Double-layered intelligent energy management for optimal integration of plug-in electric vehicles into distribution systems☆
Introduction
The global transport sector accounts for about a third of fossil fuel based energy consumption as well as a quarter of GHG emissions. Therefore, it is essential to electrify the transportation sector and the most promising way to realize this is to replace the conventional vehicles (using internal combustion engines) with PEVs [1]. Currently, the market share of PEVs is a small fraction of the global vehicle stock due to higher costs [2]. However, the supporting government policies are expected to make PEVs cost-competitive with conventional vehicles and hence large scale adoption of PEVs is envisioned which will lead to significant load addition into power grids.
The charging of large fleet of PEVs in an uncoordinated scenario may have adverse effects on various aspects of distribution network such as power loss, peak loading, voltage deviations, and the lifetime of network components [3]. As a result, many charging strategies have been developed in the literature for effectively managing the PEV load demand and minimizing its impact on the distribution system [4], [5], [6], [7]. In [4], a fast converging distributed demand response method has been proposed to minimize the peak load demand of the system by managing the PEV charging demand and shape the aggregated demand profile. The study [5] has developed smart charging strategies with an integrated G2V and V2G charging framework for optimal PEV integration within the existing distribution system infrastructure. A two-layered parking lot recharge scheduling system for PEVs is proposed in [6] which takes into consideration realistic vehicular mobility and parking patterns. The authors in [7] studied the impact of the developed PEV charging strategy on voltage security and power losses of the distribution network at different PEV penetration levels.
The formulation of charging strategies in [8], [9], [10] involves operation of PEVs in G2V and/or V2G mode. In study [8], smart charging strategies are designed to operate in a combined G2V and V2G charging approach. Similarly in [9], an intelligent two layer PEV charging algorithm was presented which was implemented using both G2V and V2G technologies. However, the study [10] proposed a bi-level optimization framework considering only G2V operating mode for charging management of PEV fleet.
Most of the studies present in the literature consider the exchange of only real power between the grid and PEVs [11], [12], [13], [14]. In [11], a charging strategy for fast charging stations was proposed which managed real power demand of PEVs by leveraging incentives from variations in electricity pricing and aimed to minimize the voltage deviations in the distribution network. In study [12], the real power demand of PEVs is managed by developing a balanced charging strategy which benefits PEV users and system operators simultaneously by saving charging costs and relieving load demand respectively. A double layer smart charging management algorithm for controlling real power demand of PEVs was developed in [13] to prevent overloading of transformers and to minimize the charging costs. The authors in [14] proposed a charging and discharging strategy for real power of PEVs which aimed at minimizing the grid operation costs and at the same time meeting the driving requirements of PEVs. It is important to highlight that the studies [4], [5], [6], [7], [8], [9], [10], similar to studies [11], [12], [13], [14], managed and controlled only the real power of PEVs.
However, PEVs can also be used to inject/absorb reactive power to/from the grid with the help of their on-board bidirectional battery chargers. In addition, frequent (dis)charge of PEV batteries for real power support affects the battery life, however, using PEV batteries for reactive power services do not affect the battery life [15].
Reactive power is an important ancillary service which can help in power loss reduction as well as voltage regulation [16]. A few studies in the literature have utilized PEVs for providing reactive power ancillary service [17], [18], [19], [20], [21], [22]. In [17], a decentralized scheduling of PEV parking and charging is formulated which aims to fulfil the grid reactive power requirement as well as minimize the PEV owners’ monetary cost. Based on the optimal scheduling for (dis)charge of PEV active power, a reactive power supply function is extracted in [18], [19] which is able to provide on-demand reactive power service with zero cost to PEV owners. In [20], PEVs’ active and reactive power flow are controlled to provide combined frequency and voltage regulation. However, the studies [17], [18], [19], [20] consider only the aggregator’s or PEV owners’ perspective which is to minimize the cost incurred for PEV (dis)charging. The perspective of DSO is not considered in [17], [18], [19], [20], which can utilize the PEV reactive power services for minimizing the power loss and voltage deviations resulting from non-PEV loads as well as PEV charging demands. Furthermore, it is assumed in [17], [18], [19], [20] that the reactive power injection/absorption requirements from PEVs are known and hence various important aspects, such as the network topology, power flow conditions, and the size and location of both PEV and non-PEV loads, are not considered.
