Dynamic Wireless Charging for Micromobility Under Electromagnetic Field Exposure Regulations: A Review of Smart Grid Control and Charging Optimisation Approaches

Abstract

Dynamic inductive power transfer (DIPT) can enable dynamic wireless charging for urban micromobility, but deployment is constrained by electromagnetic field (EMF) exposure compliance and by lateral and angular misalignment typical of two-wheeled vehicles. This review consolidates the state of the art and links these constraints to smart grid control and charging optimisation. It frames dynamic charging lanes as corridor infrastructure that behaves as a distributed electrical load whose demand depends on traffic and availability, with segmentation control as a key lever for controllability. It then synthesises practical system architectures that combine power electronics, segmented transmitters, sensing, communication, and supervisory control, because these interfaces determine which degrees of freedom are available to shape demand in space and time. The review also summarises coupler, shielding, and compensation choices that jointly determine efficiency, misalignment robustness, and EMF leakage. Finally, it surveys scheduling methods that incorporate network limits, output from distributed energy resources, and uncertainty through rolling horizon, robust, and risk-constrained formulations. The synthesis supports deployment aligned with renewable integration and sustainable urban mobility, and it highlights open needs in forecasting robustness, scalable optimisation, and secure interoperability.

1. Introduction

Electric vehicles (EVs) for micromobility, such as electric scooters, bicycles, and mopeds, are becoming increasingly widespread in urban environments as cities seek to decarbonise short-distance trips and reduce local air pollution. However, the large-scale adoption of these vehicles hinges not only on the vehicle design but also on the availability of convenient, reliable, space-efficient charging solutions that can be integrated into dense urban fabrics without compromising pedestrian space or streetscape quality.

In this context, wireless power transfer (WPT) has emerged as a particularly attractive alternative to conventional plug-in charging. By transferring energy across an air gap through inductive or resonant inductive coupling, WPT eliminates exposed conductive connectors, reduces mechanical wear, and enables chargers to be embedded in parking bays, pavements, or the roadway itself. These features are especially appealing for micromobility applications, where vehicles are frequently parked in public space and user convenience is a critical factor for adoption, as well as safety for pedestrians and users.

WPT systems for EVs can be deployed in several modes, ranging from stationary charging pads for parked vehicles to dynamic configurations in which power is transferred while the vehicle is moving. The latter, commonly termed dynamic wireless power transfer (DWPT), seeks to mitigate range anxiety, reduce the size and cost of onboard batteries, and distribute energy supply along the route rather than concentrating it at a limited number of plug-in charging stations. In low-speed urban corridors or dedicated lanes, DWPT can in principle provide quasi-continuous energy replenishment to light electric vehicles, public transport fleets, and micromobility systems, thereby supporting wider adoption of electric mobility in city centres.

Achieving energy sustainability requires tighter coordination between electric energy generation and consumption, particularly as renewable generation increases and brings variability to grid operation. In this context, demand side flexibility and controllable loads become relevant tools to reduce mismatches between supply and demand and to support reliable grid operation. Dynamic wireless charging can act as enabling infrastructure that facilitates a more harmonious integration of electrified micromobility with local distribution networks by spreading charging demand along the route. This infrastructure-level perspective also provides a basis for future scheduling and charging optimisation strategies in smart-grid contexts, even when the present work focuses on coupler-level design.

The direction of further research in this area is expected to support the development of sustainable cities and the seamless integration of local zero-emission transportation infrastructure [1,2]. At the same time, the proliferation of WPT and DWPT solutions has brought issues of interoperability, electromagnetic compatibility (EMC), electromagnetic field (EMF) exposure, safety, and efficiency to the foreground. These concerns have motivated an intensification of national and international standardisation efforts. A notable example is the SAE J2954 standard for light-duty vehicles, which defines power classes, alignment procedures and acceptable criteria for interoperability, EMF emissions and EMC, with the aim of enabling the safe and widespread deployment of interoperable wireless charging systems [3]. In parallel, the SAE J2954/3 guideline, which specifically targets DWPT for light-duty vehicles, is being developed as an extension to SAE J2954, providing application-orientated specifications for DIPT systems, but is still a work in progress at the time of this study [4].

Despite substantial progress at the system and standardisation levels, the widespread deployment of DWPT, and more specifically, dynamic inductive power transfer (DIPT) systems, still requires further research to overcome several technical challenges. Key among these are the design of coil geometries that maintain high power-transfer efficiency under lateral, longitudinal, and angular misalignments, the control of EMF leakage in nearby air regions subject to stringent exposure limits, and the achievement of robust performance along realistic vehicle trajectories and operating conditions. These challenges are particularly acute for two-wheeled micromobility vehicles, whose slender geometry and motion patterns naturally lead to larger relative misalignments and more constrained packaging space than conventional passenger cars.

Accordingly, this paper reviews DIPT technologies for electric micromobility and brings together the practical constraints that govern their design and deployment. It synthesises how robustness to lateral and vertical misalignment, stability under coupling and load variation, and compliance with electromagnetic field exposure limits jointly shape the feasible design space. Building on this synthesis, the paper frames DIPT as an integrated infrastructure and control challenge, linking hardware and protection choices to corridor operation, sensing, and power management. This perspective motivates the requirements and research directions needed to enable scheduling and charging optimisation in smart grid settings, where dynamic charging demand must be coordinated with network limits and local energy resources.

Figure 1 summarises the structure of this review and highlights how the main themes discussed in the subsequent sections connect from infrastructure and coupler design to EMF compliance, control, and urban deployment.

2. Dynamic Wireless Charging as Enabling Infrastructure for Urban Energy Systems

A DWPT system supplies electrical energy to an electric vehicle while it is moving by transferring power across an air gap from transmitting equipment embedded in the road to a receiver on the vehicle [5]. This shifts charging from a small number of stationary points to a distributed, in-use service that can be planned and operated as part of an urban energy system [6].

From a mobility perspective, the enabling value of DWPT is closely linked to service continuity: charging during motion reduces the need for long stops and can lower the battery capacity required to deliver the same daily service. This is particularly relevant for urban fleets with regular routes and predictable stopping patterns, where infrastructure placement can be tied to operational timetables and vehicle energy use [6].

Seen through the lens of urban energy systems, DWPT behaves as an electrical load distributed along a corridor, whose demand depends on traffic volume, speed, and infrastructure availability, rather than only on plug-in behaviour. This dependence creates a direct coupling between transport operations and the power system: where and when vehicles move becomes a key driver of where and when power is requested along the network [5].

Control of the transmitting infrastructure is central to making this distributed demand manageable. Charging lanes with segmentation allow sections of the transmitter to be energised selectively, which can improve efficiency compared with energising long tracks continuously, especially when no vehicle is present. However, segmentation introduces switching and power fluctuation challenges that require robust control in real time [7].

As DIPT expands charging onto the roadway, it also broadens the set of options for sourcing and scheduling energy. In a life cycle and deployment optimisation study, roadside solar panels and storage are explicitly modelled as electric energy sources for charging moving vehicles, and are reported to be important for reducing life cycle energy use and greenhouse gas emissions, albeit with higher infrastructure cost [5]. Studies focused on grid stability emphasise that dynamic charging can affect operation at distribution level and should be analysed with time-varying load profiles rather than as a static load [8].

The sustainability case for DIPT therefore depends on whole-system trade-offs across vehicles, infrastructure, and electric energy supply. Life cycle assessment of an e-road equipped with DIPT shows that manufacturing the wireless power transfer components can dominate climate-change impacts despite their small contribution to overall mass, highlighting the importance of material and manufacturing choices alongside operational benefits [9]. In parallel, infrastructure has been framed as a combined spatial and temporal optimisation problem, with results that quantify when limited coverage can yield measurable energy and emissions reductions and enable substantial battery downsizing in a defined case study [5].

Finally, DIPT infrastructure must be compatible with evolving interoperability and grid integration requirements. Recent open-access work on wireless charging impacts emphasises that rapid adoption of wireless charging at high power can pose technical risks for power quality and distribution network operation if deployment is unplanned [10]. Recent standardisation, such as SAE documentation for WPT, explicitly states a present focus on unidirectional power transfer from the grid to the vehicle, with bidirectional transfer considered a possible future extension [3].

To position DIPT within the wider technical landscape and to compare approaches across different distances and frequency ranges, the next section provides an overview of WPT technology families, followed by a classification that distinguishes radiative and coupling-based approaches before narrowing the discussion to the technologies most relevant for dynamic road charging.

Current Status of Commercial and Pre-Commercial Dynamic Wireless Power Charging Systems

DWPT has progressed beyond laboratory demonstrations, with a clear shift towards pilot and pre-commercial implementations and field trials over the last decade.

Table 1. Representative dynamic wireless charging implementations and demonstrations reported in the literature.

Organisation	Project	Year(s)	EV Type	EV Numbers	Track Length	Inverter Power	EV Power	References
ENRX/ CFX/ ASPIRE	State Road 516 electrified roadway track (U.S.A.)	2024–2026	Small EVs, large SUVs and semi-trucks)	Not disclosed	1.20 km	Not disclosed	200 kW	[11,12]
VINCI Autoroutes/ Electreon/ Gustave Eiffel University	“Charge as you Drive” pilot on the A10 motorway (France)	2023–2025	Heavy truck, bus and light-EVs	4 (pilot tests)	1.50 km	Not disclosed	200 kW	[13,14,15]
Stellantis	“Arena del Futuro” circuit in A35 (Italy)	2021–2022	General EVs	Not disclosed	1.05 km	Not disclosed	Not disclosed	[16]
Electreon	Visby public wireless electric road (Sweden)	2021–2024	E-truck and e-bus	Not disclosed	1.60 km	Not disclosed	Not disclosed	[17]
University of Minho	Laboratory prototype (Portugal)	2016–2020	E-bike	Multiple EVs possible	3.25 m	Not disclosed	250 W	[18,19]
Korea Railroad Research Institute (KRRI)	DIPT transfer demonstrator for high-speed train	2015	High-speed train demonstrator (Korea)	Not disclosed	128 m	1.00 MW	818 kW	[20]
Korea Advanced Institute of Science and Technology (KAIST)	OLEV public bus service (Korea)	2013–2016	E-buses	2 e-buses	3.46 km	Not disclosed	100 kW	[21,22]
INTIS	DIPT test platform (Germany)	2013	Light-duty EVs	Not disclosed	25 m	Not disclosed	200 kW	[23]
Bombardier Transportation	PRIMOVE (Germany)	2010	Tram	1 tram	800 m	Not disclosed	200 kW	[24,25]

summarises representative projects reported from 2010 to 2026, spanning trams, buses, light-duty passenger EVs, heavy trucks, and high-speed rail demonstrators. The dominant targets are high-utilisation fleets and corridor-based applications where routes and duty cycles are predictable, which supports controlled validation of embedded infrastructure and repeated operation under realistic conditions.

The data in Table 1 indicate two main scaling directions. First, several road pilot installations have reached kilometre-class electrified segments, including 1.05 km (Italy), 1.20 km (U.S.A.), 1.50 km (France), and 1.60 km (Sweden). Second, high-power operation has been demonstrated in multiple contexts, with EV-side power reported at 200 kW for several road field trials and a tram demonstrator, and 100 kW for an in-service bus application. A rail demonstrator reports 818 kW EV-side power with a 1.00 MW inverter rating over a 128 m track segment, which evidences that very high power transfer is technically feasible when infrastructure and vehicle integration constraints allow it. At the same time, Table 1 also highlights a persistent reporting gap, since inverter power, EV counts, and in several cases EV-side power are not disclosed. This limits direct comparison of power density, utilisation, and energy delivered per vehicle-kilometre across reported implementations.

Micromobility remains comparatively underrepresented in dynamic field implementations. Table 1 includes an e-bike entry, but it is a laboratory prototype with a 3.25 m track length and 250 W EV-side power, which is substantially smaller in both length and power than the kilometre-scale, 100–200 kW road and bus pilot installations summarised in the same table. This contrast suggests that, although DWPT maturity is increasing for heavy and light-duty road vehicles, there is still limited evidence of sustained, public-road dynamic charging installations tailored to micromobility. From an engineering and implementation perspective, e-bikes and e-scooters require different optimisation priorities that are not yet reflected in the reported field trials. These include lower transfer power with high tolerance to lateral misalignment, compact and lightweight onboard receivers, shallow and low-cost track embedding compatible with cycling infrastructure, and operational models that may favour short charging segments at constrained locations rather than continuous kilometre-scale electrified lanes.

3. Overview of Dynamic Wireless Power Transfer Technologies

Before focussing specifically on dynamic wireless power transfer, it is useful to place DWPT within the broader landscape of WPT methods and the terminology used in the literature.

