Electric vehicle as a service (EVaaS)

: Under vehicle-to-grid (V2G) concept, electric vehicles (EVs) can be deployed as loads to absorb excess production or as distributed energy resources to supply part of their stored energy back to the grid. This paper overviews the technologies, technical components and system requirements needed for EV deployment. Electric vehicles as a service (EVaaS) exploits V2G technology to develop a system where suitable EVs within the distribution network are chosen individually or in aggregate to exchange energy with the grid or individual customer, or both. The EVaaS framework is introduced and the interactions among EVaaS subsystems such as EV battery, charging station, load and advanced metering infrastructure is studied. The communication infrastructure and processing facilities which enable data and information exchange between EVs and the grid are reviewed. Different strategies for EV charging/discharging and their impact on the distribution grid are reviewed. Several market designs that incentivize energy trading in V2G environments are discussed. The benefits of V2G are studied from ancillary services, supporting renewables and environmental perspective. The challenges to V2G are studied from battery degradation, energy conversion losses and effects on distribution system perspective.


17
Due to climate change, fossil energy reserves and greenhouse gas (GHG) emission 18 concerns, efforts are currently going towards the transition to electric mobility [1]. There 19 exist several kinds of government policies designed to reduce GHG emissions and promote 20 the acceptance of electric vehicles (EVs), such as the UK Vehicle Scrappage Scheme (VSS) 21 [2], Car Allowance Rebate System (CARS) in the US [3], Zero Emission Vehicle (ZEV) 22 programs in California, China and EU [4], and the Corporate Average Fuel Economy 23 (CAFE) standards [5]. There are also government incentives designed to support policy-24 driven adoption of EVs, such as purchase rebates, tax credits, tax exemptions, and waivers 25 on charging and parking fees. However, due social obstacles, technical limitations and 26 cost premiums compared to conventional internal combustion engine (ICE) vehicles, EVs 27 have not been widely adopted [6]. The main types of EVs on the market are battery electric 28 vehicles (BEV), plug-in hybrid electric vehicles (PHEVs), extended range electric vehicle 29 (EREVs) and fuel cell vehicles (FCVs) [7]. For purposes of the article, all electric vehicles 30 with plug-in capabilities are collectively referred to as "EVs".

31
Range anxiety -fears over the distance EVs can travel between charges -is a major 32 technological barrier to large-scale adoption of EVs [8]. One of the factors influencing 33 range anxiety is the availability of EV charging points. Fears over lack of charging points 34 was identified in [9] as the biggest concern with regards to EV ownership. However, 35 with several government policies across the globe supporting the increased penetration 36 of EVs [10], there has been a rapid rise in the number of charge points. The remaining 37 driving range (RDR) -distance an EV can travel with the residual energy in the battery -is 38 While the potential benefits of V2G transition have been widely recognized, they 75 may not accrue without significant challenges. Impediments to V2G actualisation include 76 resistance from automotive and oil sectors, communication infrastructure needed for 77 information exchange, battery degradation, requirements for monetization of energy losses, 78 and technical, political, social and cultural obstacles. EV battery degradation costs happens 79 to be one of the main barriers to V2G transition. An additional issue is that energy flow 80 will become bidirectional and increasingly complex. Since the distribution grid has not 81 been designed for this purpose, service capabilities of V2G devices tend to be limited. 82 Conversely, bidirectional communications implementation in V2G infrastructures unlocks 83 new possible vulnerabilities. 84 The implementation impact of V2G technologies on the distribution system and 85 strategies for V2G interfaces for individual and aggregated EVs were studied in [19]. 86 The study in [20] discussed the operation of EVs and their impact on grid stability. The 87 methodology adopted for power flow under V2G scheme and challenges associated with 88 the commercial level adoption of V2G are described in [21]. The study in [22] inspects the 89 implementation challenges of EVs infrastructural and charging systems in conjunction with 90 several international standards and charging codes. The ancillary service potential of V2G 91 is presented in [23] and the potential impacts, challenges and future market penetration 92 capabilities of V2G technology is discussed. Several studies have been conducted to 93 evaluate the impacts of V2G integration on the utility. However, little attention has been 94 given to maximizing the full potentials of V2G from an EV-prosumer perspective. The 95 technologies, technical components and system requirements needed for the deployment 96 of EVs for grid-related services are reviewed in this article. The system architecture and 97 communication infrastructure for EV-enabled microgrids are introduced. The market 98 design and various mechanisms that motivate EV owners to participate in energy trading 99 are discussed. Optimization methods for EV charging/discharging, and benefits and 100 barriers to the deployment of EVs are presented and evaluated. EVaaS describes a system in which heterogeneous electric vehicles communicate 104 with the electricity grid to participate in grid-related services [24,25]. EVaaS exploits 105 V2G technology to develop a system where suitable EVs within the distribution network 106 are chosen individually or in aggregate to exchange energy with the grid or individual 107 customer, or both. It provides the opportunity for EV owners to benefit from an additional 108 revenue stream. EV owners can be incentivized to charge their EV batteries when the 109 energy generated exceeds demand, example, too much energy being generated from RESs 110 or off-peak hours. By contrast, EV owners who self-generate electricity from RESs or 111 connect to the grid to charge at low-demand, cheap tariff, can then market the excess or 112 unneeded energy stored in their EV batteries when energy costs are higher or during peak 113 demand. Thus, EVs can act as an energy reserve for the grid. The operation of EVaaS 114 system is a distinctive combination of EV, an energy management system and a service 115 contract which can deliver value by providing demand response services. 116 EV owners can go into a contract or agreement with the utility to make charging and 117 discharging controlled, coordinated and more predictable. The utility can offer lower energy 118 price to incentivize EV charging and battery insurance or maintenance service to incentivize 119 EV discharging in exchange for EV owners agreeing to charge and discharge the battery, 120 respectively, to meet the grid requirements. Based on this approach, the implementation of 121 a centralized charging and discharging solution becomes feasible, as well as the possibility 122 of maximizing system efficiency. Alternatively, EV owners can voluntarily participate in 123 EVaaS without making commitment with the utility. Different incentives can be offered 124 by the utility to motivate the EV owners to charge or discharge their batteries, depending 125 on the current demand and supply of the grid. The EV owners will individually consider 126 their charging or discharging options in a distributed fashion. Based on this approach, a 127 decentralized charging and discharging solution can achieve maximum system efficiency. 128

