EV Fleet Flexibility Estimation on the Distribution Network
BILAL FARES
Master of Science Thesis Department of Energy Technology
KTH 2020
EV Fleet Flexibility Estimation on the Distribution Network
TRITA-ITM-EX 2020:551
By Bilal Fares Approved
October 2, 2020
Examiner Peter Hagström
Supervisors Peter Hagström Mònica Aragüés Co-Supervisor
Fabian Rücker
Commissioner Contact person
Bilal Fares
Abstract
A new electricity market model was drafted in Denmark (1) to define the roles of different players and promote the integration of renewable generation as well as the gradual shift towards smart grids. Along with a clear political will and adjustments of some regulatory barriers, this potentially opens the market for an increased share in electric vehicles (EVs). Presenting an opportunity to analyze the role of EVs and their capacity to provide flexibility services for system operators, the distribution service operator (DSO) in particular. Specifically, it allows for an alternative route the DSO could take rather than resort to traditional measures to fix grid contingencies, which are not only costly and time consuming but have negative environmental and social impacts as well.
The thesis tackles flexibility from a fleet of EVs on the Danish island of Bornholm in a bid to estimate the value of instantaneous power that could be dispatched on the distribution grid at specific times. A real data set is analyzed corresponding to an EV fleet at 8 vehicle-to-grid (V2G) charger points, all connected to the same 400 kVA distribution transformer and already participating in energy sell-back to the grid. The fleet itself is part of the regional municipality, operating during working hours from 8am to 4pm on weekdays and providing homecare as well as social care to citizens. The data is acquired for the months of January and July 2020, which presents an added value to the analysis as peak electric demand in Denmark usually occurs during the winter season and the month of January around 7 pm in particular.
Flexibility estimation is provided through a model that is coded in python. It takes as an input a database formed from the provided data, including information on vehicle and charger models.
The code then outputs a power time series, which is the basis of the analysis in this thesis as it provides a platform that generates the possible amounts of flexible power that could be dispatched to the grid in addition to the power that is available for charging at different time slots and for a chosen duration. The analysis focuses on the flexible power that could be injected on the grid as part of services provided to the DSO to alleviate grid congestions and prevent the overloading of transformers.
The results show how the EV fleet of 8 V2G chargers can satisfy the forecasted increase in peak demand of 13% by 2030. These are estimated for the transformer operating at 70% load at peak and full load at peak, albeit at a slightly lower confidence level for the latter. Although the model does not take into consideration battery aging and charger optimization schemes, the results still provide such estimations at a 95% confidence level (i.e. -2 standard deviations) from a normal distribution curve for the month of January during peak times. With more types of EVs connected in the future, such as residential and commercial, flexibility levels can be predicted to further increase.
Abstrakt
En ny elmarknadsmodell utarbetades i Danmark (1) för att definiera olika aktörers roller och främja integrationen av förnybar produktion samt den gradvisa övergången till smarta nät. Tillsammans med en tydlig politisk vilja och justeringar av vissa rättsliga hinder öppnar detta potentiellt marknaden för en ökad andel i elfordon. En möjlighet att analysera rollen för elbilar och deras kapacitet att tillhandahålla flexibilitetstjänster för systemoperatörer presenteras, framförallt operatörer för distributionstjänster (DSO). Det möjliggör närmare bestämt en alternativ väg som DSO skulle kunna ta i stället för att tillgripa traditionella åtgärder för att fastställa beredskapsnät, som inte bara är kostsamma och tidskrävande, utan också har negativa miljömässiga och sociala konsekvenser.
Detta arbete tar itu med flexibilitet från en flotta av elbilar på den danska ön Bornholm i ett försök att uppskatta värdet av momentan kraft som skulle kunna sändas ut på distributionsnätet vid specifika tidpunkter. En verklig datamängd analyseras motsvarande en EV-flotta vid 8 laddare (fordon till rutnät) (V2G), alla anslutna till samma 400 kVA-distributionstransformator och som redan deltar i återföringen till elnätet. Själva flottan ägs av kommunen i regionen, som arbetar under arbetstid från 08:00 till 16:00 på vardagar, och som ger hemvård samt social omsorg till medborgarna. Uppgifterna förvärvas för månaderna januari och juli 2020, vilket ger ett mervärde för analysen eftersom den elektriska toppefterfrågan i Danmark vanligtvis sker under vintersäsongen och januari månad, omkring kl. 19.00 i synnerhet.