3.1. Classification of Wireless Power Transfer Technologies

The WPT technology can be classified on the basis of transfer distance and frequency of operation. There are two main groups of technologies: radiative, for considerable distances, and coupling, for short distances (Figure 2). Coupling technologies refer to a short transmission distance, transmission being carried out by a magnetic field, IPT or by an electric field, capacitive power transfer (CPT). These types of technologies can be further categorised as non-resonant and resonant [26,27]. For radiative technologies, there are two main categories that differ in operating frequency: microwave power transfer (MPT) and optical power transfer (OPT) [28].

3.1.1. Inductive Power Transfer

IPT operates on the basis of the generation of an electromagnetic field by a primary coil which induces an electric current in a secondary coil [28,30]. The primary coil is powered by an inverter operating in the 79 to 90 kHz frequency band, with a standard frequency of 85 kHz defined by SAE J2954 and IEC 61980-3 standards, and by a compensation network, which powers the electric vehicle [3,26,31]. The inverter is fed from a DC bus, which, in turn, is fed by rectification of the single-phase or three-phase grid supply. The electromagnetic field (EMF), from which the high-frequency (HF) supply is derived, induces an alternating current in the secondary coil, which is located in the EV and is connected to an AC–DC converter and compensation network. It should be noted that the position and movement of the coils directly affect mutual coupling, affecting the efficiency of the power transfer. Figure 3 illustrates a simplified block diagram of an IPT system.

In resonant IPT, compensation networks are tuned to the operating frequency to improve efficiency and tolerance to air gaps, at the expense of detuning sensitivity [29].

Three categories of IPT systems can be distinguished: stationary, dynamic, and quasi-dynamic. Stationary charging systems, commonly encountered, mostly serve as charging facilities in private garages and public parking bays, subject to the payment of a usage and consumption charge. The IPT system is used to charge a battery on board the EV. When the battery is charged, the vehicle can be driven until the battery reaches a level that requires it to be recharged. Dynamic charging is a process that applies to vehicles in circulation, in which the transmission coils are embedded in the road in a linear configuration that powers the secondary coil in the vehicle. This type of system charges an EV battery normally on fast roads or highways. Quasi-dynamic systems are used to charge EVs when they are opportunely stopped at an intersection, a traffic light, or a bus stop. They are also based on systems with embedded coils positioned where vehicles stop momentarily [26,32].

3.1.2. Capacitive Power Transfer

CPT is a type of system based on an electric field coupling created between two pairs of plates that act as two capacitors. The plate pairs are subjected to an alternating HF voltage, and a displacement current is generated between them due to the created electric field. The generated current depends on the electric field, the dielectric constant and the area where the current is generated [33].

This type of system comprises an HF inverter, a compensation network in the transmitter and receiver, and an inverter in the receiver that feeds the EV battery (Figure 4). Compensation networks contribute to decreasing system impedance and increasing efficiency.

The cost of this system is lower than the cost of an IPT system, and its efficiency is also lower. It is highly tolerant of misalignment, especially for aluminium pads [28,30,34].

There are some disadvantages to this type of technology compared to the others, namely the limited transmission distance between the plates and electrical current leaks that can cause injuries to EV users. For heavy-duty EVs, it may not be a viable system due to the low power per transmissible plate area. This is due to the low capacity because it is a simple coupling with two plates. Some studies suggest increasing the operating frequency as a solution, but the efficiency naturally decreases. There are also issues related to the appearance of metallic obstacles in the coupling area that lead to a reduction in transferred power [30].

3.1.3. Microwave Power Transfer

MPT operates in the microwave or radio frequency band, with the magnetron powered by high-voltage direct current. The signal is transmitted by an antenna and received by a rectenna consisting of an antenna and a rectification stage. The direct current is then converted and feeds a load. Figure 5 illustrates a high-level diagram of this system.

It enables the transfer of power on the order of kilowatts over a few kilometres and is applicable to DWPT systems while maintaining compatibility with other communications systems. Nevertheless, it is less efficient than IPT systems and, due to the essentially one-dimensional nature of the power transfer, it does not permit power injection from the emitter via regenerative braking. Moreover, this type of energy transfer can be harmful to the health of humans and animals upon exposure [28].

3.1.4. Optical Power Transfer

OPT systems are based on the emission of a laser beam, with the receiver consisting of a photovoltaic panel. Power transfer occurs on the order of THz, and it enables the transfer of power on the order of kilowatts over a few kilometres and is applicable to DWPT systems, reduced in size compared to MPT systems. The system consists of a laser diode that is directed at the photovoltaic (PV) panel. The light received on the PV panel is converted into an electric current to power a load. Figure 6 represents a high-level diagram of an OPT system [29,35].

This type of system require alignment between the laser beam and the PV panel, is intolerant of obstacles, and its efficiency is limited by the PV panel, whose efficiency depends on temperature [28,29].

3.1.5. Comparative Analysis and Application to Dynamic Systems

IPT systems present high safety and efficiency along with the possibility of bidirectional transmission, presenting little tolerance to metallic objects, especially resonant IPT systems. Resonant IPT systems face issues with regard to thermal dissipation and magnetic interference. CPT power transfer systems feature lightweight couplers, tolerance to metallic obstacles, and the possibility of bidirectional transmission. MPT is applicable over longer distances and for high-power applications, but has a relatively high cost and does not allow bidirectional transmission.

The aim of this study is to analyse dynamic systems to which inductive technologies are applied. This type of technology offers simplified implementation, galvanic isolation, and high efficiency, along with the possibility of bidirectional power transfer, which can enable grid services such as vehicle-to-grid (V2G) and support smart charging and grid-integration applications.

Table 2. Comparison of coupling and radiative wireless power transfer technologies [28,29,36].

Technology	Coupling		Radiative
	Inductive	Capacitive	Microwave	Optical
Efficiency	90–95%	80–90%	40–50%	1–15%
Power range	100 kW	7 kW	250 W	500 W
Frequency	kHz-MHz	kHz-MHz	MHz-GHz	THz
Air gap	30 cm	30 cm	>1 km	>1 km
Bidirectional	Yes	Yes	No	No
Relative cost	Medium	Low	High	High

presents a comparison between the categories of technologies applicable to WPT. The scalability of WPT systems is dictated by component complexity. With capacitive coupling as the cost baseline (1.0×), inductive systems typically range from 1.5× to 2.5× the cost. Radiative technologies represent the highest investment tier, with microwave systems scaling from 3.0× to 5.0× and optical solutions exceeding 10× the baseline investment. IPT technology presents a number of engineering challenges, such as, the high cost of coupler manufacturing materials due to the reduction of skin-effect losses. The creation of eddy current losses in adjacent metallic materials, and the fluctuation of the transferred power due to misalignment between the transmitting and receiving coils [29].

4. System Architecture and Implementation Framework for Dynamic Wireless Charging

This section describes the system architecture that enables dynamic wireless charging to operate as part of an urban energy system. Beyond the magnetic coupler, a practical implementation combines power electronics, segmented transmitting infrastructure, sensing, communication, and supervisory control to deliver energy only when and where it is required. These functions support interoperability and safe operation, and they define the interfaces between roadway equipment, the vehicle, and the electricity supply. From a grid perspective, the architecture determines which degrees of freedom are available to shape charging demand in space and time, which is a prerequisite for charging optimisation and coordinated operation in smart grid settings. Recent standardisation efforts provide reference points for these interfaces and operating assumptions, including wireless charging interoperability and the management of charging infrastructure [3].

In particular, standards for charging communication and infrastructure management provide a basis for control and coordination layers that can be integrated with scheduling and optimisation strategies.

4.1. Coupler Architectures for Dynamic Charging

The centrepiece of a DIPT system consists of a set of coupled coils commonly referred to as pads. For ground transport infrastructures, the primary coil is road-embedded and the secondary coil is incorporated into the vehicle. The coils have auxiliary equipment that is fundamental to their operation: magnetic material to conduct the magnetic flux and EMF shielding. The coils are made up of low-resistance conductors and allow high inductance, contributing to an increase in the quality factor and efficiency of the system (Section 4.3) [26].

4.1.1. Inductor Materials

Coils are composed of conductive materials, ranging from litz wires, magnetoplated wires, magnetocoated wires, tubular conductors, rare-earth barium copper oxide wires (ReBCO), and copper-clad aluminium wires.

Litz wires consist of a multi-stranded wire in which the various copper conductors have their own insulation (Figure 7a). The outside of the wire is also insulated (Figure 7c). This type of wire is suitable for conducting high-frequency alternating current, and due to its insulated multi-strand design, reduces the skin effect. The skin effect is the increase in current density near the surface of the conductor, causing eddy current losses. Separating the wire into smaller insulated wires reduces the total surface area, resulting in a reduction in losses [37,38]. The SAE J2954 standard specifically refers to the use of 5 mm diameter litz wire, without specifying the material [3]. This type of conductor is considerably expensive, as it requires the insulation of several conductors and the need to twist the conductors together, which also increases production costs [37].

Magnetoplated wires are multistrand wires in which each conductor consists of a conductive copper core and a thin layer of material with magnetic properties. The layer is composed, from the inside out, of nickel, iron and a layer of insulation (Figure 7b). Iron acts as a magnetic conductor, while nickel is present to improve welding performance. The presence of iron increases the inductance of the coil and, therefore, the coupling factor and the efficiency of the system. This type of wire improves magnetic coupling and promotes a reduction in magnetic losses. However, they have a more complex and expensive manufacturing process involving electroplating [28,39].

Magnetocoated wires differ from magnetoplated wires because of their manufacturing process, using chemical deposition or spray-coating methods. Similarly to the previous type of wire, this improves the magnetic characteristics of the conductor, and the durability of the coating may be lower than that of the magnetoplated wire [28,39].

Tubular conductors, usually manufactured from copper, have a lower skin effect and a lower conductive area than conductors with the same cross-sectional area. This type of conductor is manufactured by extrusion, with improvements in weight reduction and the possibility of internal cooling (Figure 8a). These types of wire are less common for this type of application [40].

ReBCO wires are formed by superconductors resulting from a combination of copper oxide, barium, and rare earths that present a ribbon-shaped shape. The manufacturing process entails epitaxial deposition on a nickel-chromium-molybdenum alloy substrate, usually Hastelloy, which is a registered alloy of Haynes International. The copper, silver and ReBCO layers are deposited sequentially (Figure 8b). This type of wire can achieve a coil quality factor Q (defined in Equation (4)) up to 7.2 times higher than copper coils, which can improve transfer efficiency at cryogenic operating temperatures around 70 K. However, they require complex cooling systems to maintain the superconducting state needed for operation. They are mechanically fragile compared to the other wire types discussed and they also have a high cost [28,41,42].

Copper-clad aluminium wires are a type of dual metal conductor constructed of an aluminium core and a copper-clad cast cladding (Figure 8c). Manufactured by lamination and extrusion, they are lightweight, suitable for use at high frequencies, and have a lower cost than copper wires [43,44].

4.1.2. Classification of Magnetic Couplers

Magnetic coupling is a fundamental element in IPT systems and typically consists of a primary coil, a secondary coil, and a shield. The coils are used to transmit power, while the shield helps to improve the distribution of the flux. There are two main types of coil that are applicable to DIPT systems: power track systems and planar coupling-based systems or pad systems [28].

4.1.3. Power Track Systems

Power track systems make use of a track coil along the fixed path of an electric vehicle, which could be an automated guided vehicle (AGV), a railway machine, or an online electric vehicle (OLEV) [45,46]. This type of transmitter allows more than one vehicle to be powered simultaneously. A single HF inverter and a compensation network power the track coil, ensuring constant mutual inductance in relation to the receiving coil [28].

It features a straightforward coil design with minimal construction complexity [28]. However, this type of coil provides insufficient tolerance for misalignment and is mostly applicable to vehicles travelling on a closed circuit. It has high maintenance costs and high energy losses in power transmission along with significant EMF emissions along the track, when the EV is not covering it [47].

The track coil can be classified according to the design of the magnetic core, which is usually made of ferrite. Initially, types “I”, “S”, “U” and “W” were proposed for power transmission on tracks [28,48,49]. Figure 9 represents the four main types of geometry for power tracks.

These types of geometries can be compared according to their EMF leakage, maximum air gap, track width, efficiency, power transfer level, and tolerance to lateral misalignment (

Table 3. Comparison of DIPT power track systems magnetic cores [26,28,48,49].