129
The architecture of an EVaaS system with interactions among subsystems is shown 130 in Fig. 1. The EVaaS system typically consists of EVs, loads (critical and non-critical), 131 charging stations, smart meters, power lines, communication infrastructure and microgrid 132 control centre or aggregators [13]. The system is remotely monitored and controlled by the 133 microgrid control centre using a supervisory control and data acquisition (SCADA) system, 134 while the subsystems communicate with each other through a communication network 135 to effectively carry out tasks and collectively achieve an objective. The communication 136 network facilitates the collection of necessary data from EVaaS subsystems and allows 137 the aggregator to efficiently optimize EV charging and discharging. However, this is 138 dependent on status monitoring and information update of both parked and moving EVs. 139 The information includes the current location of EV or where it will be in the next time 140 frame, battery capacity and state of charge (SOC). Using this information, the aggregator 141 can forecast or estimate the energy demand or supply from EVs within a specific region.  143

144
While ICE vehicles get energy from burning fossil fuel, an EV is powered from a battery. 145 Unlike the batteries used in mobile phones, laptops and other battery-powered electronic 146 devices, EV batteries are designed to achieve prolonged running time with high power and 147 energy capacity. Automakers have different EV passenger models, with battery capacities 148 boasting up to 100 kWh. The study in [26] investigates larger battery capacities (200 kWh 149 and above) for futuristic mobile usage and recommends subdividing the total battery unit 150 into mechanically separated containers. When an EV is being charged, chemical reactions 151 go one way and the battery absorbs power, these reactions are then reversed to produce 152 electricity when the EV is being discharged. Some of the EV battery technologies widely 153 deployed in the real world include lithium-ion (Li-ion), nickel-metal hydride (NiMH) and 154 lead acid [27,28]. Among them, Li-ion is the most common battery technology. Several 155 potential technologies that might be able to achieve better or comparable performance to 156 Li-ion batteries are in early stage of development. These include nickel-cadmium (NiCd), 157 sodium-sulfur (NaS), zinc bromid (Zn-Br) and aluminium-air (Al-air) [29]. The study in 158 [30] details the expected development in battery technology by 2030. As observed in Table 159 1, Li-ion is currently the widely accepted battery technology in the EV market [31][32][33][34][35][36][37][38][39][40][41][42][43]. This 160 can be attributed to its lightweight, high density, low self-discharge rate and prolonged 161 life features. The risk of explosion from overcharged cells and life cycle reduction from 162 undercharged cells are the disadvantages associated with Li-ion batteries.

163
The lifespan of the Li-ion batteries are typically estimated through calendar life and 164 cycle life. The calendar life is the retainable duration in terms of calendar years, indepen-165 dent of charge and discharge cycling [44]. The cycle life estimates the capacity retention 166 during continuous charge and discharge cycling before degrading significantly [45]. To 167 predict battery life, study its behaviour and simulate its performance under dynamic con-168 ditions, a battery model is required. The types of battery models widely studied include 169 electrochemical, experimental, mathematical and electric circuit models. The most accurate 170 is the electrochemical models, however they require in-depth knowledge of the chemical 171 reactions of batteries and complex set of equations that govern battery behaviour [46,47]. 172 Experimental models are based on experimentation to determine parameters associated 173 with battery behaviour [48,49]. Mathematical models consist of stochastic approaches that 174 capture battery behaviour [50,51]. Electric circuit models provide an equivalent representa-175 tion of battery characteristics. The basic equivalent circuit model consists of an open-circuit 176 An EV charging station, also called charge/charging point, electric recharging point 180 and electric vehicle supply equipment (EVSE), is an equipment that connects an EV to the 181 electricity grid. When EVs are plugged into a charge point, they can behave as loads or 182 generators to the electricity grid. The rapid growth in the global EV market has led to the 183 proliferation of charger designs, charging strategies, charging techniques and charging 184 networks. Techniques for charging and discharging of EVs, with emphasis on convenience, 185 simplicity, flexibility, and high efficiency, have become the motivation of current research in 186 academic and industrial communities [55]. The two main charging solutions for EVs are 187 conductive and inductive methods [56]. Conductive charging typically involves hard-wired 188 connection (electrical contact) between EV and the source of electricity, or EV and load in a 189 discharging scenario. While inductive charging works on the principle of inductive power 190 transfer (IPT), where magnetic field is used to transfer power across an air gap to a load. 191 Here no power cable, physical contact or human intervention is required. The exclusion 192 of cables, relatively low maintenance and autonomy for the driver has improved their 193 practicality in V2G systems [57]. Although there have been recent progresses in inductive 194 methods [58,59], conductive methods remains the most common solution [60]. The time it takes to charge EV batteries is currently longer than the refuelling time of 196 ICE vehicles to satisfy the same driving demands. EV charging rate is determined by how 197 many kilowatts the charge point can provide and the EV can accept -the higher the power 198 output, the faster the charge. Currently, the three main types of EV charging -representing 199 the power outputs, and therefore charging speeds -are slow, fast, and rapid. Slow charging 200 (level 1) can be done using existing electrical circuits. Level 1 chargers plug directly into 201 a standard 120 volt AC outlet, and are suitable for home or office use cases due to long 202 charging sessions. Unlike the previous case, fast charging (level 2) requires installation of 203 residential or public charging equipment. Level 2 chargers offer charging through a 240 204 volt AC plug, and are largely deployed in public places such as park and ride facilities, 205 shopping centres, car parks, airports and universities. While levels 1 and 2 charging are 206 adequate to serve the day-to-day needs of EV owners, long-distance or unplanned trips in 207 EVs need to be considered. Rapid charging (level 3) is available in a much higher voltage, 208 often charge using DC and can achieve 80% of charge in about 30 minutes, depending on 209 the capacity of the EV [20]. Level 3 chargers are often used as range extenders along major 210 roads and in urban environment to support drivers in urgent need.

