at each node i.

Fairness index for the equal proportion of EV load penetration corresponding to the base loads at all nodes

$$F_k = \frac{{P_{i,k}^{FL} }}{{P_{i,k}^{zl} + P_{i,k}^{ml} + P_{i,k}^{pl} }}$$
(6)

The inequality constraints are

$$V_i^{\min } \le |V_{i,k} | \le V_i^{\max }$$
(7)

To operate EV in the fourth quadrant, the following limits are imposed on net reactive power dispatch

$$- Q_{i,k}^{\max } \le Q_{i,k}^{FL} \le 0$$
(8)

3.2 EV Load Model

EVs arrival to the charging station is random and highly stochastic in nature in nature. Objective function for EVs charging total cost minimization

$$\psi_m = \sum_k {\rho_k } \sum_e {P_{i,e,k}^{EV} \Delta k}$$
(9)

where, ρ is energy price, e is EV number, EV is load and \(\Delta k\) is time interval.

EVs consuming power must follow:

$$P_{i,e,k}^{EV}{}^2 + Q_{i,e,k}^{EV}{}^2 \le R_{i,e}^2$$
(10)

where, R is rating of charging slot.

For EVs to operate in the fourth quadrant

$$\begin{aligned} & P_{i,e,k}^{EV} \ge 0 \\ & Q_{i,e,k}^{EV} \le 0 \\ \end{aligned}$$

Grid constraints are incorporated by

$$\begin{aligned} & \sum_e {P_{i,e,t}^{EV} \le P_{i,t}^{FL} } \\ & \sum_e {Q_{i,e,t}^{EV} \ge Q_{i,t}^{FL} } \\ \end{aligned}$$
(12)

3.3 Communication Framework for µPMU-Based Coordinated EV Charging

DSO communicates with µPMUs and exchanges the crucial information for EV integration to the grid such that the P is consumed and Q is injected to avoid adverse effects on the grid. Uncoordinating charging give rise to feeder losses, voltage deviation and overloading distribution transformers. Research studies demonstrate that 45% penetration of EVs leads to about 50% transformer overloading and 25% increase in losses in the distribution network. It is also reported in the literature that 40% uncoordinated EV penetration will result in the replacement of 50kVA transformers. EVs can supply/consume Q at any SOC. Figure 3 shows µPMU based coordinated EV charging.

Fig. 3
figure 3

Framework for µPMU based coordinated EV charging

Full size image

Multi-agent system (MAS) communication reduces the computational burden and avoids overhead bits. A conceptual view of the proposed communication framework is shown in Fig. 4. The agent-based µPMUs are placed at each bus where EV aggregators are connected [18]. The static agents (SA) will communicate this information to the mobile agent (MA) and send the same information to PDC. PDC will forward it to the DSO. DSO will send this information to the grid. EVAs are apprised of the P consumption and Q injection bounds. The agents proposed in this model are grid agent (GA), DSO Agent (DSOA), Electric Vehicle Aggregator agent (EVAA) [19,20,21,22].

Fig. 4
figure 4

Agent-based analytical model

Full size image

Mobile Agent (MA): DSO periodically generates MA and is sent to every µPMU requesting for the power transaction with the grid. After nth PMU MA dies.

Static Agents (SA): Available at both PMU and PDC.

PMU Agent (PMUA): PMUA provides information about power exchange with the grid to PDC.

DSO Agent (DSOA): DSOA is a static agent (SA) which estimates all parameters associated with power and voltage. It will estimate the amount of power that can be with the grid and subsequently send this response to GA.

Grid Agent (GA): GA receives the information from AA and acts based on the load's requirement.

4 Results and Discussion

The mathematical models were developed in General Algebraic Modeling Language (GAMS) and solved using CIPLEX and KNITRO solvers. The case study was carried out on IEEE 13 bus test systems. In coordinated and controlled charging, DSO provides bounds (P and Q) to EVA at each node. Mobile C was interfaced with MATLAB for MAS communication. IEEE 13 bus test system was used for the simulation study.

EVs were made to operate in the fourth P-Q quadrant, where EVs consumed active power and supplied reactive power to the grid. For case 1, EVA was connected at bus 2, and using the coordinated EV charging based on the information received by DSO, it is observed from Fig. 10 that there is constant reactive power support to the grid which varies from 0.536 to 0.699 MVAr. Figure 5 also depicts that EVs are charged throughout the day without violating grid constraints using coordinated EV charging. EVs are charged during the night based on DSO information i.e they draw active power to the tune of 1.9 MW from the grid with a reactive power support of 1.65MVAr. Base case results demonstrate that active and reactive power losses are 0.8 MW and 0.6278 MVAr, respectively with voltage profiles maintained at bus 1 and bus 2 respectively.

Fig. 5
figure 5

Coordinated EV charging with EVA at bus 2

Full size image

In case 1, DG with a capacity of 6 MW was connected at bus 2 and it is observed that the power losses are reduced to 0.5287 MW and 0.4151 MVAr. There is improvement in voltage level also at bus 2, bus 3 and bus 4, respectively. In case 2, DG with a capacity of 6 MW and 3 MW are connected at bus 2 and bus 11, respectively. It is very interesting to note that the voltage profile improves at all the buses as per the limits specified by IEEE 1547 and there is a significant reduction in active and reactive power losses of 0.1827 MW and 0.1443 MVAr.

In case 2, EVAs are connected at bus 2 and bus 11 operating in the fourth P-Q quadrant. From Fig. 6, it is clear that the voltage profile is well within the prescribed limits and EVs consume active power and supply reactive power. At 2 PM, EVs draw active power of 1.5 MW and supply a reactive power of 0.572 MVAr. From 3 to 5 AM, the reactive power support of 0.672 MVAr and draws active power of 0.5 MW. From 7 AM to 3 PM, the average active power drawn is 0.41 MW and reactive power support is 0.61 MVAr. From 8 to 10.30 PM, EVs charging reaches a value of 1.558 MW and supports reactive power of 0.699 MVAr. The voltage levels are maintained throughout and coordinated charging serves the purpose of promoting EVs connecting to the grid.

Fig. 6
figure 6

Coordinated EV charging with EVA at bus 2 and 11

Full size image

5 Conclusion

Mathematical models were developed and presented for EVs in the distribution grid and intermittent EV load. Multiagent-based micro PMU communication for coordinated EV charging in smart distribution network is presented. It is observed that when EVs are made to operate in the fourth P-Q quadrant, there is a continuous supply of reactive power which helps in maintaining the voltage levels and at the same time EV can consume active power as prescribed by DSO. It is found that the results obtained are significant in terms of reduced power losses and improved voltage profile. The work presented is a concept of Cyber Physical System and serves as a platform for the researchers venturing into intelligent transportation system (ITS) and EV integration to the grid.