Parameters	I-Type	S-Type	U-Type	W-Type
EMF leakage	Medium	Low	High	Low
Air gap size	Considerable	Considerable	Medium	Small
Track width	Small	Small	Considerable	Medium
Efficiency	High	Low	Low	High
Power Transfer Level	High	High	Low	Low
Tolerance to lateral misalignment	High	High	High	Low

). It should be noted that the information summarised is predominantly reported qualitatively in the available literature.

The different types of power track systems are designed for specific applications. In general, they have a high cost, mechanical fragility, considerable weight, high maintenance and sensitivity to misalignment, which directly affects the power transferred. They are not a viable alternative to urban transport, where movement is more dynamic and subject to misalignment [29]. In order to overcome these limitations and obtain greater flexibility of movement while maintaining the efficiency of the system, novel types of coils with pad geometry have been developed [28].

4.1.4. Magnetic Pad Couplers

Magnetic pad couplers can be classified according to the magnetic flux generated as non-polarised and polarised pads (Figure 10).

Non-polarised pads are defined as a single coil with orthogonal magnetic flux to the plane of the pads resulting from the central magnetic pole and the opposite magnetic pole around the outside of the coil (Figure 10a). Some examples are circular, square, and rectangular pads. Polarised pads generate both orthogonal and in-plane magnetic flux components relative to the pad plane (Figure 10b). This type of pad has opposite magnetic poles on the same transverse line and is typically realised using double-D or double-D quadrature geometries [50,51]. In this context, a polarised coupler is commonly defined as generating and coupling a flux pattern that flows dominantly along a single in-plane dimension of the pad (either its length or its width), rather than remaining directionally symmetric around the coil centre. Regarding the single-loop flux path illustrated in Figure 10b, this behaviour is obtained by arranging two adjacent coil loops on the same side (for example, a Double-D structure) and exciting them such that two opposite magnetic poles are formed across the pad width. Under this opposite-phase excitation, the magnetic flux is forced to link the two coil regions, which produces a dominant in-plane flux component between them and results in the single-loop flux trajectory shown in the figure, while the return flux closes through the surrounding air and magnetic materials [52].

The magnetic field created in a non-polarised coil is orthogonal to the plane of the pads, and the magnetic field created in a polarised coil is parallel to them. Thus, coupling will occur between a non-polarised and a polarised coil if both are centred [50,53].

Circular, square, and rectangular pads were among the earliest types of pads to be studied, and many variants of these designs have been found with optimised ferrite geometries to direct magnetic flux and increase efficiency while reducing EMF leakage. Circular pads are the most common and least complex to construct, allowing considerable power transfer (Figure 11a). They have limited tolerance for misalignment, which is reflected in a reduction in the power transferred, so their application to dynamic systems is more limited. High EMF leakage, requiring improved shielding. This type of coil is normally used as a transmitter for stationary systems. Square pads are suitable for DIPT, being interoperable with circular pads and moderately tolerant of misalignment. Regarding EMF leakage, shielding is necessary and has been shown to be effective (Figure 11b) [28,54].

Rectangular pads have moderate tolerance to misalignment and the shielding has some influence on the transmitted power. Significant EMF leakage, requiring optimised shielding. It is applicable to short transmission distances, can be used both as a transmitter and receiver, and is suitable for DIPT [28,55].

Double-D (DD) pads consist of two opposite adjacent coils, allowing energy transmission over an improved distance and moderate misalignment tolerance. It is usually only applicable as a transmitter coil and is not interoperable with non-polarised pads. This design features extremely low EMF leakage. However, DD pads have a coupling null point which limits power transfer and their tolerance to horizontal misalignment. To overcome this limitation, the Double-D quadrature (DDQ) pad was developed, which consists of placing a coil decoupled from the previous two, improving tolerance to horizontal misalignment [54].

Bipolar pads (BP) represent a simplified version of the DDQ, consisting of two overlapping rectangular coils. It offers similar improvements to the DDQ, but requires less complex fabrication and reduces the amount of material required. The mutual decoupling between the coils forming BP improves the distribution of the magnetic flux and tolerance to misalignment [29].

4.2. Power Supply and Segmented Infrastructure Architectures

DIPT systems deliver energy to electric vehicles while they are in motion, enabling a substantial reduction in the required capacity of the on-board battery and thereby lowering vehicle cost, size and weight. In such systems, grid frequency power is processed by a chain of converters comprising a rectifier, power–factor–correction stage, and HF inverter on the transmitter side, together with a rectifier and DC–DC converter on the vehicle, so that the road-embedded track is energised by a controlled HF AC current [26,56].

From the viewpoint of the ground infrastructure, two main classes of DIPT power supply architecture can be identified, depending on how the transmitter coils are arranged and energised [26,56]. In the long-track configuration Figure 12a, a single extended transmitting conductor, typically tens of metres in length, is fed from a central resonant network and HF inverter, allowing several vehicles to be charged simultaneously as they traverse the instrumented lane, but at the expense of higher reactive power and more pronounced spatial variation of the transferred power [26,56].

Alternatively, segmented coil architectures partition the roadway into multiple shorter transmitter pads whose dimensions are comparable to those of the vehicle-side receiver pad. At any instant only the segment(s) beneath the vehicle are energised, which improves utilisation of the installed infrastructure and mitigates leakage-field emissions [26]. When segmented coils are used, two main supply arrangements are commonly discussed, as illustrated in Figure 12b,c.

In the common DC-bus arrangement, a front–end rectifier establishes a DC backbone from which individual HF inverters or switching converters feed each segment, enabling independent activation and precise power control at the cost of multiple power-electronic stages and high-voltage DC cabling along the track [26].

In the common HF AC-bus arrangement, a single HF inverter supplies a bus that is routed along the track, while local switches and compensation networks connect the selected segment(s) to the bus. This reduces the extent of DC cabling and can simplify protection, but imposes tighter constraints on the design of the HF bus and on the impedance of the segments to maintain efficiency and EMC performance [26,56].

The choice between long-track and segment architectures and between common DC-bus and common HF AC-bus supplies represents a system level trade–off between infrastructure cost, converter complexity, controllability of power flow, and the ability to satisfy electromagnetic-field exposure limits along the roadway [26,56].

4.3. Efficiency Considerations at Infrastructure Level

DIPT systems can be decomposed in a sequence of energy-processing stages from the electrical source to the load (end-to-end). This sequence is composed of the inverter and respective clock, power amplifier, coupled coils, rectification and regulation stage. Figure 13 provides this cascade and the associated stage efficiencies.

The end-to-end efficiency may be expressed as the product of stage efficiencies (Equation (1)). This factorisation is useful for system-level energy accounting because it links the electrical energy drawn upstream (grid side) to the net energy delivered to the vehicle, and therefore provides a consistent basis for comparing operating strategies and charging control approaches.

$${\eta }_{ee}={\eta }_{inverter}{\eta }_{coupling}{\eta }_{rectifier}{\eta }_{regulator}$$

(1)

Each efficiency ${\eta }_i=P_{out,i}/P_{in,i}$ is defined at the electrical interfaces indicated in the diagram. Coupling, rectifier, and regulator efficiencies can be grouped as a transfer efficiency. This grouping is convenient for analysis and reporting, while the mathematical identity remains unchanged.

In the context of grid integration and charging optimisation, ${\eta }_{ee}$ determines how much upstream energy and peak power are required to meet a given in-vehicle energy demand, which directly affects scheduling decisions and infrastructure operation. The analysis that follows discusses ${\eta }_{coupling}$ in greater detail because it is the stage most sensitive to pad geometry, spacing and misalignment tolerances, shielding materials, and induced-loss mechanisms, which are central design variables for DIPT. Unless stated otherwise, frequency, compensation topologies, and load conditions are kept fixed to enable a consistent comparison across configurations and to support sensitivity analysis. Section 4.3 develops the formal definition of ${\eta }_{coupling}$.

Coupling Efficiency

An IPT system can be modelled by a circuit in which the primary coil $L_1$, powered by an alternating voltage source, induces a current in the secondary coil $L_2$. The resistance of the primary and secondary coils is given by $R_1$ and $R_2$, respectively. Figure 14a represents the model of the IPT system [57].

Considering that $M_{12}$ represents mutual inductance, the source and load voltages can be written as a function of the system impedances, according to the T-model and the simplified two-port network model in Figure 14b and Figure 14c, respectively. Equation (2) represents the voltages in the system.

$$\left[\begin{array}{c}{V}_1\\ V_2\end{array}\right]=\left[\begin{array}{cc}{R}_1+j\omega L_1& j\omega M_{12}\\ j\omega M_{12}& R_2+j\omega L_2\end{array}\right]\left[\begin{array}{c}{I}_1\\ I_2\end{array}\right]=\left[\begin{array}{cc}{Z}_{11}& Z_{12}\\ Z_{21}& Z_{22}\end{array}\right]\left[\begin{array}{c}{I}_1\\ I_2\end{array}\right]$$

(2)

The coupling between the primary and secondary coils is characterised by the coupling coefficient ${\kappa }_{12}$ given by the Equation (3) [57].

$${\kappa }_{12}={\displaystyle \frac{M_{12}}{\sqrt{L_1{L}_2}}}$$

(3)

In addition, the transmission efficiency is directly influenced by the ratio between the transmitted energy and the energy dissipated in the coil, given by the quality factor. The quality factor is a dimensionless value that depends on the inductive reactance and the resistance of the coil. Transmission efficiency is improved for a higher quality factor. Equation (4) represents the quality factor [59,60].

$$Q={\displaystyle \frac{X_L}{R_L}}={\displaystyle \frac{\omega L}{R_L}}$$

(4)

The efficiency of a coupling in an IPT system is given by Equation (5) [57].

$$\eta ={\displaystyle \frac{P_{out}}{P_{in}}}={\left|{\displaystyle \frac{Z_{12}}{Z_L+Z_{22}}}\right|}^2{\displaystyle \frac{R_L}{R_1+{\displaystyle \frac{{\omega }^2{M}_{12}^2(R_2+R_L)}{{(R_2+R_L)}^2+{(X_2+X_L)}^2}}}}$$

(5)

Therefore, the maximum efficiency for an IPT system with series-series compensation (SS) is given by Equation (6).

$${\eta }_{max}={\displaystyle \frac{{\kappa }_{12}^2{Q}_1{Q}_2}{{\left(1+\sqrt{1+{\kappa }_{12}^2{Q}_1{Q}_2}\right)}^2}}$$

(6)

Equations (3) and (6) summarise the classical two-coil analysis of an SS-compensated IPT link, in which transfer efficiency is governed by the coupling coefficient ${\kappa }_{12}$, the coil quality factors, and the load electrical characteristics [57]. In the context of DIPT, these expressions are mainly useful as theoretical background and as a reference point for interpreting trends, rather than as a complete description of on-road performance.

In practice, DIPT infrastructures are implemented with segmented and modular transmitting coils along the roadway, while the vehicle carries a receiving coil. Power transfer during motion therefore results from a sequence of couplings between the receiver and successive transmitter modules, and the effective coupling varies with vehicle position and misalignment. For this reason, recent DIPT studies report position-dependent coupling or mutual-inductance profiles to characterise how geometry, spacing, and misalignment influence the energy that can be delivered along a charging lane.

For configurations with repeated transmitting modules, it is common to describe the interaction between the receiver and each neighbouring transmitter element using coupling coefficients. Treating these couplings separately preserves physical interpretability and supports system-level discussions, because the spatial variation of coupling translates into time-varying available power and efficiency during vehicle motion. This link is particularly relevant when considering segmented infrastructure control and scheduling strategies, where the activation of transmitter segments and the allocation of power can be coordinated with traffic conditions and distribution-network constraints.

4.4. Compensation Networks

Compensation networks are applicable IPT and DIPT systems, with the aim of minimising magnetic flux leakage and reactive power in order to increase efficiency, transferred power, and system stability. The parasitic capacitance is not sufficient to maintain resonance between the two coils. Compensation networks are therefore fundamental to power transfer and are incorporated into the transmitter and receiver circuits.

Basic Compensation Networks

The basic compensation networks, which are implemented by adding a capacitor in series or parallel to the transmitter and receiver circuits. There are four basic compensation topologies to refer to depending on the position of the capacitor in the transmitter and receiver circuit: series-series (SS), series-parallel (SP), parallel-series (PS), and parallel-parallel (PP) [29]. Figure 15 represents the four basic compensation networks, where $M_{12}$ represents mutual inductance, $R_1$, $L_1$ and $R_2$, $L_2$ represent the resistance and inductance of the primary and secondary coils, respectively. $C_1$ and $C_2$ represent the compensation capacitors of the transmitter and receiver circuits, respectively.

Table 4. Parameters for determining the capacity of capacitors for basic compensation networks [28,29].