Flexibilitet-uppskattningen tillhandahålls via en modell som är kodad i Python. Som utgångspunkt används en databas som bildas från de tillhandahållna uppgifterna, inklusive information om modellerade fordon och laddare. Koden matar sedan ut en effekttidsserie, som ligger till grund för analysen i denna avhandling eftersom den ger en plattform som genererar de möjliga mängder av flexibel effekt som skulle kunna skickas till nätet, utöver den effekt som är tillgänglig för laddning vid olika tidsluckor och för en vald varaktighet. Analysen fokuserar på den flexibla effekt som skulle kunna injiceras på nätet som en del av de tjänster som tillhandahålls DSO för att lindra överbelastning av nätet och förhindra överbelastning av transformatorer.
Resultaten visar hur EV-flottan med 8 V2G-laddare kan tillfredsställa den prognostiserade ökningen av toppefterfrågan på 13% fram till 2030. Dessa beräknas för transformatorn som arbetar vid 70% belastning respektive full belastning vid topplast, om än på en något lägre konfidensnivå för den senare. Även om modellen inte tar hänsyn till batteriets åldrande och system för optimering av laddare ger resultaten ändå sådana uppskattningar på en 95%-ig konfidensnivå (dvs -2 standardavvikelser) från en normalfördelningskurva för januari månad under högtrafik. Med fler typer av elbilar anslutna i framtiden, såsom bostäder och kommersiella, kan flexibilitetsnivåer förutspås öka ytterligare.
Nomenclature
Abbreviations
BMS Battery Management System
BRP Balancing Responsibility Party
DoD Depth of Discharge
DSO Distribution Service Operator
EV Electric Vehicle
EVSE Electric Vehicle Supply Equipment
SoC State of Charge
TSO Transmission Service Operator
V2G Vehicle to Grid
List of Figures
FIGURE 1:FINAL ENERGY CONSUMPTION BY SECTOR IN DENMARK (3) ... 10
FIGURE 2:STOCK OF ELECTRIC CARS IN DENMARK FROM 2010 TO 2020(4) ... 11
FIGURE 3:FINAL ELECTRICITY CONSUMPTION BY SECTOR IN DENMARK (3) ... 11
FIGURE 4:THESIS OBJECTIVES ... 13
FIGURE 5:METHODOLOGY ... 14
FIGURE 6:ROLES AND RESPONSIBILITIES OF MARKET PLAYERS IN TRANSFERRING FLEXIBILITY (8) ... 15
FIGURE 7:LI-ION BATTERY CHARGING AND DISCHARGING CHEMICAL OPERATION (9) ... 16
FIGURE 8:BASIC CC-CVCHARGING REGIME OF A LI-ION BATTERY (10) ... 16
FIGURE 9:EFFECT OF THE MAXIMUM CHARGING LEVEL ON LI-ION BATTERY CYCLE LIFE (11) ... 17
FIGURE 10:TYPICAL LI-ION BATTERY DISCHARGE CURVE (9) ... 17
FIGURE 11:TRACTION BATTERY CAPACITY VS.LIFE CYCLE (13) ... 18
FIGURE 12:EVCHARGING LEVELS ... 19
FIGURE 13:BASICS OF EVCHARGING (9)... 19
FIGURE 14:V2GBI-DIRECTIONALITY FLOW CHART (15) ... 20
FIGURE 15:EVSESECTOR CATEGORIZATION (5) ... 20
FIGURE 16:EVSEAVAILABILITY PER SECTOR (5) ... 21
FIGURE 17:SAMPLE CONSUMPTION PROFILE IN ADANISH SEMI-URBAN SETTING (6) ... 22
FIGURE 18:DAILY AVERAGE PEAK LOAD IN BORNHOLM IN 2007(17) ... 22
FIGURE 19:POSSIBLE EVFLEXIBILITY SERVICES (20) ... 24
FIGURE 20:POSSIBLE EVCHARGING LOAD MANAGEMENT ... 25
FIGURE 21:POSSIBLE FLEXIBILITY MARKET OVERVIEW (27) ... 27
FIGURE 22:DANISH MARKET MODEL 2.0(1) ... 27