211
To achieve a refuelling time that is comparable to that of ICE vehicles, EVs would 212 need charging stations with much higher power output. Extreme fast charging (XFC) is 213 an emerging technology with potentials to address the fast charge barrier and be truly 214 competitive to the ICE refuelling experience [61]. XFC stations should be able to support 215 charging at 400 kW, recharge an EV in less than 10 minutes and provide up to 200 additional 216 miles of driving [62]. However, there are still many barriers that need to be addressed 217 towards the standardisation and successful implementation of XFC. The technology gaps in 218 XFC topology is identified in [63]. The study in [62] investigates the XFC technology-based 219 charging infrastructure which will be necessary to support current and future EV refueling 220 needs.

221
With EV batteries becoming cheaper, automakers are equipping new model EVs with 222 more battery capacities. What used to be considered fast charging for a 24 kWh battery is no 223 longer fast when the battery size reaches 60 kWh or more. To address the changing market 224 environment and meet the expectations of EV stakeholders, CHAdeMO has developed an 225 ultra-high-power charging protocol enabling 500 kW charging and allowing for maximum 226 current of 600 A [64]. This new DC charging standard aims to support shorter and safer 227 charging using ultra-fast charging technology and is another step closer to achieving a 228 refueling time that can be competitive to the ICE refuelling experience. The background 229 and technical challenges of harmonising this new DC charging standard and its impact 230 on the global EV charging infrastructure outlook is presented in [65]. EV charging station 231 characteristics are presented in Table 2.

232
There is a growing interest in the integration of solar photovoltaic into the EV charging 233 system. Solar-powered EV charging stations can help reduce GHG emissions, charging 234 costs and the impact of additional load on the grid. Different technologies for solar-powered 235 EV charging and their deployment in the real world are discussed in [66]. While solar-236 powered charging stations bring opportunities for EVs, the environment and the grid, the 237 uncertainty and intermittent nature of solar power raises challenges in timely utilization. 238 The concentration of electricity output during the daytime limits the contribution of solar 239 power in meeting a large fraction of typical energy demand. Thus, a grid connection or 240 battery bank is necessary to guarantee effective operation of the solar-powered charging 241 station.

243
Load indicates an electrical component (device or machine) or a collection of equip-244 ment that consumes electrical energy. Based on demand response management, loads in 245 a building can be divided into two categories -controllable and non-controllable. Non-246 essential loads that can be deferred or interrupted for a limited period of time with minimal 247 effect on convenience are considered as controllable loads. These include air conditioners, 248 water heaters, dish washers, clothes washers, clothes dryers and EVs. While loads such as 249 lighting, cookers, microwave ovens and other plug loads are considered as non-controllable 250 loads. Building loads in a power system can be categorized into two groups: critical and 251 non-critical loads. Critical facilities which need to be operating during power outages 252 such as hospitals, care homes, residential houses with life support equipment, water and 253 communication infrastructure, control centres, data centres, evacuation centres, emergency 254 shelters, police and fire stations, military bases and airports are considered as critical loads. 255 Non-critical loads are not essential for human health and safety, and they would generally 256 be loads not categorized under critical loads.

257
Load profile represents the pattern of energy usage of a consumer, both daily (on-peak 258 and off-peak) and seasonally (summer and winter). Modern girds are usually known to be 259 based on the behaviour of consumers to manage the load and supply in the distribution 260 network, where reliable and efficient delivery of electric services are dependent on the load 261 profile. Load forecasting is the predicting of power or energy needed to meet the short-262 term (up to a day), medium-term (a day up to a year) or long-term (over a year) demand. 263 The load profile can be forecasted using techniques such as similar-day approach, time-264 series method, regression method, neutral networks, fuzzy logic, knowledge-based expert 265 systems, adaptive load forecasting, iterative reweighted least-squares and exponential 266 smoothing [67,68]. The accuracy of forecasting is of significant importance for the planning 267 and operation of electric utilities. . AMI provides time stamped information and establishes two-way commu-276 nication between smart meter and the utility. With two-way communication, many services 277 that were nearly impossible to implement without smart metering are now applicable. 278 These services include power outage detection, power quality measurements and power 279 flow monitoring. The power flow monitoring information is important as it enables the 280 utility to react rapidly on changes in consumption levels.