Topology	Primary Capacitance	Secondary Capacitance
SS	$C_1={\displaystyle \frac{1}{{\omega }^2{L}_1}}$	$C_2={\displaystyle \frac{1}{{\omega }^2{L}_2}}$
SP	$C_1={\displaystyle \frac{L_2}{\omega (L_1{L}_2-M_{12}^2)}}$	$C_2={\displaystyle \frac{1}{{\omega }^2{L}_2}}$
PS	$C_1={\displaystyle \frac{L_1+{\omega }^2{R}_L}{{\omega }^2({\omega }^2{M}_{12}^4+L_1^2+{\omega }^2{L}_1{R}_L)}}$	$C_2={\displaystyle \frac{1}{{\omega }^2{L}_2}}$
PP	$C_1={\displaystyle \frac{L_2^3(L_1{L}_2-M_{12}^2)}{{(L_1{L}_2-M_{12}^2)}^2{\omega }^2{L}_2^2+M_{12}^4{R}_L^2}}$	$C_2={\displaystyle \frac{1}{{\omega }^2{L}_2}}$

presents the relationship between the compensation capacitances and the design variables of the transmitter and receiver. In the SS topology, the compensation capacitance of the transmitter and receiver depends directly on the desired resonance frequency and the inductance of the primary coil, and is independent of the coupling coefficient and the load. Thus, the capacitance is moderately dependent on misalignment. In the SP topology, the capacitance of the transmitter depends on the resonance frequency, the inductances of the primary and secondary coils, and the mutual inductance. It is a moderately sensitive strategy to misalignment. The PS topology depends on the resonance frequency, the inductance of the primary coil, the mutual inductance, and the load resistance. Since it depends directly on the receiver system, it is very sensitive to misalignment. The PP topology depends on the variables mentioned for the PS topology and also on the inductance of the secondary coil, and is also very sensitive to misalignment.

Basic compensation topologies depend on variables that are not stable, which explains the reason why these systems underperform in the conditions for which they were designed.

4.5. Higher-Order Compensation Topologies

Simple compensation topologies cannot accommodate variation for all misalignment and frequency-shift situations, as they are designed for a specific ideal scenario. The application of compensation topologies with multiple elements in series-parallel configurations allows the power transfer system to function satisfactorily for a greater number of scenarios.

Compensation networks are introduced to meet multiple design criteria simultaneously. In practice, single-sided compensation can offer limited degrees of freedom and is therefore often replaced by higher-order compensation networks when several criteria must be satisfied at once [61].

For DIPT, variations in coupling and effective load are expected along the lane due to position changes and misalignment. Review evidence indicates that multi-resonant networks (including LCC–LCC and LCL–LCL families) are used to extend the operation efficiency over wider ranges of coupling and loading, at the cost of added components and tuning complexity [28].

High-order compensation topologies are represented in Figure 16. The topologies illustrated can be classified in different families, such as, SP–S variants, S–SP variants, and multi-resonant LCL and LCC families [28].

The SP–S topology is described as combining series resonance compensation with series-parallel resonance compensation (Figure 16a). It offers improved misalignment tolerance and output characteristics that can be less dependent on load in the reported designs [62,63]. A reported limitation is that reactive power can increase under misalignment, which can reduce transfer efficiency and increase current stress [62,64]. Additionally, when the load-side network is parallel-compensated, the equivalent impedance seen at the electrical interface generally contains both resistive and reactive components except at a specific load condition, which must be accounted for in primary-side impedance and power-factor considerations [65].

The S–SP topology is reported to enable zero-phase-angle (ZPA) input impedance that is largely independent of load and coupling, supporting high efficiency and near-constant gain at the tuned resonant frequency, including in operating ranges relevant to higher-power applications (Figure 16b). Its main drawback is the higher tuning and design complexity [66,67].

The P–PS topology seeks to improve the misalignment performance of the PS topology (Figure 16c). A key trade-off highlighted for this modified network is the higher tuning and parameter-calculation complexity, with added elements also linked to increased system complexity, cost and potentially reduced reliability [68].

The LCL family (Figure 16d,f,g) is reported to behave as a current-source type network and to provide improved harmonic filtering, with ZPA operation achievable in the tuned designs. Within this family, LCL–LCL is reported to provide a constant-current output tendency, whereas LCL–S is reported to provide a constant-voltage output tendency. In both cases ZPA is reported as achievable. A reported concern for LCL–S is the possibility of an effective short circuit leading to large secondary currents, which is undesirable in practice. For LCL–P, the primary compensation is reported to be reflected as a current source suitable for interfacing a boost inductor in the cited implementations. Across the LCL family, the added inductors and components increase circuit complexity and can introduce additional losses at higher power, relative to simpler networks [28,69,70].

The LCC topology (Figure 16e,h) is reported to reduce inverter current stress and to compensate reactive power on the secondary side, with performance reported as less dependent on coupling and load in the reviewed designs [71,72,73,74]. LCC networks can offer high efficiency and are described as having lower cost and smaller size than some LCL implementations, while still supporting load-independent constant current or voltage output behaviour in the reported designs [70,72,74,75,76]. As with other higher-order networks, the main drawbacks are increased component count and tuning complexity [72,74,77].

From the perspective of smart grid operation, the choice of compensation topology is a practical determinant of how a wireless charger interacts with its power electronics and, indirectly, with the upstream supply. In high power WPT systems, the uncompensated coupler presents a strong reactive component. Compensation networks are designed to neutralise this reactance and to achieve a ZPA input condition, which is described as central to efficient operation. The input phase angle affects the apparent power demand placed on the supply and switching stage, so designs that maintain a near resistive input can reduce the stress on the inverter and simplify regulation. In addition, resonant compensation can be configured to exhibit load independent constant current or constant voltage characteristics at the operating point, which is discussed as beneficial for controllability in practical implementations [61]. This system level relevance is reinforced in bidirectional wireless power transfer for EVs, where vehicle to grid operation is reviewed as requiring coordinated control together with suitable compensation networks, and where compensation selection is described as a critical design task that influences overall performance [74].

These considerations become more important when wireless charging is extended from stationary points to dynamic operation, where power transfer occurs while the vehicle is in motion [78]. In this setting, predictable electrical behaviour at the converter interface supports higher level charging optimisation because it reduces uncertainty in power exchange under the coupling variations that arise during travel. Reference [79] proposes an integrated framework that uses traffic and energy demand information to site charging tracks and then applies dynamic scheduling for vehicle-to-grid operation, with the explicit aim of enabling daytime ancillary services that help stabilise the smart grid. Reference [80] formulates coordinated planning of fast charging stations and dynamic wireless charging systems by modelling them as links between the power network and the traffic network. It explicitly treats dynamic wireless charging as a means of charging vehicles during transportation within a coupled infrastructure planning problem. In conclusion, compensation and control choices, which govern efficiency and stability of the electrical interface, are enabling factors for reliable integration of dynamic wireless charging into smart grid oriented planning and coordinated charging strategies [61,74,79,80].

5. Electromagnetic Field Exposure Regulations as Design Constraints

WPT under coupling and load variation also shape the coil currents that generate the stray electromagnetic field in the surroundings [81,82,83]. EMF compliance is commonly discussed against international guidelines that provide internal basic restrictions and external reference levels [84,85]. Because adopted limits and compliance pathways can differ across jurisdictions and standards, the relevant guideline set and averaging metric should be stated explicitly before presenting numerical thresholds [84,85]. In the Section 5.1, standards and procedures to analyse exposure requirements are reviewed.

5.1. Exposure Limits and Compliance Assessment Methodology

WPT and DWPT systems have associated EMF leakages as a result of the inherent energy transfer method [86]. The impact of these emissions should be minimised through construction solutions or shielding and should always be accurately quantified according to the applicable guidelines. EMF emissions have significant biological impacts, especially on human health. This type of emission is categorised as non-ionising radiation, forming part of the spectrum that does not have enough energy to produce ionisation in matter. Exposure to EMF emissions can be categorised as direct, when the fields interact with the body, or indirect, when the body interacts with conductive materials that have a different electrical potential relative to the body. The International Commission on Non-Ionizing Radiation Protection (ICNIRP) 2010 guidelines state that, for frequencies above 100 kHz, there is an increase in the temperature of biological tissues exposed to EMF emissions [84]. It also mentions a feeling of “warmth” when there is exposure to 10 kHz EMF emission, along with effects in the nervous system [84,85]. Exposure to non-ionising radiation has effects on fertility, brain function, human behaviour, and the functioning of the cardiovascular system [87].

There are numerous international standards, guidelines, and recommendations that recommend limits on exposure to non-ionising radiation and subsequent absorption by the human and animal body. The absorption of non-ionising radiation by biological tissues can be accounted for by the Specific Absorption Rate (SAR), which is defined as the time derivative of the incremental energy consumption by heat $dW$, absorbed or dissipated by an incremental mass $dm$. The mass of biological tissue can be given as a function of its density $\rho$, contained in a given volume $dV$ (Equation (7)). SAR represents the energy absorbed per unit mass, the unit of which can be expressed as $W/kg$ [85].

$$SAR={\displaystyle \frac{d}{dt}}·\left({\displaystyle \frac{dW}{dm}}\right)={\displaystyle \frac{d}{dt}}·\left({\displaystyle \frac{dW}{\rho ·dV}}\right)$$

(7)

Biological tissues have substantial dielectric losses and unitary magnetic permeability and do not significantly interfere with magnetic fields. By deriving Equation (7), it is possible to obtain a simplified expression for SAR (Equation (8)).

$$SAR={\displaystyle \frac{\sigma ·{\left|E\right|}^2}{\rho }}$$

(8)

The conductivity and density of biological tissues are represented by $\sigma$ and $\rho$, respectively. E represents the root mean square (rms) electric field in the interior of the biological tissue [85].

The increase in temperature in biological tissues is directly correlated with SAR and is given by Equation (9), where c is the specific heat capacity of the tissue, T is the temperature and t is the time.

$$SAR=c{\displaystyle \frac{dT}{dt}}$$

(9)

To address technical challenges related to EMF emissions from WPT systems, some regulatory entities have established values to limit the effects on human health. ICNIRP 2010 guidelines set limits, in a range from 1 Hz to 100 MHz, for time-varying magnetic and electric fields. They refer to general public exposure limits and occupational exposure limits. General public exposure limits refer to the exposure of individuals of varying ages and health status who are unaware of their exposure to EMF. This type of condition requires the adoption of stricter limits compared to the occupational exposure of workers. Occupational exposure limits refer to adults working in a regular workplace, working under known and controlled exposure conditions. Generally, the regulatory limits for this type of exposure are higher than those for the general public.

Considering that IPT systems operate in the 3 kHz to 10 MHz band, the ICNIRP 2010 guidelines establish maximum limits for the root mean square electric field inside the biological tissues of the head and body, for general public and occupational exposure.

Table 5. Basic restrictions for human exposure of all tissues of head and body to time-varying electric and magnetic fields established in the ICNIRP 2010 guidelines [84].

Exposure Characteristic	Internal Electric Field E (V/m)
General public exposure	$1.35\times {10}^{-4}·f$
Occupational exposure	$2.7\times {10}^{-4}·f$

represents these maximum limits (root mean square), where f represents the field frequency [84].

Considering that the WPT operating frequency range covered by the ICNIRP 2010 guidelines is between 3 kHz and 10 MHz, there are exposure limits defined according to the electrical and magnetic properties of the EMF field. Limits for these properties are established for the general public and types of occupational exposure.

Table 6. General public exposure limits established in the ICNIRP 2010 guidelines for frequencies between 3 kHz to 10 MHz [84].

Reference Levels	General Public Limits
Electric field strength E (V/m)	83
Magnetic field strength H (A/m)	21
Magnetic flux density B (μT)	27

represents the electric field strength, magnetic field strength, and magnetic flux density limits established by the ICNIRP 2010 guidelines, which must be respected when designing WPT systems [84].

IEEE C95.1-2019 standard establishes exposure limits for the general public, setting regulatory root mean square levels for the strength of the electric field, the strength of the magnetic field and the density of the magnetic flux for specific areas of the human body (

Table 7. General public exposure limits established in the IEEE C95.1-2005 standard for frequencies between 3.35 kHz to 5 MHz and between 3 kHz to 100 kHz [88].

Frequency Range	3.35 kHz to 5 MHz		3 kHz to 100 kHz
Body Parts	Head and Torso	Limbs	Whole Body
Electric field strength E (V/m)			614
Magnetic field strength H (A/m)	163	900
Magnetic flux density B (μT)	205	1130

) [88].