FIGURE 23:ELECTRICITY PRICES FOR HOUSEHOLD CONSUMERS IN EUR PER KWH (29) ... 28
FIGURE 24:BARRIERS FOR V2GSERVICES (25) ... 29
FIGURE 25:FLEXIBILITY MARKET PLATFORM ... 30
FIGURE 26:FLEXIBILITY BILATERAL CONTRACTS ... 30
FIGURE 27:LOCATION OF CHARGERS (31) ... 33
FIGURE 28:BASIC DATABASE STRUCTURE ... 34
FIGURE 29:DATABASE TABLE LAYOUT ... 35
FIGURE 30:CHARGING AND DISCHARGING EFFICIENCY MAP (32) ... 39
FIGURE 31:SAMPLE EVCHARGING CURRENT VS.SOCLEVEL ... 39
FIGURE 32:CVPHASE INTEGRATION ... 40
FIGURE 33:DATA ASSUMPTIONS ... 40
FIGURE 34:DATA CORRECTION POINTS ... 42
FIGURE 35:CHARGING ERROR ... 42
FIGURE 36:BATTERY MODEL ... 43
FIGURE 37:FLEXIBILITY MODEL SCHEMATIC ... 44
FIGURE 38:CODE STEPS ... 44
FIGURE 39:POWER-TIME SERIES MODEL OUTPUT SAMPLE ... 45
FIGURE 40:MODEL OUTPUT ... 46
FIGURE 41:TRANSACTION BEHAVIOR FROM CHARGER 1 FOR JANUARY ... 49
FIGURE 42:TRANSACTION BEHAVIOR FROM CHARGER 1 FOR THE SECOND WEEK OF JANUARY ... 50
FIGURE 43:TRANSACTION BEHAVIOR FROM CHARGER 1 PER DAY IN JANUARY ... 50
FIGURE 44: TRANSACTION BEHAVIOR FROM CHARGER 2 PER DAY IN JANUARY ... 51
FIGURE 45:TRANSACTION BEHAVIOR FROM 4CHARGERS FOR JULY ... 51
FIGURE 46:FLEXIBILITY POTENTIAL FROM CHARGER 1 ... 52
FIGURE 47:FLEXIBILITY POTENTIAL FROM 4CHARGERS ... 53
FIGURE 48:FLEXIBILITY POTENTIAL FROM 8CHARGERS ... 53
FIGURE 49:FLEXIBILITY POTENTIAL FROM 8CHARGERS PER DAY ... 54
FIGURE 50:V2GSERVICES ANALYSIS PLATFORM ... 54
FIGURE 51:AGGREGATED FLEXIBILITY POTENTIAL AT 1-MINUTE RESOLUTION ... 55
FIGURE 52:AGGREGATED FLEXIBILITY POTENTIAL AT A 15-MINUTE RESOLUTION ... 55
FIGURE 53:AGGREGATED FLEXIBILITY POTENTIAL AT A 30-MINUTE RESOLUTION ... 56
FIGURE 54:AGGREGATED FLEXIBILITY POTENTIAL AT A 60-MINUTE RESOLUTION ... 56
FIGURE 55:AGGREGATED FLEXIBILITY POTENTIAL AT A 120-MINUTE RESOLUTION ... 57
FIGURE 56:EFFECT OF INCREASED RESOLUTION ON ESTIMATED FLEXIBILITY ... 57
FIGURE 57:AGGREGATED FLEXIBILITY DISTRIBUTION DURING WORKING HOURS 8 AM –4 PM ON WEEKDAYS AND FOR THE LAST 3WEEKS OF JANUARY... 58
FIGURE 58:AGGREGATED FLEXIBILITY DISTRIBUTION DURING OFF-WORKING 6 PM –11 PM ON WEEKDAYS AND FOR THE LAST 3WEEKS OF JANUARY... 59
FIGURE 59:EFFECT OF RELAXED CHARGER LIMITS ON FLEXIBILITY POTENTIAL ... 60
FIGURE 60:DATA DISTRIBUTION FOR RELAXED CHARGER LIMITS FOR THE LAST 3 WORKWEEKS OF JANUARY DURING PEAK TIMES ... 60
FIGURE 61:ESTIMATED FLEXIBILITY POTENTIAL AS A PERCENTAGE OF TRANSFORMER CAPACITY ... 62
FIGURE 62:ESTIMATED FLEXIBILITY POTENTIAL AS A PERCENTAGE OF A 400 KVATRANSFORMER CAPACITY ... 63
List of Tables
TABLE 1:DATA DESCRIPTION ... 32
TABLE 2:DATA FIELDS ... 32
TABLE 3:DATABASE FIELD DESCRIPTION ... 35
TABLE 4:EVLOAD AS A PERCENTAGE OF TRANSFORMER CAPACITY ... 61
Table of Contents
Abstract ... 2
Abstrakt ... 3
Nomenclature ... 4
List of Figures ... 5
List of Tables ... 7
Table of Contents ... 8
1 Introduction ... 10
1.1 Background ... 10
1.2 Project Description ... 12
2 Objectives ... 13
3 Methodology ... 14
4 Literature Review ... 15
4.1 Defining EV Flexibility ... 15
4.1.1 Battery Operation ... 15