281
Unlike traditional meters, smart meters are self-reading meters which give more 282 detailed information on energy usage in near-real time. The smart meter stores various 283 types of data, such as executed or received commands, event logs, time of use tariffs and the 284 firmware. The smart meter has either an Ethernet interface to connect to wireline services or 285 a direct interface to a wireless service. Data of the smart meter is collected and transmitted 286 to the utility using wide area network (WAN) connection. Smart meter connections to 287 home area network (HAN) are fundamental to residential or building management and 288 allow appliances to respond to time-based pricing signals or other triggers carried over 289 the grid. Key features of smart meters include load limiting and balancing for demand 290 response applications, remote command (turn on/off) operations, power outage detection, 291 time-based pricing, power quality monitoring (active and reactive power, phase, voltage, 292 current and power factor) and power consumption measurement for utility and consumer. 293

294
EVaaS communications enable data and information sharing among EVaaS subsystems, 295 and it consists of communication infrastructure, such as wired and wireless networks, and 296 processing facilities, such as data centre and cloud computing. The smart meter facilitates 297 the transmission of data through commonly available fixed wired and wireless networks, 298 such as Fixed Radio Frequency, Power Line Communication (PLC), Broadband over Power 299 Line (BPL), as well as public networks such as cellular, landline and paging. Consumption 300 data from the smart meters are received, stored and analyzed to provide useful insights to 301 the utility. The smart meter also responds to remote command from the utility. V2G communications enable EVs and the grid to interact and exchange information. 304 This is crucial to solve problems related to V2G management. By enabling real-time and 305 reliable communication between EVs and the grid, energy resources distributed over large 306 geographical areas can be managed effectively to enhance the overall system performance. 307 The communication network in V2G systems must be bidirectional to ensure substantial in-308 formation exchange [72]. The system needs information control that is aware of EV location, 309 battery capacity, battery efficiency, SOC, energy price and transportation cost. Transmitting 310 this information and receiving commands over efficient bidirectional communication links 311 is an essential requirement for successful V2G integration.

348
EVs will be an integral part of the modern era of low latent wireless communications 349 that promises to provide low-latency and ultra-reliable transmissions [89]. 5G network aims 350 to support the deployment of vehicles to everything (V2X) technologies, carter to explosive 351 ever-growing data traffic and enable users to indulge in gigabit speed immersive services 352 capable of extremely low response time, regardless of geographical and time dependent 353 factors. V2X technology will facilitate autonomous energy trading, where EVs in parking 354 lots can autonomously charge and discharge their batteries, while self-driving EVs can be 355 routed to appropriate charging stations to participate in EVaaS activities. EVaaS requires 356 much shorter network response time and big data analytics to enable rapid reactions and 357 intelligence across the network. There is no doubt that large amounts of data will be 358 generated by sensors, smart meters, cameras, maps, on-board electronic control unit and 359 battery management system of EVs, databases and more.

360
EV data which can be used to monitor, analyse and make decisions relating to charg-361 ing/discharging, energy trading and range estimation mostly come from the on-board 362 electronic control units and battery management system. EV data can be categorized as into 363 three types, namely standard, historical and real-time data. Standard data include technical 364 specifications from manufacturer and the usual driving time to destination according to 365 Google Map. Historical data include battery management system logs showing start and 366 end times of journeys, as well as SOC information like connect and disconnect times of 367 charges and discharges. Real-time data include SOC of EV battery, GPS location of EV and 368 data closely related to emergency issues, such as unplanned road closures and real-time 369 traffic/weather condition. Internet of things (IoT) enables the recording and transmitting 370 of detailed EV data in on-board computers or cloud computing infrastructure. In the 371 context of the smart grid, IoT is built by integrating internet-connectivity into all grid 372 subsystems, connecting them in intelligent networks, and utilizing data analytics to extract 373 meaningful and actionable insights from them [90]. Cloud computing provides the virtual 374 infrastructure for data collection, analysis and visualization in the current architecture of 375 IoT.

376
During mobility, autonomous EVs can generate data up to thousands of gigabytes, 377 where the volume of data is dependent on the variety of sensors and cameras used for 378 autonomy. Data generation is not expected to be enormous during EVaaS, but the various 379 sensors collecting data from grid, EVs, smart meters, charging stations and drivers will need 380 solutions from the big data domain. Effective integration of data from different sources 381 is possibly an enormous task; however, with the right tools and solutions from the big 382 data domain, valuable insights can be drawn [91]. Prioritizing the intercepted information 383 is essential and means of prioritization should be investigated, as decision-makers can 384 only digest a certain amount of information and draw insights based on it. Furthermore, 385 the processing of EV data by the aggregator make EVs vulnerable to security and privacy 386 concerns, which are yet to be addressed in the domain of big data analytics for EVs.

