Smart charging refers to an optimal or intelligent way of charging of electric vehicles (EVs) considering the condition of the electricity grid and charging needs of EV users. As elaborated in the last article on smart charging, managed/control-led charging would facilitate in meeting the escalating energy demands as well as maintaining the grid constraints. Smart charging can have a wide range of levels from the most basic on/off control to the advanced bidirectional V2X application. The key players in the smart charging ecosystem includes the grid operators/ system operators, EV users, and aggregators. EV aggregator is an entity between the system operator and EV user that helps to monitor, manage, and control the charging. Smart charging strategies defines a set of approach that help in optimising the vehicle charging process according to the distribution grid condition, renewable energy availability, EV user preferences, and the electric vehicle supply equipment (EVSE) site condition. The smart charging strategies discussed in this article applies to the unidirectional controlled charging (V1G) of EVs. The selection of best strategy ensures that each of the key stakeholders associated with the EV charging infrastructure gets benefit by enabling optimised charging process.
Classification of smart charging strategies
The smart charging strategies can be classified based on different attributes such as topology/architecture, location, ownership, methodology/approach, price structure, and objective(s).
Figure 1: Classification of smart charging strategies based on different attributes
Source: (Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), 2021)
The catogorisation of smart charging strategy is mostly based on the location of the EV charging station rather than the type of EV in particular. The key players in the ecosystem are directly involved in most of the strategies. The charging strategy is generally selected based on the computational and communicational infrastructure requirements.
- EV Smart charging strategies based on the control architecture:
Based on the control architecture of smart charging, EV charging strategies can be categorised into five sub-categories: centralised control, decentralised control, distributed control, hierarchical control, and local control. The network operator, aggregator, and EV owners are involved in the exchange of information or control signal according to the strategy. The control strategy can be achieved either by the EV users responding to the price signals, by the automated response of EVSE to the control signals generated by market and the grid situations, or by a combination of two taking care that due consideration is given to the needs and availability of EV users.
In centralised control strategy, the aggregator determines the pattern of EV charging considering the system operator constraints and charging energy demanded by the EV users. Further, the aggregator’s role in the strategy is to maintain the system stability while fulfilling the energy demand of the EVs (Nimalsiri, 2020). This control strategy takes care of congestion and faults in the system and finds an optimal solution for charging considering all the stakeholders. However, the controlling unit in the centralised control does not permit the plug-and-play mechanism, which might discourage the owners due to the lack of assurance on immediate starting of EV charging. Since the main charging decision is taken by the aggregator, this control strategy requires high computational power and communication infrastructure.
In decentralised control strategy, the EV users decide the charging pattern. The aggregator/ system operators try to influence this charging pattern of the EV users by providing varying energy prices, certain incentives, and potential revenues (Gan, 2018). This control architecture provides a plug-and-charge facility to the users, and it is relatively popular among EV customers. Since the decision-making unit is the EV user itself, the computational power required for this control strategy is relatively low. However, unlike the centralised control architecture, the decentralised charging approach does not guarantee the global optimum solution for the system.
Distributed control strategy is an enhanced edition of decentralised strategy. In contrast, here the aggregator communicates with each other to interpret an optimum operating point considering the system stability. It ensures plug and charge mechanism, which encourages participation of EV customers in smart charging.
In hierarchical control strategy, the architecture is divided into a central aggregator, subordinate layers of sub-aggregators, followed by EV user layer. The decision is taken at the central aggregator level and passed on to the sub-aggregators. Sub-aggregator optimally fixes the charging schedule for the EVs. It combines the benefit of centralised and decentralised strategies of directly controlling the charging and transferring the computational requirement for decision-making to the subordinate layer. Each layer of the architecture takes its own decision for achieving the desired objective without disturbing the other entities’ objectives.
In local control strategy, the EV users are accountable for maintaining the local system constraints and finalising the charging pattern. Local control only considers the local parameters, constraints, and pricing signals for making charging decisions (Kevin Mets et al., 2010). Since the decision signal is taken locally, the required communicational infrastructure and computational power is relatively low in this strategy. This control strategy is mainly applicable to private charging stations and especially home charging points.