4.1.2 V2G Chargers ... 18
4.1.3 Sector Categorization ... 20
4.2 Load Analysis and Peaks... 21
4.3 Markets for EV flexibility ... 23
4.3.1 Benefits for the DSO ... 24
4.4 Role of the Aggregator ... 26
4.5 Regulation Barriers ... 27
5 Flexibility Market Platform ... 30
6 Database ... 32
6.1 Description of Data ... 32
6.1.1 Location of Chargers ... 33
6.2 Database Schematic ... 34
6.3 Data Information ... 38
6.3.2 Data Assumptions ... 38
6.3.3 Data Cleaning ... 41
7 Flexibility Model ... 43
7.1 Battery Modeling ... 43
7.2 Defined Parameters ... 43
7.3 Model Description ... 44
7.4 Model Calculations ... 46
7.4.1 Subplot 1: Available Capacities ... 47
7.4.2 Subplot 3: Capacity Flexibility ... 47
7.4.3 Subplot 4: Power Flexibility ... 47
8 Results and Analysis ... 49
8.1 Transaction Times and Charger Behavior ... 49
8.2 Aggregated Flexibility Potential ... 52
8.2.1 Potential from Chargers ... 52
8.2.2 Increasing Resolution ... 55
8.2.3 Effect of Day and Time ... 58
8.2.4 Relaxed Charger Limits ... 59
8.3 Analysis and Potential DSO Market ... 60
9 Conclusions ... 64
10 Future Improvements & Integrations ... 65
11 References ... 66
1 Introduction 1.1 Background
The EU has set sustainability targets with an aim to reach carbon neutrality by 2050. Taking the Danish market in particular, by 2030 it is expected to have a 55% renewable energy share accompanied by a phase out of coal powered plants (2). This increase in share would be ideally accompanied by optimized usage and flexible storage in order to harness the full potential of some renewable generation, especially that of wind and solar, that are volatile by nature. While many forms of storage are possible and studied in literature, such as compressed air and pumped hydro, the focus in this thesis is towards battery storage in electric vehicles and their capacity to provide some services on the distribution network.
A closer look at the Danish energy consumption shows the transport sector representing almost 25% of the total (3), a significant amount as shown in Figure 1. Noting that the transport portion in the graph excludes the aviation and marine sectors.
Figure 1: Final Energy Consumption by Sector in Denmark (3)
Establishing the Ministry of Climate, Energy and Utilities in 2015 for promoting a more sustainable society can be seen as a major motive for an increase in EV penetration. Since then initiatives have been presented to facilitate the transition in the transportation sector including a stoppage to the sale of diesel and petrol vehicles by 2030 (2). The increasing number of EVs on the Danish market is confirmed in Figure 2, with a significant jump from 2016 onwards. The registered number of vehicles increased by around 12,500 in the last 5 years alone.
Figure 2: Stock of electric cars in Denmark from 2010 to 2020 (4)
However, as illustrated in FIGURE 3, a very small portion of the transport sector has been electrified, only around 35 ktoe (~407 MWh) out of the total 4,422 ktoe (~51.4 GWh) consumed in 2018 (3).