388
Large-scale deployment of EVs will result in higher demands on distribution systems, 389 which were not originally designed to withstand a high level of EV penetration [92]. With 390 the expected rise in penetration levels, future EV charging scenarios could be accompanied 391 by numerous challenges. EV charging profile has an effect on the distribution system. The 392 increasing number of charging EVs adds extra load on distribution systems, which can 393 drastically impact electricity grid stability. These impacts include power quality issues, 394 phase imbalance, transformer degradation and failure, higher system losses and increased 395 operational cost [93]. We review different charging strategies and their impact on distri-396 bution systems. This review classifies the EV charging strategies into uncoordinated and 397 coordinated strategies. 398

399
Uncoordinated charging describes a scenario where the EV batteries either start charg-400 ing immediately EVs are plugged into a charge point or after a user-adjustable fixed delay, 401 and continues charging until the batteries are completely charged or unplugged. In uncoor-402 dinated charging, EV charging is presumably at Level 1 with no coordinative control action. 403 Thus, its impact on distribution systems is primarily driven by the stochastic behaviour 404 of the EV user [94]. Load at peak hours tend to increase with uncoordinated charging 405 operations. An increase in peak load can cause severe network stress and overloads in 406 the local distribution grid. Random uncoordinated charging may lead to increased power 407 losses, overloads in transformer and cables, poor voltage profiles, degraded power quality 408 and an overall reduction in the reliability and economy of the distribution grid [95]. 409 An analysis into the impacts of random uncoordinated EV charging on the perfor-410 mance of distribution transformers was carried out in [96]. Results revealed that even under 411 low EV penetrations, transformer load surging and voltage deviations were significant. 412 Load growth on transformers for low penetration level of 17% to 31% showed a 37% to 74% 413 increase in transformer load current. In [97], a test model using household load profiles 414 for Belgium reports voltage deviations close to 10% during evening peak for a penetration 415 level of 30%. A typical UK distribution system is studied in [98] to determine the impact 416 of uncontrolled domestic charging on the distribution system. Results show up to 17.9% 417 increase in daily peak demand at 10% penetration rate of EVs, while the peak load would 418 increase by 35.8% at 20% EV penetration. In [92], the impact of EV penetration on existing 419 electricity distribution infrastructure was analysed using data for the Netherlands. Results 420 show that at 30% EV penetration, uncoordinated charging would increase national peak 421 load and household peak load by 7% and 54%, respectively, which may exceed the capacity 422 of the distribution system. The utility operator will have to increase peak generation if the 423 load exceeds peak capacity. The cost of additional generation capacity during peak period 424 is then passed on to EV owners. In [99], uncontrolled charging was shown to cause a 22% 425 increase in the monthly energy bill, even at just 10% EV penetration. 426 Some energy suppliers in the UK offer EV tariffs to help reduce peak demand, redi-427 recting it to off-peak times [100]. The two-rate tariff, which offers cheaper rates during 428 off-peak times (overnight), is designed to encourage EV owners to charge when the energy 429 demand is low, and generation is mostly base load. The study in [98] showed that overnight 430 charging increases off-peak energy consumption but it had no impact on the daily peak 431 load. In [92], off-peak charging at 30% penetration rate of EVs was reported to cause a 432 20% higher, more stable base load and no additional peak load on the national grid. Thus, 433 with the introduction of off-peak charging, additional generation capacity would not be 434 required for low EV penetrations. 435

436
Coordinated charging is being investigated as an alternative and possible solution to 437 random uncoordinated charging and its associated problems, respectively. EV charging 438 is most likely at Levels 2 and 3 in coordinated charging [19]. By utilizing the control and 439 bidirectional communications infrastructure of smart grids, smart coordinated charging and 440 discharging of EVs can reduce transformer load surges, line currents, voltage deviations 441 and daily energy costs [95,97,101]. It can also provide efficient energy usage [98] and 442 flatten the voltage profile of a distribution node [102]. Incremental distribution network 443 investment and energy losses costs can be avoided with the implementation of smart 444 charging strategies. Results in [103] showed the possibility of avoiding up to 60%-70% of 445 the required incremental investment with smart charging. The results of the study in [104] 446 reveal that coordinated charging of EVs minimizes system losses and improves voltage 447 regulation in the distribution grid.

448
Smart charging and discharging where EVs charge their batteries from RESs and 449 discharge them during peak demand is reported to offer the best possible utilization of RESs 450 for cost and emission reductions in the smart grid [105][106][107]. It can improve operational 451 performance in stand-alone operation mode and increase the quantity of RESs installed in 452 islanded microgrids [101]. A control strategy was implemented in [108] to coordinate the 453 charging and discharging of EVs to support a grid with high penetration of wind energy. 454 The obtained results showed that the total power imbalance in the system was significantly 455 suppressed. In [109], coordinated EV charging and discharging was implemented on an 456 Australian distribution grid with solar power generation. The proposed control method 457 was able to cope with solar power uncertainty and efficient in improving grid performance, 458 reducing energy cost and mitigating grid imbalance.

459
Coordinated charging can be categorized into two types, namely centralized and 460 decentralized approaches. In centralized approaches, EV charging is directly controlled by 461 a centralized unit (microgrid control centre or aggregator). Centralized approaches offer 462 full support for ancillary services. However, only a limited number of charging EVs can 463 be accommodated. Another drawback of this approach is that it involves higher order 464 complexity. In decentralized approaches, the power of decision-making with regards to 465 EV charging is distributed among individual EVs. The charging behaviour of EVs can be 466 directly influenced by a price signal. Decentralized approaches offer greater scalability 467 and lesser computational complexity. Considering EVs only have to exchange limited 468 information with the aggregator, their privacy is preserved [110]. A drawback in this 469 approach is the need for EVs to collect and store the trip history [111]. Compared with 470 centralized approaches, the decentralized approaches are more scalable, flexible, and 471 enables EV owners to partake in the decision-making process of EV charging.