In [21], the real power load demand of PEVs is managed to perform demand response and the reactive power injection capability of PEVs is utilized for voltage regulation. However, the operation of PEVs in V2G mode for real power support is not considered in [21]. A two-stage optimization method based on simultaneous active and reactive power management of PEVs is presented in [22] for minimizing the energy losses in a microgrid. The method for PEV charging coordination developed in [22] considers that the aggregators (present at each node of the distribution system) aim to flatten the load demand of respective nodes. However, the aggregators are more likely to minimize the costs incurred for PEV (dis)charging which is not considered in [22]. Moreover, the utility of reactive power services of PEVs for voltage regulation is also not demonstrated in [22].
Lastly, as aforementioned, operation of PEVs for reactive power compensation can aid in voltage regulation. Therefore, it is believed that the PEV penetration within acceptable voltage limits in a distribution system can be increased by utilizing the reactive power injection/absorption capability of PEVs’ on-board chargers. It is noted that an investigative study for the evaluation of acceptable PEV penetration level considering voltage constraints and utilizing reactive power services of PEVs has not been carried out in the literature so far.
In this paper, smart PEV charging is performed which allows utilization of PEVs for both real and reactive support to the distribution network. The main contributions of the paper are as follows:
- (a)
A DIEM approach, based on two layers of optimization framework, is proposed for optimal integration of PEVs into the distribution system. The first optimization layer is designed for real power management at the node level of the distribution system and represents the nodal aggregator’s perspective, whereas the second optimization layer is designed for reactive power management at the network level of distribution system and represents the DSO’s perspective. The objective function for the first layer of optimization is to minimize the total daily cost incurred for PEV (dis)charging by the nodal aggregator. However, the objective function for the second layer of optimization is to minimize the network power loss.
- (b)
An investigative study corresponding to the proposed DIEM approach is conducted for the evaluation of acceptable PEV penetration which can be integrated into the distribution system within the permissible voltage levels.
To the best of our knowledge, this is the first time the capability of PEVs as reactive power compensators is utilized to demonstrate the possibility of accommodating higher PEV penetration within acceptable voltage levels. Furthermore, this is also the first time PEV charging coordination is considered simultaneously from perspectives of nodal aggregators and DSO for managing real and reactive power respectively. A GA with heuristic initialization and a DE algorithm are employed as the optimization methodologies for the first and second layer of optimization, respectively. Lastly, the performance of the proposed DIEM approach is assessed against a SIEM approach which involves only first optimization layer, i.e. real power management of PEVs.
Section snippets
Problem overview
In this paper, the focus of study is to coordinate the (dis)charging of real and reactive power to/from PEVs. The PEVs are considered to be located in industrial, commercial and residential areas of a distribution system. It is assumed that a total of 1000 cars can be parked in the distribution system. All PEVs belonging to one node are contracted to a nodal aggregator called RPM present on the same node. All the RPMs are contracted to a DSO called RQM which is present at the network level. A
Evaluation of PEVs’ energy demand
The evaluation of the energy demand of PEVs is essential for enabling the formulation of the proposed DIEM approach which deals with the management of active and reactive power flow to\from PEVs. The energy requirement of a PEV is dependent on the initial state of charge (SOC) while plugging-in and the desired SOC which is required while plugging-out. PEVs are assumed to be present in industrial, commercial and residential areas of the distribution system. The arrival and departure times for
Problem formulation
In this section, the problem formulation corresponding to the proposed DIEM approach which includes a double-layered optimization framework is presented. The nodes in the distribution system are denoted by a vector . The time horizon is defined by a vector having 24 time slots. The number of PEVs arriving and plugging-in for charging at hour ‘’ and node ‘’ in the distribution system are represented by a vector . The parking duration of th PEV is
Implementation of the proposed DIEM approach
The proposed DIEM approach with double-layered optimization framework is implemented using GA with heuristic initialization (first optimization layer) and DE (second optimization layer) [23]. It is noted that the execution of first and second layer of optimization takes place sequentially.