Figure 2: Key smart charging control strategies
Source: (Nimalsiri, 2020), (Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), 2021)
- EV smart charging strategies based on the objectives:
The objective based strategies are applied with an aim to meet the desired objectives of the key stakeholders in the ecosystem. The system operator intends to maintain the stability of the system, whereas the EV owner needs the mobility requirement to be satisfied with minimum possible charging cost. The aggregator is the entity that ensures both the objectives of system operator, and the EV owner is satisfied.
Table 1: Achievable objectives from smart charging strategies
| Architecture/ Approach | Achievable objectives from the strategy |
| Centralised | · For load levelling
· For minimisation of energy loss · Controlling battery voltage constraint · Maximise charging rate · For optimal charging point allocation |
| De-centralised | · Minimising battery degradation cost
· For supporting fixed energy demand · For system cost and electricity consumption cost minimisation · Maximise renewable energy utilisation · Improving voltage profile and minimise line voltage · Minimise the system power loss |
| Distributed | · For tracking variety of charging power required
|
| Hierarchical | · For frequency regulation
· For maximising profit · Load curve valley filling |
| Local | · Load flattering
· Finding optimum charging station |
Source: (MuhammadAmjada, 2018)
- EV smart charging strategies based on optimisation algorithms:
This strategy categorisation is based on the computational algorithm used for optimising and managing of EV loads. Optimisation methods, data-driven method, artificial intelligence (AI) and machine learning (ML) based methods, fuzzy method, model predictive control method are some approaches for finding optimal EV schedules while maintaining network constraints.
Table 2: Key smart charging algorithms
|
Smart Charging Algorithm |
Linear Programming |
| Quadratic Programming | |
| Dynamic Programming | |
| Convex Optimisation | |
| Metaheuristic approach | |
| Mixed Integer Programming | |
| A* algorithm | |
| Integer Linear Programming |
Source: (Zhang, 2015), (Han, 2010)
Data-driven AI and ML-based solutions are another approach for scheduling and coordinating EV charging. In this approach, the models are built and trained to understand the behaviour of participating entities considering the other network constraints/data. Based on the available training sets, the AI/ML-based strategy is categorised into supervised and unsupervised learning. K-means clustering, Gaussian mixture model, Kernel density estimator are some of the methods used in literature for unsupervised learning. Linear regression model, decision tree, random forest, space vector machine are some supervised methods. The drawback of this methodology is the training data set applicable for one geographical region may not be applicable to other. The training data set needs to be updated for each region under analysis. Secondly, the availability of the standard labelled data required for the AI/ML based strategy is very limited.
- EV smart charging strategy using pricing mechanism:
Smart charging strategies can also be implemented by using certain pricing mechanism. The EV user charging behaviour could be influenced by changing the price of electricity. This varying price could be determined in advance (also called as static) or the prices could be set according to the current system operating condition (also called as dynamic). Some of the price-based mechanisms are real time price, time-of-use, critical peak price, and peak time rebate.
In real-time pricing mechanism, the electricity price is varied at each time stamp according to the network charging conditions. The electricity prices are updated based upon various parameters including the charging behaviour, available energy mix, allowable maximum power limits etc. Factors such as transformer burden and current feeder line loading also affects the electricity cost. In this mechanism, the optimisation is done to reduce the real time charging cost. The decentralised and distributed control strategies generally adopt the real time pricing mechanism as their control principle.
In time-of-use (TOU) tariff mechanism, a fixed price is allotted to each of the time slot. This mechanism performs smart charging without varying the charging rates. These fixed prices are issued on the actual day of operation, and the users have the flexibility to shift their use of appliances to those slots where the electricity prices are low. TOU indirectly facilitates increased renewable energy utilisation, reduced transformer loading and peak load, but does not ensure an optimal solution and grid stability. TOU tariff is used in centralised charging where the aggregator considers this tariff to optimise the charging to reach the desired objective. TOU helps in major cost minimisation compared to uncontrolled charging.