473
Advances in V2G promises unprecedented improvements in operational efficiency. 474 This unlocks the possibility of prosumer and consumer participation in energy trading. 475 Consumers equipped with rooftop solar power system can emerge as EV-prosumers and 476 self-supply during peak period or power outages using V2H integration [112]. This can 477 lead to reduced household energy costs, maximum utilization of solar power generation 478 and minimum dependency of domestic loads on the grid. In an EVaaS energy trading 479 scenario, a grid manager (aggregator) has a demand target and manages individual or 480 aggregated EVs to fulfill the demand. Energy trading can be categorized according to the 481 market design. We review two types of energy trading in V2G environments: traditional 482 bilateral energy trading used in conventional energy industry and futuristic energy trading 483 with increased distributed influence, based on blockchain technology.

485
Traditional energy industry has operated on a centralized energy trading model for 486 decades. In centralized approaches, one user or a centralized controller dictates to a group 487 of users, while acting collectively as one entity. The central controller, acting as an energy 488 broker, is assumed to know all the information about trading entities and tries to match 489 demand and supply. The appropriate framework to employ in a centralized market would 490 be single objective optimization models such as swarm, stochastic or convex optimization, 491 or social welfare maximization [113]. Decentralized energy markets enables scalability and 492 competitiveness amongst self-interested EVs compared to their centralized counterparts. 493 Thus, it is important to investigate distributed economic approaches which incentivizes 494 energy trading between EVs and the grid [114]. 495 Auction is a promising market mechanism used to sell (forward auction) or buy 496 (reverse auction) energy in smart grids, with the aggregator acting as auctioneer. In a 497 scenario where energy trade is incentivized, buyers pay a discount in forward auction 498 while sellers receive a premium in reverse auction as compared to the clearing price. The 499 amount of energy to be traded and the final price to be paid is the outcome of the auction. 500 Based on the final payment, there are different auctions schemes, namely first price auction, 501 second price auction and uniform price auction [115]. Utility-maximizing bidders could 502 misrepresent their valuations (individually or collusively) by not bidding truthfully, which 503 could harm the fairness and efficiency of the trade. Vickrey-Clarke-Groves (VCG) auction 504 is effective in ensuring the properties of truthfulness [116][117][118]. An auction scheme that 505 enables EVs and batteries in swap stations to trade energy is proposed in [119]. A double 506 auction-based approach for enabling EVs to trade their excess energy to the grid is studied 507 in [120]. Double auction mechanism has also been studied in [121,122] for energy trading 508 in a two-layer V2G architecture, made up of grid-aggregator and aggregator-EV layers. 509 In [123], a group-selling strategy for V2G demand response management is implemented 510 through a two-layer reverse auction.

511
Game-theoretic approach is another promising solution which has been used in numer-512 ous applications to study the interactions among self-interested and independent agents. A 513 game is made up of three essential elements: a set of players, a set of actions (strategies) 514 and a set of payoffs (utility functions). The payoff obtained by the players is the value of the 515 game. One major strategy for game theory is the Nash equilibrium, where no player has 516 any incentive by unilaterally deviating from its strategy. Based on players coordinating or 517 competing with themselves, games can be categorized into two types, namely cooperative 518 games and noncooperative games. Noncooperative games are appropriate in distributed 519 energy trading scenarios between competitive trading entities. While cooperative games 520 are ideal in scenarios where trading entities cooperate with the aid of communication 521 networks, in other to optimize the efficiency or social welfare of the collaborators. In [124], 522 an analytical framework that captures the interactions between a smart grid and EV groups 523 is modelled using a noncooperative Stackelberg game. The interactions and energy trading 524 decisions of geographically distributed storage units, such as EVs, is studied in [125] using 525 a noncooperative game. An incentive-based V2V game theoretic approach that captures 526 the coordination strategies of EVs and battery swapping station aggregators is modelled in 527 [126]. In order to incentivize EV participation, each battery swapping station aggregator 528 implements a noncooperative game among the EVs in its range through a smart pricing 529 scheme. Collaborative and non-collaborative approaches which considers energy trading 530 and residential load scheduling with EVs is proposed in [127]. The collaborative approach 531 is based on social welfare maximization, while the non-collaborative approach utilizes a 532 noncooperative game. Besides auction and game theory, incentive-based approaches such 533 as pricing, bargain and contract theories, which are able to study the interaction between 534 self-interested participants and improve the efficiency of energy trade, have been widely 535 deployed [114].

537
With the rising penetration of EVs, satisfying the ever-increasing energy demand of 538 V2G applications remains a challenge for the distribution system. To address this challenge, 539 recent studies have exploited blockchain technology for energy trading in V2G environ-540 ments. Blockchain technology enables increased distributed influence in the distribution 541 system, while preserving privacy and maintaining transparency and system security [128]. 542 This will improve the flexibility of the conventional energy market, enable a consumer-543 centric energy market and support prosumer participation. EVs, acting as prosumers or 544 consumers, will be able to trade energy in a peer-to-peer (P2P) manner without third-party 545 intervention as shown in Fig. 2.