The indicated values are higher than ICNIRP’s. Furthermore, this IEEE regulation is not binding on the European Union, but is valid for the United States of America and other countries that follow IEEE standards. It should be emphasised that Directive 2013/35/EU transcribes the limits indicated in the ICNIRP 2010 guidelines, making it binding to the European Union [89]. Directive 2013/35/EU was transposed into Portuguese law by Law 64/2017 of 7 August [90].

However, the guidelines and directives covered until now focus on general public and occupational limits and are not directly related to the occupation of EVs, pedestrians, or cyclists in the surroundings or interference with implanted medical devices. They must, however, be observed during the design and production process of EVs with WPT. The SAE J2954, IEC 61980 and ISO 19363 standards stand out in this regard, as they present guidelines focused on the automotive sector, regulating emissions and exposure to EMF for stationary WPT systems.

Despite the fact that the SAE J2954 standard is not binding in the European Union, it is recognised internationally as a technical reference and follows the ICNIRP 2010 guidelines and the IEEE C95.1-2019 standard. The SAE J2954 standard establishes an industry-wide specification that establishes criteria for interoperability, electromagnetic compatibility, EMF exposure and emissions, minimum performance, safety and testing for WPT systems. The standard outlines regulations on EMF emission and exposure for EVs with input power up to 11.1kVA. It also establishes regulatory limits for EMF exposure for implanted medical devices (IMD) and other cardiac implantable electronic devices (CIED) [3].

Three regions are defined in relation to EV, according to the SAE J2954 standard, where different regulatory limits are applied for exposure to EMF (Figure 17).

Region 1 is the entire area underneath the vehicle, including the wireless power assemblies and the surroundings, not extending beyond the structure of the lower body. Region 2 is the peripheral region of the vehicle, with the border with region 1 extending vertically down the side of the vehicle to the ground surface. Region 3 is the interior of the vehicle.

Region 1 requires increased safety measures to prevent human exposure to values higher than those mentioned in

Table 8. EMF exposure limits established in the SAE J2945 standard for frequencies between 79 kHz to 90 kHz [3].

Region	Reference Levels	Magnetic Flux Density B (μT)
1	IEEE C95.1-2019	Head and torso: 205 Limbs: 1130
2,3	ICNIRP 2010	27

. Prevention of exposure requires active or passive measures to restrict access to the energy transfer zone of the WPT system. Alternatively, the system should be equipped to detect and deactivate in the presence of humans to prevent exposure, taking into account the speed of approach specific to each class of the system. Compliance with the regulatory limits outlined in Table 8 during normal operations is essential in the absence of previous security measures. Regions 2 and 3 must comply directly with the regulatory limits defined in Table 8.

SAE J2954 standard also establishes regulatory limits for regions 2 and 3 in the case of people with CIEDs placed on their torso, in compliance with the ISO 14117 (

Table 9. CIED EMF exposure limits established in the SAE J2945 standard for frequencies between 79 kHz to 90 kHz [3].

Regions	Magnetic Flux Density B (μT)
2,3	15

) [91]. Additional limits for other IMDs are still being considered by the committee responsible for SAE J2954 [3].

The values in Table 8 and Table 9 apply to sinusoidal modulated WPT fields. When non-sinusoidal, assessments should be conducted considering peak limits equal to the root mean square multiplied by $\sqrt{2}$.

IEC 61980-1 states, in section 11.8, that WPT systems and analogues should not expose humans to dangerous EMF above the limits established in international guidelines, usually ICNIRP and IEEE [92]. In addition, ISO 5474-4 states in section 10.4.3 that regions 2 and 3, which are equivalent to those referred to in SAE J2954, must not exceed the regulatory limits defined in the ICNIRP 2010 guidelines [93].

The standards observed apply to stationary systems, and the regulations for dynamic systems are lacking. There are some initiatives and standards to extend existing regulations for stationary systems to dynamic systems. One of these is the SAE J2954/3 standard, which establishes specifications for airworthiness criteria for light-duty EVs and heavy-duty EVs. However, in its current scope, the standard does not address electric two-wheelers, such as scooters or motorbikes, highlighting a gap in existing regulations for DWPT in the micromobility sector. This standard is still classified as Work In Progress (WIP), and there is no date set for its application. However, it will establish standards for interoperability, electromagnetic compatibility, EMF, minimum performance, safety, and testing for dynamic wireless power transfer. It will not yet address the possibility of bidirectional power transfer, which will be reserved for a future standard [4].

The IEEE/IEC 63184 standard establishes assessment methods of human exposure to EMF from stationary and dynamic WPT systems. It also presents the intention to regulate EMF exposure related to CPT systems operation in future editions [94].

To ensure the safety of EV drivers and pedestrians, DIPT systems must comply with international guidelines on human exposure to electromagnetic fields, such as those defined by ICNIRP. However, compliance with these limits can be challenging, particularly under misalignment conditions or at higher power levels. As a result, shielding strategies play a crucial role in reducing unintended EMF emissions and ensuring compliance with regulatory thresholds. Section 5.2 the design and implementation of magnetic shielding techniques designed to decrease stray fields in DWPT applications.

5.2. EMF Mitigation and Shielding Strategies for Compliant Deployment

Shielding strategies are designed to control EMF leakage. Limited EMF leakage affects adjacent equipment and interferes with human health, fauna, and flora. EMF leakage must be controlled by the IPT system in order not to exceed the maximum limits allowed by international standards such as SAE J2954 or IEC 61980 [3,95]. Shielding is usually placed adjacent to the coils to help improve efficiency and coupling [29,96]. Shielding types can be classified as active, passive and reactive, with the passive category being sub-classified as magnetic, conductive or magneto-conductive (Figure 18).

Active shielding consists of adding extra windings to the coils in order to generate an inverse magnetic field to the power transfer magnetic field, thereby mitigating EMF leakage [97,98]. Naturally, there is the design problem of placing the windings to reduce EMF leakage without interfering with the magnetic field and, consequently, the efficiency of the system. The extra energy consumed to power the shielding coils also contributes to a reduction in efficiency, along with an increase in the cost and size of the system [28,29].

Conductive passive shielding uses the effect of magnetic induction on conductive materials, where eddy currents are created. The currents created in turn induce a magnetic field opposite to the field that generated them, counteracting it. EMF leakage is reduced, with efficiency losses of around 1% to 2% due to Joule losses related to eddy currents [28]. There are examples of EVs that employ the chassis and body itself to reduce EMF leakage while still affecting the efficiency of the system [97,99]. The SAE J2954 standard recommends an aluminium shield with dimensions of 800 mm × 800 mm × 0.7 mm [3].

Magnetic shielding uses materials that have high magnetic permeability and are non-conductive, helping to diffuse the electromagnetic flux to enhance power transmission. This solution has been shown to increase system efficiency while reducing EMF leakage. A practical example is the use of adjacent ferrite, which contributes to the above effects, but has the downside of increasing the cost of the IPT system. Some variations of ferrite application range from radial bars to rings around the coil [28,29,98,100].

Most passive shielding applications use solutions that combine conductive and magnetic subtypes. As a rule, conductive shielding is used in conjunction with a ferrite body in order to improve efficiency while reducing EMF leakage [28].

Reactive shielding operates in a similar way to active shielding, by adding coil windings to reduce EMF leakage. However, the extra windings are not powered by the system, but are isolated from the primary and secondary coils and in a closed circuit with capacitors in parallel. The current flowing through the shielding coil originates from the magnetic field induced by the primary and secondary coils, resulting in a magnetic field that reverses the original one, thereby reducing EMF leakage. There are no controls over the reverse magnetic field, which is possible with active shielding [29,86,101]. Other strategies employ a phase difference to reduce MF leakage [82].

Passive shielding is effective in reducing EMF leakage to levels below those mentioned in international standards for powers below 100 kW. For higher powers, active and reactive solutions are more efficient [28].

6. Scheduling and Charging Optimisation of Dynamic Wireless Charging in Smart Grids

6.1. Forecast Informed Scheduling

6.1.1. Forecast Inputs and Uncertainty Modelling

Forecast-informed scheduling in DWPT differs from conventional smart charging because the electrical demand is distributed along roadway segments and varies with traffic. As a result, the scheduling layer must translate forecasts from both the power system and the transport system into segment-level energisation and power allocation decisions, while preserving distribution network operability [102,103,104]. This workflow is summarised in Figure 19.

In DWPT corridors, the relevant electrical inputs are not limited to load forecasts at the feeder level, but also include forecasts of loading at the supplying transformers and feeders, together with the expected distributed energy resources (DER) output, as these quantities determine the margin available to accommodate additional, traffic-driven charging demand [105]. Consistent with the standard definition of net load as system demand minus non-dispatchable generation, a convenient operational quantity is the time-indexed net load (Equation (10)) [106].

$$P^{\mathrm{net}}\left(t\right)=\left[P^{\mathrm{base}}\left(t\right)+P^{\mathrm{DWPT}}\left(t\right)\right]-P^{\mathrm{DER}}\left(t\right)$$

(10)

In this equation, $P^{\mathrm{base}}\left(t\right)$ denotes the predicted background demand, $P^{\mathrm{DWPT}}\left(t\right)$ the anticipated incremental demand arising from DWPT operation along the corridor, and $P^{\mathrm{DER}}\left(t\right)$ the expected local variable generation. Recent work has modelled aggregated DWPT loads and proposed risk-constrained operating strategies to coordinate uncertain DWPT demand with renewables in associated distribution systems [105].

The main steps used to obtain the time-varying corridor charging demand from mobility information and an aggregated load model are summarised in Figure 20.

In practice, $P^{\mathrm{DWPT}}\left(t\right)$ can be predicted by coupling a mobility forecast with an aggregated DWPT load model, where the mobility forecast provides the expected corridor utilisation over time [105,107]. For example, GPS trajectory data can be processed to obtain road segments and time-dependent vehicle presence statistics from segment entry and exit times, which can then inform the expected traffic intensity on electrified segments [107]. A simple stochastic representation of this mobility input models the number of vehicles arriving in a time interval $\Delta t$ as a Poisson process [105]. This is given in Equation (11), where $N\left(t\right)$ is the cumulative number of arrivals up to time t, $\xi$ is the realised number of arrivals within $\Delta t$, and $\lambda$ is the average arrival rate [105].

$$\mathbf{P}\left[N(t+\Delta t)-N\left(t\right)=\xi \right]={\displaystyle \frac{e^{-\lambda \Delta t}{(\lambda \Delta t)}^{\xi }}{\xi !}},\xi =0, 1, 2,\dots$$

(11)

When vehicle arrivals follow a Poisson process, the headway h between consecutive vehicles can be modelled by a negative exponential distribution. This is given in Equation (12), where $\beta$ is the headway distribution parameter [105].

$$f_H\left(h\right)=\beta e^{-\beta h},h\ge 0.$$

(12)

Given realised arrivals and headways, the charging start time of each vehicle w can be denoted as $T_{\mathrm{start},w}$. A per-vehicle charging power profile can then be represented as a time-indexed function, as expressed in Equation (13), where $P_{\mathrm{EV},w}$ is the charging power assigned to vehicle w [105].

$$P_w\left(t\right)=\left\{\begin{array}{cc}0,\hfill & t<T_{\mathrm{start},w}\hfill \\ P_{\mathrm{EV},w},\hfill & t>T_{\mathrm{start},w}.\hfill \end{array}\right.$$

(13)

The aggregated corridor-level DWPT demand is obtained by superimposing the power profiles of all vehicles, as given in Equation (14), where $N_{\mathrm{total}}$ is the total number of vehicles considered in the modelling horizon [105].

$$P^{\mathrm{DWPT}}\left(t\right)=\sum _{w=1}^{N_{\mathrm{total}}}{P}_w\left(t\right).$$

(14)

This formulation links mobility-derived corridor utilisation to a time-indexed electrical demand profile. It is also consistent with scenario-based operation, because $\lambda$ and $\beta$ can be updated from current-period traffic statistics and used with Monte Carlo sampling to generate multiple realisations of $P^{\mathrm{DWPT}}\left(t\right)$ in a rolling optimisation framework [105].

Mobility forecasts determine where and when vehicles interact with electrified road segments, thereby defining the expected temporal windows of power transfer and the spatial distribution of DWPT demand. This motivates coupled power transport formulations in which traffic conditions and route patterns influence both the planning and operational scheduling of DWPT infrastructure [102,103,104]. In particular, frameworks that jointly consider placement and scheduling highlight that forecasting route usage and corridor utilisation is essential to anticipate segment load and to allocate charging capacity in a manner consistent with network limits [79,104].

DWPT systems are commonly implemented using segmented transmitter arrays embedded in the roadway, where segments are energised selectively as vehicles pass. Consequently, the scheduling problem depends on the availability of each segment and on the effective coupling window expected during vehicle traversal, both of which shape the achievable transferred energy in operation. Segment availability considerations also interact with activation strategies and the discretised nature of power delivery in dynamic conditions, reinforcing the requirement for forecasts that describe not only traffic presence but also the operational availability of the infrastructure along the corridor [26,108].