Simulation results and case studies
In this section, the implementation of the proposed DIEM approach is presented for managing the flow of real and reactive power to/from PEVs which are parked in a 12.66 kV 33-bus distribution system, shown in Fig. 3. The system and load parameters of the 33-bus system are taken from [24]. The backward-forward sweep method is utilized in this study to perform the load flow analysis [25]. Fig. 4 shows the typical non-PEV load demand profiles for industrial, commercial and residential nodes of the
Discussion
The proposed DIEM approach utilizes PEVs for both real and reactive support of the grid and therefore will find a lot of applications in real world scenarios. With the widespread adoption of PEVs globally, there are two main challenges in front of the grid operators. The first challenge is to integrate more PEVs into the existing power grid infrastructure with minimal investment on reinforcements. The second challenge is to manage the charging load of growing number of PEVs without affecting
Conclusion
This paper proposes a DIEM approach which is based on two layers of optimization for managing the real and reactive power flow to/from PEVs at the nodal and system level of a distribution system, respectively. The objective function for the real power management in the first optimization layer is designed from RPMs’ perspective (nodal level) and is aimed at minimizing the total daily cost incurred for PEV (dis)charging. However, the objective function for reactive power management in the second
References (29)
The use of parking lots to solar-charge electric vehic-les
Renew Sustain Energy Rev
(2016)- et al.
A bi-level optimisation framework for electric vehicle fleet charging management
Appl Energy
(2016) A charging pricing strategy of electric vehicle fast charging stations for the voltage control of electricity distribution networks
Appl Energy
(2018)- et al.
Balanced charging strategies for electric vehicles on power systems
Appl Energy
(2017) - et al.
Adouble-layer smart charging strategy of electric vehicles taking routing and charge scheduling into account
Appl Energy
(2016) - et al.
An improved PSO-based charging strategy of electric vehicles in electrical distribution grid
Appl Energy
(2014) - International Energy Agency, Global EV Outlook 2016, May 2016 [Online]. Available:...
- et al.
A comprehensive study of the impacts of PHEVs on residential distribution networks
IEEE Trans Sustain Energy
(2014) Demand response for residential electric vehicles with random usage patterns in smart grids
IEEE Trans Sustain Energy
(2015)- Mehta R, et al. Smart charging strategies for optimal integration of plug-in electric vehicles within existing...
A smart parking lot management system for scheduling the recharging of electric vehicles
IEEE Trans Smart Grid
Probabilistic analysis of plug-in electric vehicles impact on electrical grid through homes and parking lots
IEEE Trans Sustain Energy
Integration of plug-in hybrid electric vehicles into residential distribution grid based on two-layer intelligent op-timization
Smart Grid IEEE Trans
Cited by (62)
Integration of electric vehicles into transmission grids: A case study on generation adequacy in Europe in 2040
2022, Applied EnergyCitation Excerpt :This comes at a cost of limiting the area of study (to a country or a smaller region) and the length of the period studied. Most studies on the interaction between EVs and distribution grids rely on this modelling approach, especially when scoped to local scale [13,28,47,48], but very few have studied the impact of EVs on transmission grid flows [49]. In this work, the scope of study has been set on EV flexibility from the second perspective (‘unit commitment and dispatch’), as it has been identified as one of the main research gaps for mid-term studies [50,51].
- ☆
This work was supported by Ministry of Education, Singapore under the grant R-263-000-B14-279.