Critical peak pricing (CPP) is CPP works under the same TOU principle. The difference is that CPP is applied for a period of high demand. It is not decided on historical data, but rather forecasted data is used to apply and issue the prices quickly. The electricity price is very high in CPP compared to TOU, so it is more effective than TOU for peak load reduction (Newsham, 2010)
In the peak time rebate tariff structure, the utility provides a rebate to the customer to limit consumption within a predefined limit. Customer views it as a gain. However, shifting load to off-peak time is considered a loss. The economic effectiveness of the scheme is dependent on the predefined critical baseline load as it requires development of precise baseline load.
Figure 3: Electricity pricing mechanisms for smart charging strategies
Source: (Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), 2021)
Comparative analysis of different EV smart charging strategies:
The Table 3 below shows a comparative analysis of the major smart charging strategies. The comparison is done based on attributes including computational power, communication, role/need for smart charging algorithm, EVSE control features, aggregator involvement, nature of solution and the primary stakeholder.
Table 3: Comparative analysis of different EV smart charging strategies
| Attributes | Centralised | Decentralised | Distributed | Hierarchical | Local |
| Computation power | High | Medium | Moderately High | High | Low |
| Communication | High | Medium | Moderately High | Moderately High | Low |
| Role/need for smart charging algorithm | High | Medium | Medium | Moderately High | Low |
| EVSE control features | High | Medium | Medium | Moderately High | Low |
| Aggregator Involvement | High | Low | Medium | Moderately High | Low |
| Nature of solution | Global optimal solution | Sub-optimal solution | Sub-optimal solution | Optimal solution | Sub-optimal solution |
| Focused primary stakeholder | Grid operator | EV user | EV user | Grid operator/ EV user/ Both | EV user |
Source: (Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), 2021)
Detailed information on each of the smart charging strategies can be obtained from Part B – (Page 101) of A critical Review – Smart Charging Strategies and Technologies for Electric Vehicles.
Conclusion
Development of a reliable framework for selection of best suited smart charging strategy is crucial for coping with the increasing EV adoption. The comparative analysis of the strategies discussed shows that the entralized strategy facilitates direct control of global network constraints and thus provides a global optimal solution. Each strategy has its own pros and cons. The understanding of the objectives of key players, computational requirement, control feature provided is critical for the selection of strategies.
Irrespective of the exact nature and performance of each strategy, understanding the impact of the selected strategy on the battery and charging performance is a key requirement that may ultimately underpin consumer acceptance. The analysis of charging pattern under each strategy and their impact on the battery life is necessary.
This article is fifth in the series that elaborates the classification of smart charging strategies based on different attributes and a comparative analysis of the strategies used in EV smart charging ecosystem. The next article would discuss on the EV battery ecosystem and the status quo of traction batteries in e-mobility applications.
References
Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), 2021. A critical Review – Smart Charging Strategies and Technologies for Electric Vehicles, New Delhi: s.n.
Gan, L. T. U. L. S. 2., 2018. Optimal decentralized protocol for electric vehicle charging. IEEE transactions on Power systems, Volume 28, p. 940–951.
Han, S. H. S. S. K., 2010. Development of an Optimal Vehicle-to-Grid Aggregator for Frequency Regulation. IEEE Transactions on Smart Grid, pp. 65-72.
MuhammadAmjada, A., 2018. A review of EVs charging: From the perspective of energy optimization, optimization approaches, and charging techniques. Transportation Research Part D: Transport and Environment, Volume 62, pp. 386-417.
Newsham, G. B. B., 2010. The effect of utility time-varying pricing and load control strategies on residential summer peak electricity use. A review. Energy Policy, Large-scale wind power in electricity markets with regular papers, Volume 38, p. 3289–3296.
Nimalsiri, N. I. P. M. L. R. S., 2020. A Survey of Algorithms for Distributed Charging Control of Electric Vehicles in Smart Grid. IEEE Transactions on Intelligent Transportation Systems ( Volume: 21, Issue: 11, Nov. 2020, 21(11).
Zhang, W. G. W. H. M. J. J., 2015. Optimal day-time charging strategies for electric vehicles considering photovoltaic power system and distribution grid constraints.. Mathematical Problems in Engineering.