This opens the market for significant increase in electric vehicles on the market and thus tackling their flexibility services on the grid. Especially as studies show that vehicles are used for active transportation only 5% of the time, opening up a lot of possibilities for V2G services (5)
Figure 3: Final Electricity Consumption by Sector in Denmark (3)
Nonetheless, flexibility services must be justified for grid operators such as the DSOs. With more electrification expected in the future the challenges to maintain the grid become more complicated.
Up until now, the measures taken to tackle load increases and voltage drops have mostly been done through grid reinforcements that include transformer and cable upgrades. However, and especially in urban areas, it mostly translates to inefficiencies related to cost and time as well as environmental and social impacts (6).
Therefore, before moving to such traditional grid alleviation methods, it is beneficial to analyze potential flexibility services from all energy resources on the grid. In particular, EVs come with certain flexibility measures thanks to their ability to store energy in their traction battery. This can be utilized in several scenarios on the grid such as maximizing the utilization of renewable resources and providing services for protecting the distribution grid from possible contingencies.
1.2 Project Description
The thesis focuses on the Danish electric market in a bid to identify possible power estimations that could be provided from EV flexibility to the DSO. It tackles meter readings from 8 V2G chargers located in the Danish Island of Bornholm and connected to a single 400 kVA distribution transformer. The charging profile of these charger is that of a fleet that operates on the during regular workday timing by providing social services to citizens.
However, power estimations are done after understanding the potential markets for EV flexibility and sketching the possible interactions between the different market players. Mainly, identifying the roles for the V2G aggregator, flexibility market operator as well as the DSO.
Flexibility estimations from the EV data is then done through an analysis platform that is modeled through python. It aggregates EV charging/discharging power and SoC mean values from multiple chargers to generate a power time series based on selected parameters such as time period and resolution. The basis of the model though is a database that is formed to provide a flexible structure and allow for future additions and expansions, with an aim to provide a unified input.
It is thus with this EV flexibility platform that an analysis can be targeted on multiple loading scenarios of the distribution transformer. Importantly, whether EV flexibility is capable of satisfying DSO services in the future with an increased electrification accompanied increased load and peak demand.
2 Objectives
With increasing EV penetration on the grid and the aim to decarbonize the transportation sector by 2050 it becomes practical to tackle the flexibility services that could be provided by EVs. Thus, the main objective from this thesis is to estimate the possible amount of power flexibility available at EV charger points and assess its potential at a DSO level.
This is done through a python model that allows for the aggregation of EV charger meter data and the creation of a power time series. The series itself will be based on specific input parameters and provide aggregators with a platform for forecasting and estimating the potential from EVs based on historical data. More specifically it allows the aggregator to estimate the amount of power flexibility that could be provided to the DSO during selected time slots, whether it is for charging or discharging purposes. The focus for the analysis, however, will be directed towards the discharging power flexibility that can help alleviate grid congestions during peak times.
A more detailed set of objectives is identified in Figure 4 with each one elaborated on in more details under the relevant sections throughout the report.
Figure 4: Thesis Objectives
3 Methodology
The basis of the thesis has been a thorough literature review that enabled an understanding on current EV charging technologies and their operation as well as the different actors and markets for EV flexibility. After that, a sample market interaction platform was sketched to designate the beneficiaries of the model and better illustrate how the interactions will take place between the different market players.
Once EV data has been received, it was analyzed in order to draw out the possible outcomes and how to utilize it as an input to the created model. Therefore, it became necessary to define some parameters, make some assumptions and clean the data from any anomalies. This step was crucial to a more efficient and valuable result and conclusion. The data itself was transformed into a database that allows for several models of EVs and chargers integrated. Although the current set of data has only one EV and V2G charger model it can form a useful base for future integrations.
The next step involved creating the flexibility model through python and performing quality analysis to validate its output. Through this analysis a list of errors was spotted within the data and cleaned accordingly. After the model was running seamlessly results were analyzed based on different parameters and inputs, with a conclusion drawn in the end on the potential of EV flexibility on the distribution network.
The methodology is summarized in Figure 5. However, a more detailed description of each block can be found in the relevant sections throughout the report.
Figure 5: Methodology Literature Review
Market Interactions and Platform
Modeling
Preliminary Data
Analysis Data Assumptions
and Cleaning Database Creation
Flexibility Modeling Quality Analysis
Results Analysis
Conclusions
4 Literature Review
This section is designated to provide necessary information on EVs and V2G operation, allowing for the estimation of EV flexibility and the analysis of its potential on the distribution grid.
