Project Topics
Get Complete Project Material File(s) Now! »
EV status in the grid
V2G-able EVs face great diculties regarding their connection requirements and legal status as exibility providers. Connection requirements can be burdensome, as V2Gcapable EVs need to comply with requirements both as producers and consumers, as well as administrative procedures to declare and allow distributed sources to participate as exibility providers. Legal status of V2G installations should also be claried and aligned with that of storage, with taris and charges that prevent double taxation.
Regulators, system operators, and EV and EVSE manufacturers need to work to standardize interconnection requirements to ensure system and end-user safety, while easing administrative procedures. For example, the French regulator issued a series of recommendations regarding the interconnection requirements, mainly for the denition of the decoupling protection5, as well as simplication of administrative procedures [131]. In 2019, Delaware state passed legislation that dened the perimeter of V2G, dened clear interconnection procedures (adopting SAE J3072 safety for on-board bidirectional chargers [132]) and allowed net-metering to provide a level-playing eld with utility-scale storage [133]. These measures have been suggested to other states as well [134].
Interactions with grid operators
An important aspect is how the dierent stakeholders interact along the exibility value chain. There are interactions between exibility providers and exibility customers, in this case EV users and DSOs respectively, and interactions between DSOs and TSOs as potential exibility customers, where their level of coordination and cooperation will aect how local exibility is used.
EV users-DSO interaction
DSOs can procure exibility from end-users directly or indirectly. As mentioned in Section 4.1, DSOs can procure exibility using dierent solutions. By using direct obligations (grid codes) for exibility provision or contract arrangements (such as interruptible contracts), DSOs will directly interact with EV users acquiring permission to directly control the EV charging process.
On the other hand, market-based procurement via exibility platforms usually needs an aggregator that would gather multiple exibility resources. This is currently the case for ancillary services and BRP energy arbitrage as done by existing EV aggregators. It could be expected that a growing number of EVs will become associated to an aggregator’s program, therefore likely to meet communication and control requirements for the smart charging process. This will allow the provision of market-based exibility services to DSOs by EV eets.
Positioning with respect to the state-of-the-art
In the remainder of this thesis we will center on two of the research gaps identied: how EV user charging and driving patterns aect EV integration, and what mechanisms/ frameworks are emerging for exibility procurement at the distribution level. In Chapter 3, we will analyze and model the charging behavior of EV users (i.e., how often they plug-in their vehicle). Then, in Chapter 4 we will study how local mobility patterns and the spatial distribution of EVs aect EV integration and the coupling with renewable energies. Finally, in Chapter 5 we analyze emerging exibility mechanisms at the distribution level, with a particular focus on long-term tenders, and how EV aggregators can participate in them.
The Electric Nation trial
We used real-world data from the Electric Nation project to calibrate the plug-in decision module. The Electric Nation project was a large-scale smart charging trial in the UK that ran from 2016 to 2018. The full dataset contains information for 153621 charging sessions, including starting time, ending time, and energy consumed for each session, by 601 unique users with a variety of BEVs, PHEVs, and REX marks and models.
To calibrate the model, we only considered BEV users that stayed in the trial for more than 3 months. After cleaning the dataset, we obtained 52822 charging sessions for 265 unique users, encompassing a wide range of EV brands and battery sizes, as shown in Figure 3.4 and Table 3.2. Two distinct EV groups can be observed, the rst composed of small EVs with battery sizes between 20{35 kWh, and a second group composed of large EVs with battery sizes around 75 kWh.
Plug-in decision model calibration
We calibrate our model to match the average plug-in frequency found in the Electric Nation trial for three cases representative of small (25 kWh battery), average-size (50 kWh), and large (75 kWh) EV eets. Using the EV model presented in Section 3, a simulation for 1,000 EVs and 12 weeks was carried out with a range anxiety factor () of 1.5, and for levels of the plug-in parameter () varying from 10-2 to 102. Each simulated EV has a daily distance sampled from a lognormal distribution, as in [40]. Table 3.4 shows the main parameters of the simulations used to calibrate the model, and Figure 3.6 shows the average frequency of charging sessions for a sweep of the parameter for the three cases and the selected value to match the observed data. The selected parameter ranges between 0.89 for small EVs, to 1.31 for the average eet.
Impact of non-systematic plug-in behavior on EV grid integration studies
We evaluated the impact of considering non-systematic plug-in behavior in EV grid integration studies. For this purpose, we analyzed two aspects: the impact of EV charging in power systems through the peak load created by EV eets, and the exibility potential to assess the time and accessible storage capacity that EV eets can use for smart charging or V2G-based exibility services. Simulations using the EV model were carried out for a eet of 10,000 EVs and combinations of battery size, charging power, plug-in behavior, and charging strategies. We considered three battery sizes, i.e., small (25 kWh), medium (50 kWh) and large (75 kWh) in line with current trends, and three charging power levels, i.e., 3.7 kVA and 7.4 kVA, re ecting standard single-phase chargers, and 11 kVA three-phase charger, all with a 0.95 power factor. Charging choices were considered via systematic plug-in (i.e., every day) and three non-systematic plug-in behaviors, an average case given by the calibration with the Electric Nation trial (=1.31), a high plug-in case (=3.4), and a low plug-in case (=0.5) to account for dierent charging choices. Two charging strategies were analyzed: uncontrolled charging, where EVs are charged as soon as they are plugged in, and smart charging, where EVs charge during an o-peak period between 10pm and 6am. Finally, arrival and departure times at the charging locations are given by joint probability distributions derived from the Electric Nation trial (probability distributions shown in 3.8).
Table of contents :
Acknowledgements
Lists of Acronyms
Nomenclature
Publications
1 Introduction: towards future smart grids
1 Towards a low-carbon future
2 Smart grids and the need for exibility
3 Thesis objectives
2 Active integration of EVs into distribution systems
1 Methodology
2 Technical aspects
3 Economic aspects
4 Regulatory aspects
5 End-user aspects
6 Discussion
7 Partial conclusions
3 Plug-in behavior of EV users: modeling, insights from a large-scale trial and impacts for grid integration studies
1 Introduction
2 Literature review
3 EV simulation model
4 Insights from a large-scale EV trial and model calibration
5 Impact of non-systematic plug-in behavior on EV grid integration studies
6 Partial conclusions
4 Assessing EV integration in distribution grids: a data-driven approach.
1 Relevant works on EV integration into distribution grids
2 A data driven methodology to build realistic case studies
3 EV charging impact at the primary substation level
4 EV and PV integration in realistic MV grids
5 Partial conclusions
5 Participation of electric vehicle eets in local exibility tenders: Ana- lyzing barriers to entry and workable solutions
1 Introduction
2 Looking for decentralized exibility markets
3 Methodology and case study
4 Results
5 Partial conclusions
6 Final conclusions
A Detailed models for EV charging
1 Cost-optimization charging
2 Decentralized valley lling
B Grid reconstruction from GIS data
1 Datasets
2 Grid reconstruction methodology
C Computational times
D Complementary results from La Boriette case study
1 Spatial distribution of EVs and PV installations
2 Maximum line loading
E Resume en francais.