Forecast errors propagate directly into distribution constraints in DWPT, because uncertainty in traffic and coupling affects the realised DWPT load, while uncertainty in DER output affects the headroom available to supply it. Practical uncertainty representations therefore include probabilistic forecasts and prediction intervals, scenario generation for rolling-horizon optimisation, and risk-aware or robust formulations that control constraint violation under heterogeneous uncertainties [105]. Risk-constrained operating strategies using coherent risk measures have been proposed to manage the operational risk associated with uncertain dynamic charging loads and renewables in distribution systems, providing a direct pathway to embed uncertainty within DWPT scheduling objectives and constraints [105].

6.1.2. Scheduling and Decision Layers

Recent work indicates that the operational management of dynamic wireless charging should be framed as a multi-timescale scheduling problem with clearly separated horizons. This structure is summarised in Figure 21.

A day-ahead layer is typically used to establish a baseline plan from demand projections that consider the traffic and grid constraints. An intraday layer updates that plan as realised vehicle flows, renewable output, and background demand deviate from forecasts. This separation is motivated by the fact that forecast error generally grows with lead time, which encourages re-optimisation rather than committing to a fixed schedule over long horizons [105]. Complementary evidence from coordinated fleet charging shows that a structured combination of day-ahead scheduling with intraday adjustments can improve operational performance relative to single-horizon schemes [109]. At the actuation level, dynamic charging additionally requires fast control to manage engagement and disengagement events and to maintain stable battery charging behaviour during motion, with model predictive control being proposed and evaluated for this purpose in DWPT settings [110]. In practice, this layered structure is reinforced by the emergence of dedicated dynamic wireless charging load forecasting methods, which are explicitly positioned as inputs to day-ahead and intraday scheduling [111].

From a systems perspective, recent studies also support the decomposition of decisions into hierarchical layers spanning allocation on the network level. At this level, the principal concern is the compatibility of aggregated dynamic charging demand with distribution system constraints and the availability of flexibility from distributed energy resources, while controllers on the corridor level translate mobility patterns into spatially and temporally power requests. Hierarchical operational frameworks have been proposed in which coordinated control is performed across these layers to ensure that dynamic wireless charging demand is served without violating the limits of the distribution system, while exploiting available flexibility where possible [112]. Beyond purely operational coordination, coupled power transportation models have introduced tri-level formulations in which higher-level policy decisions such as time-varying tolls influence route choice and, consequently, the spatio-temporal charging demand that the power system must accommodate [113]. Finally, multistage robust formulations for dynamic charging lane planning further emphasise the value of explicitly separating strategic decisions from operational recourse, particularly when uncertainty depends on the decisions and affects both infrastructure utilisation and grid impact [114].

6.1.3. Objective Functions for Cost, Carbon Dioxide Emissions and Quality of Service

Recent optimisation formulations for dynamic charging corridors increasingly consider cost and carbon dioxide emissions as coupled objectives. Studies on DWPT infrastructure that consider carbon dioxide emissions include dynamic tariffs and schedules that explicitly account for emissions constraints and uncertainty by embedding emission considerations alongside uncertain price trajectories in tariff design and scheduling [115]. In parallel, the control strategies that consider carbon dioxide emissions are gaining traction, including approaches that combine time-of-use pricing with marginal carbon dioxide emission factors as time-varying signals for smart charging [116]. However, recent studies show that strategies that consider emission factors can become unreliable at scale when too many vehicles follow the same signal, motivating fleet-partitioned or cascading signalling so that carbon-aware charging remains effective under increasing EV adoption [117]. These economic and carbon objectives are typically co-optimised with grid impact terms such as peak management, technical losses, and distribution operating limits, and dedicated models have been proposed for the coordinated operation of dynamic wireless charging loads together with distribution networks [112,118,119]. In addition, infrastructure health can be incorporated directly, with controlled charging under demand-charge exposure used to align bill reduction with asset ageing considerations [120].

The main objectives are reported in optimisation-based scheduling and control of DWPT corridors are summarised in Figure 22.

User service and system efficiency objectives further extend the scheduling problem beyond feeder-centric metrics. At the service level, formulations commonly enforce range assurance through state-of-charge targets by deadlines [121]. Recent work also formalises fairness and quality of service explicitly, including interpretable policy design methods and quality of service models [122,123]. At the system level, dynamic charging corridors behave as capacity-constrained service systems, so segment utilisation, queueing delay, and throughput become optimisation targets alongside energy delivery [124]. Traffic-side interventions such as lane control and dynamic pricing have been proposed to balance transport performance with charging efficacy, including corridor control formulations for wireless charging lane operations [125,126]. Finally, because DWPT transfer efficiency depends on coupling that varies with trajectory and misalignment, coupling scheduling benefits from models that map mutual inductance uncertainty to delivered power and efficiency, whilst converter-level methods support simultaneous power regulation and efficiency maximisation under changing coupling [127,128,129].

6.1.4. Constraints and Feasibility

At grid level, recent hosting capacity and congestion evidence frames electrified transport demand as admissible only if distribution network security constraints remain satisfied, with binding technical limits including nodal voltage bounds and thermal ratings of network assets [130,131]. In coupled electrified road settings, these requirements are increasingly handled through hierarchical or bi-level formulations that explicitly link corridor demand to active distribution network operation under uncertainty, thereby making network feasibility a first-class constraint rather than a post-processing check [132,133]. At asset level, transformer loading alone is insufficient to represent risk because ageing is driven by temperature-dependent mechanisms, so scheduling studies embed thermal models and explicit ageing proxies such as highest temperature and lifetime cost terms [120,134,135].

At DWPT layer, segmented infrastructure introduces hard limits on segment power and energisation duty cycle, and it creates position-dependent dynamics because switching patterns modify the effective coupling seen by the receiver, which can lead to transferred-power pulsation that must be controlled [7,108]. Consequently, recent optimisation load models use segment-wise representations of delivered power over the traversal of a segment. They integrate them within rolling optimisation to maintain tractability while preserving the time-space structure of corridor demand [105]. Vehicle feasibility further constrains the admissible power request through battery state-of-charge bounds, charge-acceptance limits, and mobility requirements, which couple the corridor schedule to transport-side behaviour in integrated transport-power frameworks [132,134].

6.1.5. Optimisation Formulation and Solvers

Deterministic scheduling formulations for electric-vehicle charging and dynamic charging corridors are typically cast as linear programming or quadratic programming problems when the control variables are continuous set-points, and as mixed-integer linear programming problems (MILP) when discrete decisions must be represented explicitly, such as assignment, switching, and infrastructure activation. In this setting, it is technically useful to draw an analogy with unit commitment, because both problem families combine inter-temporal coupling with binary decisions and benefit from carefully chosen mixed-integer programming formulations and solution enhancements that improve tractability [136]. Within coupled power transportation perspectives, optimisation scheduling has also been used to coordinate charging and discharging with representations thar consider the network and price signals. This thereby links the vehicle side decisions to the distribution network operation [137]. For dynamic wireless charging deployments, integrated MILP models have been reported that connect timetable or service feasibility with charging infrastructure planning decisions, reflecting the practical need to co-design operational schedules and electrified corridor assets [138].

When uncertainty in renewable output, demand, market prices, or EV availability is material, stochastic and robust formulations become central, because they preserve operational viability and economic performance under imperfect forecasts. Two-stage stochastic unit commitment variants have been proposed that combine scenario modelling with decomposition strategies to improve computational scalability as the scenario set grows, which is directly relevant when scheduling must be refreshed frequently [139]. In risk-averse charging and bidding, mean conditional value at risk (CVaR) linear programming has been used to encode explicit risk preferences for EV fleet aggregators participating in the day-ahead market, using price scenarios to produce charging templates and bidding curves [140]. Beyond scenario-based stochastic programming, with robust distribution optimisation has been applied to hedge against ambiguity in the underlying probability law, including distributionally robust optimisation (DRO) formulations based on the Wasserstein-metric for unit commitment and related operational problems [141,142]. Robust distribution-system operation models that co-ordinate aggregator charging with other flexibility actions have also been reported, combining two-stage structures with robust and cut-based solution procedures [143].

Learning-augmented scheduling is increasingly reported as a complement to mathematical programming when real-time decisions must be taken repeatedly under uncertainty and at large scale. Offline reinforcement learning has been proposed for EV charging scheduling, with the problem posed as a constrained Markov decision process and with explicit mechanisms intended to support constraint compliance during deployment [144]. For large-scale coordination, recent work integrates graph neural networks with reinforcement learning to exploit the graph structure induced by charging networks and grid constraints, reporting improved scalability and generalisation in experimental studies compared with baseline reinforcement learning approaches. In this line of work, the technical point is not that optimisation models become obsolete, but that learning scheduling can be used to deliver fast sequential decisions in settings where repeated MILP re-optimisation may be operationally impractical [145].

6.1.6. Data Requirements and Operational Infrastructure

Operational control of dynamic electrified infrastructure depends on heterogeneous telemetry, spanning grid-side measurements and mobility-side observations. In distribution systems, state estimation is explicitly framed as a prerequisite to enable automatic control and intelligent functionalities, because operators need a reliable view of network operating conditions rather than isolated point readings. Recent distribution system state estimation work also underlines that distribution networks are inherently unbalanced and therefore require accurate three-phase models, since common simplifying assumptions can degrade estimation accuracy [146]. In parallel, real-time topology detection and state estimation have been formulated using micro-phasor measurement unit and smart meter data, motivated by the practical reality that distribution networks have limited sensor coverage and that communication links to switching devices may be weak or not well established [147]. Taken together, these results make the data requirement concrete, as operational decisions must be supported by time-aligned measurements and estimators that remain valid under low observability and unbalance [146,147].

The communication layer must then deliver these data products to controllers within the latency and reliability envelope required by time-sensitive services. A recent standards-based study of long term evolution (LTE) and new radio (NR) vehicle-to-everything (V2X) describes the main V2X classes, including vehicle-to-infrastructure (V2I) where vehicles communicate with roadside transmission nodes installed along roads. It also notes that V2X use cases demand network requirements such as low latency and high reliability. The same work evaluates reliability through packet delivery ratio and reports that centralised resource allocation achieves better performance than decentralised mode 4, at the cost of requiring cellular coverage, while LTE alone is characterised as suitable only for basic V2X use cases. It further details that if sidelink control information is not correctly received, the associated transport blocks must be discarded, which operationally links packet loss directly to application-layer reliability. To meet stricter future requirements, it proposes multi-radio access technology (multi-RAT) dual connectivity that combines LTE and NR, explicitly positioning heterogeneous connectivity as an architectural lever to achieve demanding latency and reliability targets [148].

6.2. Control Strategies for Grid Integration

In dynamic wireless charging corridors, the segment-level layer is fundamentally a power electronics control problem in which IPT converters must regulate delivered power despite coupling changes and EV side variability. Recent primary-side control work for EV-oriented resonant IPT charging explicitly frames duty-ratio control and frequency control as common levers to regulate charging current and voltage under unavoidable misalignment and parameter variation, so that constant current (CC) and constant voltage (CV) modes can be maintained [149]. The same study further reports fixed-frequency phase-shift (PS) strategies alongside asymmetric clamped mode and asymmetric duty cycle control, and evaluates them in terms of output-voltage range and the ability to retain zero-voltage switching across the full control range [149]. Complementarily, constant-voltage and constant-current regulation has been demonstrated via secondary-side actuation using full-bridge synchronous rectification, where the transition between CC and CV output is achieved by adjusting only the duty cycle of the rectifier switches, and validated on an 11 kW IPT prototype [150]. These results justify a control hierarchy in which fast inner regulation at the segment level is paired with slower supervisory decisions that determine admissible set-points and operating envelopes for each segment.

At the grid supervision layer, the technical requirement is to translate corridor-level charging demand into set-points while respecting feeder constraints and exploiting flexibility from distributed assets. A recent dynamic wireless charging operation model makes this explicit through a hierarchical architecture in which local controllers convert route traffic flows into charging requests at distribution nodes, and a central controller coordinates set-points by enforcing an optimal power flow model and coordinating distributed energy resources (DERs) [112]. In distribution-operation practice, such grid-level coordination is increasingly discussed in terms of software platforms that manage heterogeneous DER flexibility, including EV charging, within operational limits. In this regard, a hybrid distributed energy resource management system (DERMS) framework is proposed as a tool for safe mass EV integration, and a real feeder case study reports the use of aggregated EV flexibility via DERMS to manage overloads and voltage violations in real time [151]. Consistently with hierarchical control views of EV charging stations, the energy management system layer is described as providing management signals to lower-level controllers on the order of several minutes, thereby separating fast regulation from slower optimisation and constraint management [152]. Finally, droop control remains relevant as a coordination mechanism for inverter-interfaced resources, and recent work shows that droop coefficients can be adapted by deep reinforcement learning to address reactive-power sharing limitations, which is directly aligned with supervisory coordination concepts [153].

