Planning and operation of EV charging stations by chicken swarm optimization driven heuristics

Successful deployment of electric vehicles demands for establishment of simple reachable charging stations (CSs). Scheduling and action of CSs is a composite problem and that should not affect the smooth operation of the power grid. The present paper attempts to solve the planning and operation of CSs by a novel chicken swarm optimization-based heuristics. The placement of CS is modelled in a multi-objective framework as cost-effective parameters secures the operation of the power grid. Further, the operation of CSs is examined for three scenarios such as uncoordinated charging, coordinated charging, as well as bidirectional vehicle to grid. The proposed approach is tested on IEEE 33-bus, and on a distribution network of Guwahati, India.


INTRODUCTION
In recent years, researchers and environmentalists are preoccupied with fossil fuel depletion, degradation of air quality, and energy crisis. Electric vehicles (EVs) are a clean mode of transportation and are viable alternatives to deal with the aforementioned problems. However, successful deployment of EVs calls for enlargement of charging station (CS). The planning and operation of CS are critical aspects. Improper planning and operation of CS may be detrimental to the power grid resulting in voltage instability, degraded reliability, increased power losses, and harmonic distortion [1][2][3][4][5]. Globally, the planning and operation of charging stations have attracted much attention from researchers to deal with various problems [6,7]. Despite their advantages, EVs are not becoming widespread at the desired level since there are no common charging stations, and the reason for this fear is that the private traditional vehicles are on the road [8,9]. To alleviate this problem, it is assumed that car parks can be used as charging places. Normally, the EVs are not used for a long time as they are often left in parking lots. For this reason, these long times can be measured as a prospect to recharge EVs in smart car parks [10]. The focus of this technology is to prevent damage to the grid by multiple EVs and HEVs being charged simultaneously. The aim of this paper is to ensure the satisfaction of EV and HEV users while eliminating the negative effects [11]. There is a need for a control mechanism to control the power supplied to parked vehicles which is fed from the grid as well as by other forms of electricity production [12][13][14].
The operation of charging stations signifies the charging strategy that will be adopted in the charging stations such as uncoordinated charging, coordinated smart charging etc. In [15,16], authors have analysed the advantages of smart charging schemes and found that coordinated charging is beneficial. In [17], authors provided a DR strategy of EV CS by using dynamic programming. In [18], the authors presented a two-stage linear programming-based approach for the operation of charging stations. In [19], authors have proposed an adaptive strategy to manage EV charging load. Further, in [20], the authors presented a load management strategy in EV charging stations in the presence of renewable energy sources.
Researchers have made significant attempts to improve energy efficiency of CSs. In a category, researchers have considered different technologies such as renewable energy sources, ESS and DR programs in studying the operation of energy systems in the presence of EVs. The authors have proposed a  [17] Operation A DR strategy of EV CS by using dynamic programming [18] Operation A two-stage linear programming-based approach for operation of charging stations [19] Operation An adaptive strategy to manage EV charging load [20] Operation A load management strategy in EV charging stations in presence of renewable energy sources multi-objective model for a PV-based, intelligent electric-vehicle CS in connection with a demand-response program in [21] to satisfy both the environmental and economic issues of CSs. A multi-agent approach for coordination of EVs, combined with a renewable-based micro grid, is proposed in [22] for providing vehicle-to-home service. The multi-agent coordination is performed by exchanging information among various control elements. The authors, in [23], have presented a self-supporting model for smart grids to minimise costs during outages by using EV batteries to fulfil the demands of the grid. Accordingly, a multi-agent scheme consisting of a micro grid, home, and EV is designed for managing outages in the smart grid. Also, the penetration of fuel cell technology as a hydrogen-storage system and electrolyze, along with time-of-use DR, have been investigated [24] in conjunction with optimal scheduling of CSs. A summary of research works on planning and operation of charging stations is presented in Table 1. From Table 1, it is observed that most of the research works have dealt with placement and operation of charging stations separately. This work considers planning and operation of charging stations under a single framework. Thus, the major contributions of this work are: • A robust framework has been considered regarding placement and operation of charging stations; • Multi-objective modelling of charging infrastructure planning has been considered using economic factors as well as the secure operation of power grid; • A novel chicken swarm optimization (CSO)-based heuristics has been used to solve the planning and operation of charging station problem; and • Three scenarios such as uncoordinated charging, coordinated charging, as well as bidirectional V2G are examined for the operation of charging stations.

PROBLEM FORMULATION
The charging station scheduling problem is mainly concerned with two activities such as location and operation of CS. The

Placement of charging stations
The CS placement is dealing with the locations and number of CSs to be placed at the respective locations. The placement of CS is divided into two stages. In Stage I, the placements of CSs are screened by using a probabilistic approach based on Bayesian network (BN). In Stage II, optimization is performed to compute the exact location and number of CSs to be placed.