546
The application of blockchain technology for EV-enabled energy trading in smart grids 547 is briefly discussed in [129]. A decentralized security model based on the lightning network 548 and smart contracts is proposed in [130] to protect energy trading transactions between 549 EVs and charging stations. A localized P2P energy trading model based on consortium 550 blockchain is proposed in [131]. The model uses iterative double auction mechanism to 551 maximize social welfare of charging and discharging EVs. Consortium blockchain has 552 also been exploited in [132] to propose an energy trading model applicable in general 553 scenarios of P2P energy trading. The model uses Stackelberg game to maximize economic 554 benefits. Blockchain technology was applied in [133] to establish a trusted environment for 555 energy trading between EVs and critical loads and a prototype was developed for remotely 556 monitoring of energy trading activities.

557
While P2P energy trading is promising, one of its major impediments is regulation. 558 Currently, decentralized energy trading is prohibited by regulation in the UK and some 559 other EU countries, however this could change in the future. Business owners or individuals 560 who generate electricity are limited to use it on site or sell directly to the utility grid 561 for a nominal price. This poses a major barrier to P2P energy trading which enables 562 direct trade between prosumers and consumers, instead of selling to and buying from 563 the utility grid, respectively. Ideally, the authorization of P2P energy trading will create 564 a competitive energy market, allow prosumers generate revenue on their excess energy 565 and consumers obtain cost effective energy. It is expected that energy prices will drop as 566 a result of eliminating the middle man and more individuals incentivized to partake in 567 microgeneration. V2G systems can maintain the balance between demand and supply in electricity grids 576 by injecting power. While the supply capacity for individual EV is small, an aggregated 577 capacity can be significant to provide value to the grid. Aggregators are expected to collect 578 EVs into a group to create a more desirable, larger electricity generation capacity for the 579 utility. For example, by simultaneously discharging their batteries, aggregated EVs will be 580 able to provide additional power required by commercial building during peak demand, 581 acting like a spinning reserve power generation source in the existing distribution system. 582

583
V2G systems are capable of regulating voltage and frequency in electricity grids. Fre-584 quency regulation provides active power support in the electricity grid. The exact amount 585 of electricity being used needs to be matched by generation, if there is an imbalance it 586 can affect the frequency of the electricity grid. For example, if electricity demand is more 587 than supply, frequency will fall. If there is too much power being generated in relation to 588 demand, frequency will rise. The frequency will not stabilize until the system in balanced. 589 In the UK, anything just 1% above or below the nominal frequency of 50Hz risks damaging 590 electrical equipment and infrastructure, including appliances of end users. Currently, fre-591 quency regulation is achieved mainly by turning on fast-responding generators to increase 592 power generation, which is costly. Alternatively, fast charging and discharging rates of EV 593 batteries can help to increase the load demand and generation, respectively. This makes 594 V2G a promising alternative for frequency regulation [134,135]. Voltage regulation provides 595 reactive power support in the electricity grid. Reactive power can be controlled by selecting 596 the current phase angle to provide inductive or capacitive action. The consumption of 597 reactive power is mostly through inductive load, which requires the addition of capacitive 598 reactive power to balance the demand. Traditionally, reactive power support is injected 599 at the transmission or distribution grid stage, with no involvements from the end-users. 600 However, with increased EV penetration, V2G can provide the necessary reactive power 601 support to the grid. V2G systems are capable of levelling peak loads in electricity grids. Peak load shaving 604 is achieved through a control strategy that manages EV charging and discharging. In this 605 technique, controllable and aggregated EVs can charge when demand is low (off-peak 606 hours or overnight) and discharge during high demand (peak hours). In scenarios where 607 the generation capacity does not match the peak demand, several problems such as voltage 608 fluctuation, instability and total blackout could possibly occur. Therefore, by shaving peak 609 load, the reliability and stability of the grid is maintained and supply shortage is mitigated. 610 Previous studies had proposed different peak shaving and valley filling techniques through 611 V2G to alleviate the generation-demand imbalance [104,136,137]. This function of V2G can 612 provide economic benefits as it limits the need to use high-priced peak generators.

614
V2G systems are capable of restoring supply during prolonged grid outages and can 615 improve the grid capabilities to withstand unexpected contingencies. Power systems must 616 not only operate reliably in response to foreseeable contingencies, but also resilient to high-617 impact low-probability events [138]. Keeping critical loads operating during prolonged grid 618 outages is a key resilience feature for mitigating its consequences [139]. Thus, following 619 an unexpected system failure, fast recovery is very essential to enhance grid resilience. 620 EVs, as mobile power generation and storage resources, can distribute the existing energy 621 produced or stored in the local region. Aggregated EVs will be able to provide backup 622 power required by critical loads during a blackout. In a scenario where the distribution 623 network is partly damaged during an extreme event and regular supply cannot reach 624 critical loads, EVs can be deployed to individual locations of the critical load to restore 625 supply [133]. For instance, 630 peak solar radiation precedes peak demand by a few hours -solar peak power is at noon, 631 peak demand is typically between mid-afternoons to late afternoons. On the other hand, 632 the stochastic nature of wind power is due to unpredictable variations in wind speed. Wind 633 generation fluctuates and cannot be turned up when energy demand increases, leading 634 to imbalances. At low scale penetration, existing mechanisms for managing supply and 635 demand fluctuations can handle the intermittency of renewable energy. However, at high 636 levels of penetration, additional resources are needed to match the fluctuating supply to the 637 already fluctuating demand. If there is too much energy being generated from renewable 638 sources, generation from conventional power plants must be curtailed to restore balance. 639 EVs can help match generation and consumption by charging and discharging so the utility 640 does not consider decreasing the power output. Thus, V2G increases the flexibility of the 641 grid to support intermittent renewables. 642