6.2.1. Coupling Variation Control and Efficiency Constraint Management

Coupling variation is a first-order driver of efficiency in dynamic wireless charging corridors, because the mutual inductance and coupling coefficient shape the reflected impedance and therefore the achievable operating point. Recent contributions emphasise that real-time identification of mutual inductance and load is technically valuable for control and condition monitoring, including hybrid approaches that combine equivalent-circuit relations with data-driven regression and that can operate with limited measurements such as direct current input current and a single root mean square voltage measurement [154]. Complementarily, maximum efficiency tracking control with real-time mutual inductance estimation has been reported for inductive charging systems [155]. For coupling estimation itself, Reference [156] reports coupling-coefficient estimation through damped frequency, while learning-based estimators have also been demonstrated for mutual inductance estimation using an artificial neural network in EV charging. In DWPT, misalignment adaptation can be framed as an online estimation problem coupled to actuation, and receiving-side current based lateral misalignment estimation with automated steering control has been reported to reduce misalignment during motion. Related work also reports misalignment tolerance extension using slightly frequency-detuned compensation [157].

Thermal constraints and battery ageing considerations further bound admissible set-points even when coupling is favourable, as losses in magnetic components and power electronics translate into temperature rise and accelerated degradation if unmanaged. A multi-physics transient thermal simulation methodology for an IPT pad embedded in asphalt has been reported, motivated explicitly by the risk that thermal losses can exceed safe operating temperature and by the need to reproduce long-duration transient temperature rise and drop with validated accuracy [158]. An integrated electro-thermal analysis for inductively coupled power transfer embedded in pavement, combining laboratory evaluation with numerical modelling to analyse both transmission efficiency and thermal behaviour under realistic embedment and offset conditions [157]. On the battery side, charging strategies that explicitly account for battery ageing have been proposed by adapting charging thresholds and establishing an optimal adaptive depth-of-discharge range to improve battery health and lifespan [159]. Taken together, these works justify treating thermal limits and battery ageing objectives as explicit constraints in efficiency management, alongside coupling and misalignment adaptation [157,158,159].

6.2.2. Interoperable Control Interfaces and Security for Cyber Physical Dynamic Charging Infrastructures

Interoperability in dynamic charging corridors depends on aligning interfaces across the charging station, the back-end management system, and the vehicle-side charging communication stack. The Open Charge Point Protocol version 2.0.1 (OCPP) is analysed in the recent security literature as an interface between charging stations and their management system, and it is reported to incorporate device management, smart charging functions, and support for International Organization for Standardization standard 15118 (ISO 15118) in addition to updated security functions [160,161]. In parallel, Plug and Charge communication has been studied through the design and implementation of the Transport Layer Security (TLS) handshake in the ISO 15118-20 context, with test bench validation reported for successful authentication and encrypted message exchange [162]. Taken together, these strands indicate that interoperable control should be framed as an end-to-end contract across protocols and roles, where charging transaction control, identity binding, and cryptographic session establishment are treated as coupled requirements rather than independent design choices [160,163].

Cyber physical security then requires an explicit threat model that covers both cyber manipulation of control, billing messages and the connectivity that carries telemetry and authorisation. For OCPP version 2.0.1, STRIDE is applied to classify threats and the associated study reports that, despite added security measures, security risks remain and require additional protection [160]. At the corridor level, denial of service can be realised through radio interference when vehicle to everything (V2X) links are part of the operational architecture, and recent New Radio vehicle to everything (NR V2X) results show that jamming can severely undermine communication reliability in both simulation and experimental perspectives [164]. For dynamic wireless charging, the authentication literature reports protocol designs based on elliptic curve cryptography, physical unclonable functions, and also proposes post quantum secure authentication, which together illustrate the direction of travel toward stronger cryptographic identity and session establishment under mobility [165,166,167]. From a deployment perspective, guidance such as the National Institute of Standards and Technology Cybersecurity Framework Profile for EV extreme fast charging can be used to structure risk management across the ecosystem, including cloud or third-party operations and utility interfaces [168].

6.2.3. Concluding Remarks

Charging optimisation and control for corridor-scale DIPT is inherently a multi-layer decision problem that couples grid objectives, infrastructure limits, and vehicle level constraints. At the planning level, uncertainty in demand and operating conditions motivates rolling horizon formulations that combine day ahead allocation with intraday updates so that set points can be revised as information improves. At the operational level, the coupling variation induced by motion means that efficiency oriented regulation benefits from real time estimation of mutual inductance and load, while admissible operating points must explicitly account for thermal limits in embedded pads and power electronics, and for battery ageing constraints so that short term energy delivery does not compromise lifetime. Finally, the reviewed interoperability and cybersecurity evidence indicates that corridor operation depends on an end to end control and communication contract in which authenticated control, cryptographic identity, and session establishment are treated as operational requirements under mobility.

7. Discussion: Implications for Charging Optimisation in Urban Energy Systems

This section synthesises the technical findings into implications for charging optimisation in urban energy systems. Section 7.1 discusses how urban load patterns and demand that is specific to micromobility shape scheduling and control objectives. It then links these demand drivers to infrastructure planning decisions, including segment placement and sizing, and to efficiency at the scale of the overall system. It also addresses how corridor operation should be coordinated with other urban electrification processes, including public transport electrification and distributed energy resources. The role of exposure and safety compliance as design drivers that directly constrains operating points is analysed in Section 7.2. It also discusses interoperability and standardisation, including the role of harmonised interfaces and authentication. It then connects these requirements to grid code compliance and power quality constraints, and to data governance topics such as privacy, secure telemetry, and operational accountability.

Section 7.3 focuses on limitations and implications for future smart grids coordination. It covers forecast and model limitations, scalability and computational constraints, communication and cyber physical risks, infrastructure and adoption uncertainties, and research directions for coordination methods that remain robust under these practical constraints.

7.1. Integration of Electric Micromobility with Urban Power Grids

7.1.1. Urban Demand Patterns and Implications for Micromobility Charging Loads

Shared micromobility trip demand exhibits systematic temporal and spatial regularities that are directly relevant when anticipating where and when charging demand may concentrate. A full-year empirical analysis of shared e-scooter operations reports a pronounced weekday bimodal structure, with activity rising during morning and afternoon commuting periods, and it also reports that trip origins and destinations are strongly concentrated in highly accessible central zones [169]. At the metropolitan scale, evidence from 124 European cities further indicates that e-scooters act as a feeder mode to public transit with integration ratios that vary widely across cities, and with reported temporal patterns that differ between first-mile and last-mile trips [170]. At the city scale, a spatio-temporal analysis of 322,000 shared e-scooter trips identifies that 25% of trips are access or egress trips to rail transit, which is consistent with demand concentrating around multimodal hubs [171]. When charging is organised around fixed or incentivised charging locations, this spatial concentration becomes a constraint in infrastructure planning, and recent work therefore formulates charging station planning for shared electric micromobility as an optimisation of station locations and capacities with explicit representation of battery use and charging processes [172]. From the distribution-network perspective, coincidence-factor thinking is widely used for feeder sizing in EV charging and is tied to simultaneous peak demand [173]. Empirical results for domestic electric-vehicle charging show that aggregation can substantially reduce coincidence factors, which provides an operationally useful analogy when assessing how clustered micromobility charging could translate into localised peaks on specific urban feeders [174].

To support readability and to link these empirical regularities to power-system considerations,

Table 10. Links between micromobility demand regularities, distribution-level impacts, and planning and operation implications.

Demand Regularity	Distribution-Level Implication	Planning and Operation Implication
Weekday bimodal demand with commuting peaks [169]	Higher probability of time-localised peaks when charging is synchronised with commute-driven activity	Time-dependent scheduling and incentive design to avoid peak coincidence in constrained periods
Spatial concentration in highly accessible central zones [169]	Localised loading on specific feeders near activity hotspots	Grid-aware siting and sizing to prevent clustering on weak feeders and to prioritise locations with hosting headroom
Feeder-mode integration with public transit, with first-mile and last-mile temporal differences [170]	Demand concentration near transit corridors and hubs, with timing that depends on trip purpose	Siting near multimodal nodes with capacity planning that accounts for directional, time-dependent usage
Access or egress trips to rail transit concentrated around multimodal hubs [171]	Spatially concentrated charging demand around hubs can create recurring stress on local assets	Arrangement with explicit local constraint checks and operational control during peak periods
Charging organised around fixed or incentivised locations [172]	Concentration becomes a binding constraint that can dominate infrastructure feasibility	Joint location-and-capacity optimisation with explicit representation of battery use and charging processes
Coincidence-factor perspective from EV feeder sizing [173] and empirical reduction under aggregation [174]	Aggregation can reduce coincidence, but local clustering can still yield feeder-specific peaks	Use coincidence-aware design at feeder scale, complemented by spatial controls to mitigate localised clustering

consolidates the main demand patterns reported in the literature, their distribution-level implications, and the corresponding planning and operational responses that are typically required for grid-aware deployment.

Overall, these regularities indicate that micromobility charging demand is not only time-varying but can also be spatially concentrated, so local feeder constraints can become binding even when system-level aggregation effects are present. This motivates grid-aware infrastructure planning and siting, discussed next, where candidate locations and capacities must be screened against distribution feasibility while meeting coverage and accessibility objectives.

7.1.2. Infrastructure Planning and Siting

Infrastructure planning and siting for urban charging corridors can be formalised as a coupled problem in which candidate locations are filtered by grid hosting limits and then selected to meet coverage and equity objectives under uncertainty. Recent grid aware planning work demonstrates this coupling by forecasting charging demand and then validating additions against distribution network operating limits, where grid hosting capacity is assessed on a feeder model using Open Distribution System Simulator (OpenDSS) and peak load voltage checks against baseline voltage thresholds [175]. At the system level, integrated expansion frameworks such as CHARGE-MAP report multi criteria formulations that link spatio-temporal demand modelling with infrastructure expansion decisions, which is consistent with corridor selection practices that must balance demand coverage with grid feasibility [176]. Hosting capability can also be shaped by siting decisions that explicitly incorporate flexibility assets, since a recent planning study reports strategic allocation of charging stations with an energy storage system and a stochastic Monte Carlo demand model, with hosting evaluation that accounts for voltage constraints, thermal limits, power losses, and voltage imbalance [177]. Equity and accessibility objectives can be embedded directly in the siting model, as shown by a data driven multi-objective formulation that trades off site development cost, demand coverage, and social equity access, solved using Non dominated Sorting Genetic Algorithm II (NSGA-II) [178]. Complementary evidence on inequitable access strengthens the planning motivation, because a national analysis reports that disadvantaged communities have fewer public EV charging stations per capita and experience more reliability issues, while a curbside planning study reports that a high need scenario targeting underserved areas delivers the greatest equity benefits among its scenarios [179].

7.1.3. End-to-End Efficiency and Carbon Intensity Implications for Urban Charging Operation

End-to-end efficiency in urban charging systems is best treated as a grid-to-battery accounting quantity that aggregates conversion losses in the grid interface and power electronics, transfer losses in the charging interface and time integrated fixed losses such as standby draw over the operating window [180]. Recent experimental WPT hardware that targets interoperability reports grid to battery efficiency of 91.7% to 94.3% under specified aligned operating conditions, which provides a concrete benchmark for what is achievable when the complete conversion chain is engineered rather than only the link [181]. Emissions outcomes then depend on when charging energy is drawn, because grid carbon intensity varies over time and the signal used to represent that variation changes the incentives seen by a scheduler. A data driven analysis shows that average and marginal carbon intensity signals can be statistically different and weakly correlated across regions, such that optimising against one signal can increase emissions when assessed with the other [182]. At higher adoption and under widespread coordination, modelling results with real charging data indicate that simple emissions factor driven charging can fail to reduce emissions reliably, while a cascading marginal strategy reduces added emissions by 10% to 28% across tested scenarios [117]. These results motivate optimisation objectives that account for time varying carbon intensity explicitly, including recent stochastic tariff design that incorporates forecast uncertainty and enforces an emissions budget through chance-constrained optimisation [115].