4.1 Defining EV Flexibility
The Danish Energy Association defines flexibility in a context where 3 market players engage in electrical transfer on the power system. From one end the private players (i.e. flexible loads such as heat pumps and EVs) provide flexible services to system operators through commercial players such as aggregators (7). These roles and responsibilities are illustrated in Figure 6. Taking the case of EVs in particular, the flexibility services they can provide to different market players are discussed in more details in 4.3. A closer look is taken in the next section on the operation of the EV’s traction battery, the source of this electric flexibility.
Figure 6: Roles and Responsibilities of Market Players in Transferring Flexibility (8)
4.1.1 Battery Operation
Lithium-ion batteries are present in most electrical equipment nowadays due to their high energy density and abundant raw materials (9). It is also the predominant direction taken by EVs with the choice of material at the cathode varying between different manufacturers. The Nissan Leaf for example utilizing a Lithium Manganese Oxide while the Tesla Model S utilizing a combination of Lithium Nickel Cobalt Aluminum Oxide. Nevertheless, the EV battery models follow the same basic chemical reactions during charging and discharging operations as highlighted in Figure 7.
Lithium ions transfer from the negative anode to the positive cathode during discharging while the transfer direction is shifted during charging, with an input source providing the power.
Figure 7: Li-ion Battery Charging and Discharging Chemical Operation (9)
Some terminology on the battery’s technical specifications before moving forward:
• SoC: The percentage of the available capacity as part of the maximum battery capacity.
• DoD: The percentage of the maximum battery capacity that has been discharged.
• EQFC: Equivalent full cycles (With over 1000 desirable (9)).
A common charging regime is the CC-CV, which is utilized at the chargers analyzed in this thesis.
It can be split into two basic phases: constant-current charging and constant-voltage charging, known as the CC and CV phases respectively. This is illustrated in Figure 8, where during the CC phase the constant current charges the battery at an increasing voltage that will bring the SoC to a certain level. The second phase continues the remaining battery charge with a constant voltage and a decreasing current.
Figure 8: Basic CC-CV Charging Regime of a Li-Ion Battery (10)
An advantage of Li-ion batteries is that they do not need to be fully charged and it is considered better for battery life to not have an SoC of 100% as high voltage stresses it out (11). As highlighted
in Figure 9 overcharging a battery reduces its useful operation cycles and keeping it within lower SoC levels improves the lifetime dramatically. Significantly, if charged to only 80% SoC the number of useful cycles could even go to around 1500.
Figure 9: Effect of the maximum charging level on Li-ion Battery Cycle Life (11)
Moving on to the discharging operation, battery life is extended non-linearly when it is not discharged completely and EV manufacturers may choose to set DoD levels at around 80% to increase service life (12). They may size the traction battery range according to this DoD level in order to guarantee customers the advertised km range. Keeping a reserve charge in the battery of around 20% also prevents dipping through non constant low voltage levels, which can affect battery life, as shown in Figure 10.
Figure 10: Typical Li-ion Battery Discharge Curve (9)
Nonetheless, the traction battery gradually fades at the end of life operation even with grace capacities reserved at chosen SoC and DoD levels, as shown in Figure 11. This grace capacity starts to become consumed and part of the driving range as the battery is utilized (13). A trend that can be observed in some vehicles. While several other factors affect battery life, prolonging it to the maximum becomes part of the job of a Battery Management System – BMS – with a focus on minimizing cost and maximizing efficiency (9).
Figure 11: Traction Battery Capacity Vs. Life Cycle (13)
4.1.2 V2G Chargers
Charging stations or Electric Vehicle Supply Equipment (EVSE) are the new components on the grid that will be responsible for supplying power to the EVs. Several types of charging levels are currently on the market as illustrated in Figure 12. In general, these levels are split between AC charging and fast DC charging.
The first level is a single-phase charger operating at a voltage of 120 Vac and is typical for applications that involve large charging times such as in homes and residences. Such levels, however, are not present in Europe due to the 220V electric supply. The second AC charging level meanwhile charges at the standard 230/240 Vac and is generally faster than the first level. It can be seen in different European sectors, such as residential and commercial, as well as public stations (14). On the other hand, the DC charging category is a much faster method due to the ability to bypass the on-board battery charger in the EV and transfer the energy through a direct current.