643
V2G systems can offer societal benefits regarding climate change, GHG emissions 644 and air pollution. Climate change benefits come about via controlled charging (or peak 645 shaving) to limit usage of high carbon energy sources, decarbonisation of the ancillary 646 service market and electrification of the transport sector. The carbon benefits of V2G are 647 mostly dependent on the electricity generation mix of the grid. In electricity grids with high 648 polluting sources, V2G providing ancillary services has potentials to increase total carbon 649 emissions [143]. EVs cannot guarantee decarbonisation since they are not generation. Despite the many benefits V2G offers, a major concern has been its impact of on the 660 degradation of EV batteries. V2G operation imposes more use (and stress) on EV batteries 661 compared to daily driving, which likely accelerates the aging of EV batteries [147]. This can 662 be associated with the increase in charge cycle, where a charge cycle is a complete charge 663 and discharge process on the battery. EV battery usage is limited to a fixed number of cycles, 664 over time the capacity (amount of energy that can be stored or extracted from the battery) 665 degenerates significantly. Recent studies have found that degradation costs are a substantial 666 barrier to the grid, while others found degradation to be minimal. Determination of V2G 667 impact on EV battery degradation is still at the research stage with recent studies arriving at 668 contradictory conclusions. While some studies found degradation costs to be a substantial 669 barrier to V2G [148,149], others found degradation to be minimal [150,151]. Nevertheless, 670 even in the best-case scenario, participation in V2G operation accelerates battery capacity 671 degradation beyond what is required to satisfy the driving demands [152]. Consequently, 672 EVs may be expected to undergo battery replacement multiple times over their service life. 673 Thus, V2G influences the frequency of battery replacement and associated costs. 674

675
In V2G systems, energy losses occur between the grid connection point and the EV 676 battery. Each time EVs are charged or discharged, energy losses occur in the EV and its 677 supporting electrical infrastructure such as charging station, breakers and transformer. Each 678 stage of storage, conversion and transmission contribute to the losses. This is considering 679 the efficiencies of system components such as EV battery, power electronic unit, charging 680 station, breaker panel and transformer. The impact of different levels of EV penetration on 681 the distribution network was studied in [103]. Under studied conditions, obtained results 682 showed that energy losses could increase up to 40% with the respect to the level of EV 683 penetration. In [153], energy losses from electricity grid to EV battery and back to the grid 684 were measured experimentally. The measured total one-way losses were up to 36%, under 685 studied conditions. Although studies have reported round-trip losses for EVs and related 686 V2G infrastructure, efficiency values are case dependent and will differ among EVs and 687 electrical circuits. Nevertheless, they can serve as a reference point for future studies on 688 economic analysis of V2G. The increasing penetration of EVs is likely to have considerable impact on the dis-691 tribution system. Since the distribution grid is still focused on conventional design and 692 operational rules, service capabilities of V2G devices tend to be limited. The charging 693 and discharging of EVs introduces a change in the overall load profile of the distribution 694 system. Uncontrolled EV charging adds to the pre-existing peak load, especially during 695 fast charging. The load demand is centralized at the fast charging station and fast charging 696 mainly occurs during the daytime, allowing EVs draw high power larger than a regular 697 household load [154]. The interconnection of fast charging stations with the grid might 698 create negative impacts on the distribution system [155]. Fast charging of EVs could result 699 in detrimental effects on distribution transformers, lower operational efficiency of the distri-700 bution network equipment and rise in energy losses. Fast charging also has adverse effects 701 on the voltage profile and power quality of the network. Additional EV load increases 702 transformer temperatures, which contributes to insulation breakdown and may decrease 703 the life expectancy of the transformer [156]. The impact of different penetrations of EVs on 704 a residential distribution transformer was studied in [157]. This revealed that high pene-705 tration of EVs can have significant impact on the electricity grid, particularly in scenarios 706 with uncoordinated charging. In [158], an investigation was carried out to evaluate some 707 of the effects of EV deployment on existing distribution network. This revealed that high 708 deployment of EVs could result in supply and demand matching and statutory voltage 709 limits violations, as well as voltage imbalance and power quality problems. In order to 710 help the distribution circuit to accommodate EV penetration, a demand response strategy 711 is proposed in the context of a smart distribution network in [159]. Thus, the effect of V2G 712 on the distribution network is greatly influenced by the charging strategies and vehicle 713 aggregation [160].

715
V2G technology enables EVs deployment as loads to absorb excess production or as 716 generators to feed-back surplus energy to the distribution grid during peak demand or 717 system failure. This paper has presented an EVaaS system where EVs, individually or as 718 part of an aggregation, can provide service to the grid, individual customers, or both. The 719 EVaaS system architecture and interactions among EVaaS subsystems such as EV battery, 720 charging station, load and advance metering infrastructure has been discussed. The infras-721 tructure and processing facilities for bidirectional communications in V2G environments 722 has been explained. Several potential battery technologies that might be able to match the 723 widely accepted Li-ion batteries were highlighted. Methodology to enhance grid resilience 724 through building load categorization was introduced. The impact of coordinated and 725 uncoordinated and fast charging on the distribution grid was discussed. The challenges 726 associated with timely utilization of solar-powered EV charging stations was examined. 727 The centralized structure of conventional energy markets does not allow the full potential 728 of V2G to be realized. The current energy market is not consumer-centric and does not 729 support prosumer participation. Policy change, supporting infrastructure and incentives 730 would play a huge role in maximizing the market opportunities presented by V2G.