7.2. Regulatory Constraints as Enablers of Sustainable Deployment

7.2.1. Exposure and Safety Compliance as a Design Driver

For dynamic wireless charging corridors, exposure compliance is typically operationalised by translating electromagnetic field limits into explicit engineering constraints that must hold under credible worst case operating conditions and misalignment states. The ICNIRP guidelines define basic restrictions and reference levels for time varying fields that are used in compliance assessment, which in turn motivates measurement planes and conservative operating envelopes in deployment practice [84,85]. Recent WPT studies address a practical difficulty of this translation, namely that leakage fields around couplers are strongly non-uniform, by proposing spatial averaging procedures intended to better relate external magnetic field measurements to quantities used in induced field reasoning [183]. In parallel, assessments carried out explicitly in the context of the SAE J2954 wireless charging standard show how compliance judgement depends on where and how magnetic fields are measured around the vehicle and infrastructure, which reinforces the need to treat coil geometry, shielding, and power set-points as compliance relevant design variables rather than post hoc validation checks [81].

7.2.2. Interoperability and Standardisation

Interoperability is largely defined through the standards stack that specifies electrical and geometric compatibility between the ground assembly and the vehicle assembly, including operating frequency bands and alignment related constraints that must be satisfied for interchangeability [3,184,185]. SAE J2954 specifies a WPT operating frequency range of 79 kHz to 90 kHz and provides requirements that include alignment methodology and interchangeability aspects for light duty wireless charging [3]. ISO 5474-4 similarly specifies safety and interoperability requirements for WPT systems for road vehicles, which positions conformance as a prerequisite for multi vendor deployment [93]. Recent surveys and reviews synthesise how SAE J2954, ISO 5474-4, and the IEC 61980 series jointly shape vehicle infrastructure compatibility, test methods, and certification practices, while also emphasising that interoperability is not achieved by frequency selection alone and requires consistent conformance testing and verification protocols across the ecosystem [92,184,185].

7.2.3. Grid Code and Power Quality Requirements

Grid code compliance for electrified corridors must be treated as both a converter emissions problem and a distribution network operating limits problem, where voltage constraints, thermal limits, and harmonic distortion bounds can all become binding as charging demand scales [186,187]. A recent distribution network review consolidates evidence that EV charging can raise peak demand and aggravate voltage instability and thermal stress in distribution feeders, which motivates regulated charging and network management in planning and operations [186]. On the emissions side, recent measurement work on smart charging reports that total harmonic distortion (THD) can increase as charging current decreases and that aggregate current THD can reach 25% under worst case multi vehicle conditions, which strengthens the argument that harmonic compliance must be evaluated across operating points rather than only at rated power [187]. For wireless chargers specifically, recent electromagnetic compatibility focused experimentation evaluates conducted and field bound interference emissions and rates measured magnetic fields against proposed limits from standards. This studies also show that details such as direct current supply cable geometry can affect compliance outcomes, which further motivates standardised test setups and explicit filtering and layout constraints [188].

7.2.4. Data Privacy and Operational Accountability

Regulatory and contractual accountability in public charging ecosystems depends on the ability to demonstrate correct operation, correct billing, and security of the cyber physical interfaces that connect chargers, control platforms, and grid facing services [189,190]. In depth security analysis of EV charging station management systems reports remotely exploitable vulnerabilities and discusses attack implications that include service disruption and potential impacts on the power grid, which supports treating secure telemetry, authenticated control, and incident traceability as operational requirements rather [189]. From the privacy perspective, published analyses map data flows in the charging ecosystem and evaluate privacy preservation relative to charging use cases and standards adherence, which directly motivates data minimisation, anonymisation, and purpose-constrained processing when mobility linked charging records are handled [190]. In this context, audit logs and traceability can be justified as mechanisms that support post event investigation and compliance reporting, provided that logging is itself designed to minimise unnecessary personal data exposure [189,190].

7.3. Limitations and Implications for Future Smart-Grid Coordination

7.3.1. Key Limitations for Coordination of Dynamic Charging with Smart Grids

Coordination based on forecast depends on models that remain valid as urban demand evolves. Online probabilistic forecast combination explicitly targets concept drift in load time series by updating combination weights online rather than relying on a static ensemble [191]. Related smart grid studies propose concept-aware lightweight adaptation mechanisms to maintain forecasting accuracy under evolving operating regimes [192].

In practice, forecast pipelines must also be robust to data gaps and malicious perturbations that can occur in metering and communication chains. An Applied Sciences study proposes an ensemble design aimed at resilience to missing values, adversarial attacks, and concept drift in load forecasting [193]. Even with robust forecasting, operational models can be mismatched to field conditions when they linearise inherently non-linear distribution behaviour or rely on uncertain parameters. A data driven linearisation approach has been reported for analysing three phase unbalance in active distribution systems, which illustrates both the value and the dependence on representative training data [194].

7.3.2. Scalability and Computational Constraints

Scheduling and coordination problems grow quickly with the number of assets, time steps, and network constraints. Large scale EV charging scheduling with V2G services and reactive power management and solves it using the alternating direction method of multipliers to support tractable coordination. Such decomposed approaches are attractive for corridor control as they can split computations across sub-problems while enforcing feeder and aggregate constraints through coordination variables [195].

Real time feasibility is often limited by solver runtime and by the need to refresh decisions on rolling horizons. A real time warm start strategy for solving multiperiod alternating current optimal power flow problems has been presented, targeting moving horizon operation where each horizon must be solved on a seconds to minutes cadence. This shows how tracking approximate solutions can reduce per-interval computational effort when the optimisation is repeatedly re-solved with slowly varying data [196].

7.3.3. Communication and Cyber Physical Risks

Communication links that deliver metering and status signals can degrade coordination performance when data is missing. A recent forecasting study explicitly targets missing values and adversarial attacks, which are consistent with failure modes that occur in operational telemetry pipelines [193]. This motivates validation against degraded-data scenarios and the design of fallback operating modes that remain feasible under reduced observability [193]. It also motivates monitoring that distinguishes genuine behavioural change from artefacts caused by data loss [191].

Security threats expand the operational risk envelope because compromised control messages can cause unsafe actuation or violate grid constraints. An analysis of the OCPP identifies threats and proposes countermeasures, showing that even widely used interoperability protocols require explicit security engineering rather than implicit trust. For corridor operation, this implies that authentication, integrity, and audit logging should be treated as part of the control system design, not as afterthoughts. It also implies that incident response and fail safe defaults should be specified so that charging service degrades gracefully under detected anomalies [160].

7.3.4. Infrastructure, Economics, and Adoption Uncertainties

Economic uncertainty arises from both investment costs and ongoing operating costs, which interact with utilisation and reliability targets. Planning models that trade efficiency against equality have been proposed for public charging station siting, which formalises the tension between high utilisation locations and equitable coverage. In dynamic corridors, analogous trade-offs appear between electrifying high-demand segments and extending service to underserved areas, and these trade-offs depend on uncertain cost assumptions and business models. Because these costs are incurred over long horizons, coordination schemes should be stress-tested against adoption rates that deviate from planning assumptions [172].

Adoption dynamics are shaped by heterogeneous user behaviour, which affects local queuing and peak coincidence. An empirical study using a large dataset of charging events in Brussels shows that habitual charging patterns contribute strongly to energy consumption and utilisation while also increasing inconvenience, and that inconvenience can be spatially concentrated and unevenly distributed across user groups. The same study highlights that charging behaviour may change as the user base diversifies, which complicates both forecasting and roll-out strategies. For dynamic charging corridors, this motivates modelling demand uncertainty with behaviourally grounded scenario sets rather than relying on a single representative profile [197].

7.3.5. Research Directions for Coordination

Progress on integrated coordination will benefit from open data and reproducible evaluation. A city-level dataset for EV charging load forecasting and analysis has been released as UrbanEV, supporting consistent benchmarking of forecasting and scheduling methods across common inputs [198]. Complementary high temporal resolution datasets that include charging profiles and charging station occupancy enable joint modelling of energy demand and service availability at fine granularity [199]. Together, these datasets support standardised reference scenarios for testing coordination algorithms under realistic variability and utilisation patterns [198,199].

Another direction is to embed explicit fairness constraints or equity objectives directly into control and planning. A study on fairness and equity in EV charging with mobile charging stations formulates scheduling with equity considerations, which provides a template for corridor-level resource allocation under constrained capacity [200]. Planning work that balances efficiency and equality similarly illustrates how objective functions can incorporate public value alongside utilisation [201]. Benchmarks should therefore report both technical performance and distributional outcomes so that trade-offs are transparent [200,201].

7.3.6. Concluding Remarks

This section highlighted that moving from pilots to routine operation depends on treating dynamic charging corridors as part of a broader urban infrastructure system rather than as a stand alone energy transfer technology. Reliable deployment requires clear operational requirements for monitoring and maintenance, consistent interoperability across vehicles, chargers, backend platforms, and utility interfaces, and power quality and protection provisions that remain robust under highly variable use. It also reinforces that data handling, privacy, and accountability must be addressed as engineering requirements because operational decisions rely on telemetry and user level interactions. Taken together, these aspects show that technical performance is necessary but not sufficient, and that scalable implementation is ultimately determined by how well electrical design choices are integrated with operational governance and ecosystem coordination.

8. Conclusions

In this paper, we evaluated the impacts of DIPT for micromobility when deployment is constrained by EMF exposure compliance and by the practical misalignment that arises for two-wheeled vehicles. Section 2 framed dynamic wireless charging as corridor infrastructure that behaves as a distributed electrical load whose demand depends on traffic and availability, and it highlighted the role of segmentation control in making that demand manageable. Section 3 then positioned DIPT within the wider WPT landscape to clarify terminology and scope before the subsequent sections consolidated the core design and operational constraints into an implementation oriented perspective for smart grid control and charging optimisation.

A central conclusion from the infrastructure perspective is that corridor design choices and operational strategy are inseparable. The review shows that segmentation can improve efficiency compared with energising long tracks continuously, but it introduces switching and power fluctuation challenges that place stronger requirements on real time control. At the same time, the system level architecture involves explicit trade offs between long track and segmented approaches, and between supply bus choices, which jointly affect cost, converter complexity, controllability, and the ability to satisfy EMF exposure limits along the roadway. These interactions support treating DIPT as a whole system optimisation problem across vehicles, infrastructure, and electric energy supply rather than a standalone coupler design task.

At the electromagnetic design level, misalignment robustness and EMF leakage must be treated as joint objectives in pad and compensation selection for micromobility. Section 4 reviewed pad geometries and showed that different pads exhibit distinct combinations of misalignment tolerance and leakage tendencies, with examples ranging from early circular pads with limited misalignment tolerance and higher leakage, through to polarised geometries such as double D quadrature and bipolar pads that improve tolerance to horizontal misalignment via coil decoupling and flux distribution. The circuit analysis and compensation review further supports that efficiency and deliverable power remain governed by coupling and quality factor fundamentals, while basic compensation topologies are sensitive to variable conditions and therefore underperform away from their design point. Higher order multi resonant compensation networks are reported as a common route to extend acceptable operation across wider ranges of coupling and loading, at the cost of additional components and tuning complexity.

For compliant deployment, Section 5 emphasised that meeting EMF limits is challenging under misalignment and at higher power, which makes mitigation and shielding a first class design requirement rather than a late-stage add-on. The paper classified shielding strategies into active, passive, and reactive approaches. It highlighted that passive conductive shields reduce leakage through eddy current cancellation with reported efficiency penalties on the order of 1% to 2% due to Joule losses, while magnetic shielding such as ferrite can both reduce leakage and improve transmission by directing flux, albeit with increased material cost. The discussion also placed these design choices in the context of evolving standards and guidance, noting the role of SAE J2954 for light duty wireless charging and that SAE J2954/3 for DWPT was still under development at the time of the study.

Finally, the paper connected electromagnetic and infrastructure constraints to smart grid integration, concluding that effective DWPT operation requires layered scheduling and hierarchical control that can accommodate forecast uncertainty, coupling variation, and distribution network constraints. Section 6 summarised evidence that forecast error growth with lead time motivates rolling horizon approaches that combine day ahead planning with intraday updates, and that corridor operation benefits from hierarchical decompositions spanning network level allocation, corridor level set points, and vehicle level policies. It also identified enabling control ingredients, including real time estimation of mutual inductance and load for efficiency tracking, and explicit inclusion of thermal limits and battery ageing constraints when setting admissible operating points. Section 7 then extended the argument to urban deployment, highlighting the need for telemetry, state estimation, and secure interoperable control interfaces, alongside governance concerns such as privacy and accountability, and it outlined limitations for future coordination that include forecast drift, data gaps and adversarial risks, scalability constraints, and adoption uncertainty, with open data and fairness objectives positioned as practical research directions.