This is further illustrated in Figure 13 where power in an AC level charger has to pass through the on-board charger of the EV to convert it to DC before supplying it to the battery. While in the case of DC charging the power goes straight to the battery (9).
Standard connectors are utilized for EV charging and discharging schemes. The SAE J1772 connector is currently utilized in most of modern chargers and vehicles for AC level charging.
While DC charging has several standardized communication protocols such as CHAdeMO, CCS and Tesla’s supercharger.
Figure 12: EV Charging Levels
Figure 13: Basics of EV Charging (9)
As the focus in this project is on grid services, a V2G charger is needed. It can be described as the flow of energy in two directions, in and out of the EV’s traction battery. This bi-directionality, as illustrated in Figure 14, is mostly accessible through DC chargers; however, AC chargers can be developed for such applications granted the vehicle is accompanied with a bidirectional invertor.
Compatible electric vehicles also need to be connected to these chargers for providing flexibility services, specifically EVs that are capable of V2G communication, such as the Nissan leaf.
In brief, AC power is delivered from the grid to the EVSE through a controller that manages the charging profiles for several charging stations. This AC power is converted to DC and energy is stored in the EV’s traction battery. Conversely, when power is required on the grid, energy is dispatched from the EVSE after it is converted from DC to AC.
Level 1: AC Charging
A slow charging
level of around 1
kW.
Level 2: AC Charging A higher AC
charging level of around 3-20
kW
Level 3: DC Fast Charging A fast DC
charging level with over 20 kW
Figure 14: V2G Bi-directionality Flow Chart (15)
In Denmark, 10 kW bidirectional chargers are running with a CHAdeMO protocol, which will be the basis of the data in this report. Some projects analyzing V2G services, such as the Parker project, even set the limits at ±8.5 kW due to software limitation.
4.1.3 Sector Categorization
For assessing grid services from EVs, their operation and availability becomes critical. In general, EVSEs can be categorized under 4 types: Residential, Fleet, Workplace and Commercial. Whilst the focus in this report will be on fleet EVSE due to the analyzed data, Figure 15 provides a description of each category. Noting that some of these categories could be also utilized for other types of EV flexibility services such as V2H – Vehicle to Home – and V2B – Vehicle to Building – as is the case for Residential and Workplace EVSEs respectively.
Almost 19% of the passenger vehicles in the United States are of the fleet category (5), it therefore becomes a significant sector for grid services. Especially when considering that the fleet owner has the possibility of charging more than one vehicle at each charger.
Figure 15: EVSE Sector Categorization (5)
•Chargers at homes dominated by AC charging levels. Generally connected off-working hours from 6 pm to 8 am.
Residential
•Chargers that are placed at the employer's facility in a similar way to the workplace category. However, the fleet is utilized for working within the day such as school buses, pubic utilities and deliveries.
Fleet
•Chargers at work sites and offices. Generally connected during working hours from 8 am to 5 pm.
Workplace
•Chargers at remote locations where users do not spend a prolonged period of time such as malls and restaurants.
Commercial
The availability potential of each EVSE sector on a typical working day is illustrated in Figure 16.
As can be seen, the different sector profiles allow for different scenarios and potential services.
Although some periods of low unavailability overlap, mainly around 8 am and 5 pm, which refer to the times when people go to and leave from work. Nonetheless, it still shows that EVs are connected and could be available for some flexibility services throughout the day. This will be studied in this thesis with real data from fleet EVSEs, noting that grid services from EVs are dependent on having appropriate V2G chargers.
Figure 16: EVSE Availability per Sector (5)
4.2 Load Analysis and Peaks
The highest demand for electricity in Denmark generally occurs during the winter season (6), which is confirmed through a study case done in 2012-2013 on a semi-urban LV grid in a Danish town, as shown in Figure 17. On a more national scale, peak demand occurs during the month of January around 7 pm (16). This presents a perfect scenario for analysis as the data gathered from the chargers in this thesis is for the month of January.