Stage I
In Stage I, the candidate locations of the power distribution network where charging stations can be placed is screened by BN. The competence and potential of the BN in handling uncertainty and interaction among different events is effectively utilised in [25][26][27]. The BN used for computing the candidate locations is as shown in Figure 1. A summary of parent nodes and child nodes of the BN is presented in Table 2. VSF, charging demand, and failure probability are parent nodes. The probability of being a candidate location for placement of CS is dependent on VSF, charging demand, and failure probability. Hence, the probability of being a candidate location is the child node of VSF, charging demand, and failure probability. VSF measures the change in the bus voltage upon increasing the active power or loading and is computed by Algorithm 1. Charging demand of a particular point is computed by an improved random forest algorithm as depicted in Algorithm 2. The failure probability of the buses of the power distribution network can be found in the logbook of the substations [28][29][30]. The probability of being a candidate location for placement of CS is:

Stage II
In Stage II, optimization is performed to find the optimal cite, number and type of CSs to be placed in the distribution net-ALGORITHM 2 Pseudo-code for computing for charging demand [26,27] Input historical feature data and corresponding charge of EVs Step the input charging amount data work. The details of the objective function are given in Table 3. It is depicted as: Subject to: S min < S p ≤ S max and s min < s p ≤ s max , System Average Interruption Frequency Index (SAIFI), System Average Interruption Duration Index (SAIDI), Customer Average Interruption Duration Index (CAIDI) are the typical power distribution network reliability indices [28].

Charging strategy
Three charging strategies via uncoordinated charging, coordinated charging, as well as bidirectional vehicle to grid (V2G) are examined in this work. An overview of the aforementioned charging strategies is provided in Table 4.

METHODOLOGY
The placement and operation problem reported in Section 2 is solved by a novel CSO-based heuristics approach elaborated in [31]. CSO is a bio-stimulated procedure that mimics the Voltage stability index Voltage stability index (VSI) is regarded as a tool for evaluating the proximity of a given operating point to voltage instability

Reliability
The probability of a system under which it operates satisfactorily is termed as reliability  food searching phenomenon of chicken in a swarm. The swarm is divided into dominant roosters that lead the food searching process, hens following the roosters, and chicks following the mother hen. The algorithm also mimics the competition between hens in the quest for food. The main advantage of CSO is that it utilises the intelligence of chicken swarm in an effective manner and maintains a good balance between randomness and determinacy while finding the optimal solution. A multi-objective modelling considering economic and security issues jointly has made the problem more complex. However, the proposed model is more realistic and considers planning and operation of charging stations simultaneously. The CSO-based heuristics proposed in the work has the capacity to take into account the computational burden of the proposed model by sustaining a stability among exploration and manipulation. The detailed solution procedure is shown in Figure 2.

NUMERICAL ANALYSIS
The proposed formulation is validated on IEEE 33-bus and distribution network of Guwahati as depicted in Figures 3 and 4, respectively. The bus, line, outage data of the two test systems shown in Figures 3 and 4 can be found in [4,32,33]. As elaborated in Section 2, the first step of the placement problem  Figures 5 and 6. Also, it is observed that bus 14 and bus 19 are the weakest buses of test system 1 and test system 2, respectively. The charging demands of the test systems are computed by Algorithm 2 and depicted in Figures 7 and 8. The failure probability of test system 1 can be found in [4] and the failure probability of test system 2 is taken from the logbook of substations. The probabilities of being candidate locations for the placement of charging stations computed by BN are depicted in Figures 9 and 10. It can be observed that locating the    3, 6, 7, 8, 9, 10, 14} charging stations at various buses is better than concentrating on few buses which will create voltage deviation, reliability problem and increased losses. If any node has no space for charging the vehicles, the neighbouring bus can be selected. Accordingly, another favourable position for appropriating the charging station is making the charging station available to a bigger number of EVs handling in various routes. This will decrease the congestion of the particular routes in which charging stations are concentrated. The set of candidate locations having probability more than 0.6 is shown in Table 5. The optimization problem reported in Section 2 is solved by CSO. The input parameters of the optimization problem are same as in [31]. The general as well as algorithm-specific parameters of CSO are also taken from [31]. The optimization yielded six non-dominant solutions (NDSs) as reported in Tables 6 and 7. The selection of the best plan among the six alternatives depends upon user requirements. Further, three charging scenarios namely uncoordinated charging, coordinated charging and V2G are compared based on  VRP index proposed in [4]. A novel index named VRP index taking into account the voltage stability, reliability and power loss is also formulated. The novelty of VRP index lies in the fact that it has the capability of considering voltage stability, power loss, and reliability together under a common frame. The VRP indices for two test systems in case of the three aforementioned scenarios are shown in Figures 11 and 12, respectively. The advantages of coordinated charging and V2G

CONCLUSION
In recent years, both limited fossil fuels and environmental factors have forced automotive manufacturers to produce more efficient and greener vehicles. In this context, the production and use of hybrid and electric vehicles are increasing rapidly. With this rapid change in vehicle technology, user habits have raised important issues, such as the need to increase the numbers of charging stations, updating the electricity grid and increasing their capacities. It will, therefore, become inevitable that existing traditional car parks and fuel stations will have to be equipped with charging units and smart energy management algorithms will have to be developed to ensure their effective use. The formulations consider economic factors as well as secure operation of the power network. Further, the framework compares three charging strategies such as uncoordinated charging, coordinated charging and V2G. The solution methodology is based on a CSO driven heuristics. The simulation consequences indorse the efficacy of the anticipated framework and the advantages of coordinated charging and V2G over uncoordinated charging. The important issues such as quantitative analysis of charging strategies, operation of V2G enabled CSs, techno-economic analysis of smart charging as well as V2G will be addressed in future work.