Figure 17: Sample Consumption Profile in A Danish Semi-Urban Setting (6)
Although summer peaks on the island might be coming from the tourism sector, an overview of the Bornholm power system still suggests a peak demand during the month of January. As is shown in Figure 18, with a peak demand almost around 56 MW back in 2007 (17). That is why January will be considered the worst-case month in terms of electric load and the one that the analysis will be based on.
Figure 18: Daily Average Peak Load in Bornholm in 2007 (17)
The danish electric system itself is expected to have an increase in peak load by 2030 averaged around 13%, which mainly comes from the conventional increase in demand as well as the projected injection of EVs and heat pumps. As they are considered flexible resources, EVs and heat pumps can play a role in how much this peak demand increases especially in January when heat pumps are mostly required. One study estimates the increase in peak load to fluctuate between
8 and 16 % depending on the usage of these flexible loads. Specifically EVs, if the charging profiles were optimized the peak load increase could be reduced by as much as 4% (18). EV charging would have to managed anyways for avoiding high voltage drops across the secondary low voltage lines.
Thus shifting the EV load in a controlled algorithm to off-peaks or take advantage of surplus renewable production can help mitigate such issues (19).
Nevertheless, the 13% increase in peak demand will be the value used when estimating the potential of an EV fleet to tackle such load increase.
4.3 Markets for EV flexibility
Several Danish projects run with a focus on V2G integration. Some of those include the Parker project, Nikola and Edison, with the latter in particular focusing on the Danish island of Bornholm.
Possible markets for EV flexibility, as shown in Figure 19, can be split under 3 categories:
transmission, distribution and user. The transmission services, provided to the TSO, are mainly related to the balancing of regional power systems with increased renewable penetration. These include power balancing schemes that provide instantaneous power balancing such as frequency containment as well as energy balancing schemes such as the ones utilized by BRPs for regulating grid supply and demand.
DSO services are more focused on local distribution grids and neighborhoods. They include grid contingencies that allow for a proper and reliable operation of the distribution grid as well as allowing some sort of energy autonomy where part of the energy is produced and consumed by the local community. As the data analyzed in this thesis is part of a local distribution transformer the focus will be on such services in 4.3.1.
The last category involves the user, which could be a building, house or commercial site. In such cases there might be benefits for an extra security of supply with an islanded operation, whereby the user can self-sustain the electric consumption through flexibility and without a connection to the grid. It could be seen as a backup power or even a completely off-grid system. Another option to tackle flexibility in this sector would be in terms of mobile load serving or operating loads from EVs. Examples of such applications could be vehicle-to-tool or vehicle-to-vehicle operations.
Figure 19: Possible EV Flexibility Services (20)
4.3.1 Benefits for the DSO
With distributed generation on the rise added to the increased penetration of electric vehicles on the distribution grid as well as an increase in electricity demand, a new mechanism is needed to optimize the use of such resources and loads. Traditionally, with an old type of single-direction electric network the DSO would plan network reinforcements as well as upgrades accordingly. But with the new shift towards smart grids, flexibility services from EVs in particularly could provide another way of tackling expansions and issues on the distribution grid.
The main flexibility analyzed in this thesis is on the discharging potential of EVs. In other words, the injection of power on the distribution grid from this flexible source. However, another main asset from EVs is their flexibility in charging as well, which can be modelled as a load: shifted and adjusted to minimize technical losses as well as costs as shown in Figure 20. The main increase in EV load in this case comes from the residential sector where charging is done at home. Instead of simultaneous charging, optimized models are run to prevent any grid congestions and avoid reaching the limit power capacity.
Figure 20: Possible EV Charging Load Management
Nonetheless, DSOs will have to face other major load increases from heat pumps in addition to electrification in other sectors in general. Thus, it becomes critical to assess EVs as a source of power as well. The first need for EV flexibility services for the DSO comes in the form of grid contingencies, which can be split under congestion and voltage issues. Traditionally these issues have been dealt with by the DSO through the following methods:
• Cable reinforcements to minimize voltage drops across the feeding lines and to accompany a transformer upgrade with an increased load
• New transformers or transformer upgrades based on an increased load
• Capacitor banks to regulate the voltage at points on the low voltage network where voltage drops below standard values (21)
• Adjusting transformer taps to regulate the voltage (21)
However, such methods are deemed outdated with distributed generation and EVs (22). More importantly replacing electric network components is not only costly but also time consuming and labor heavy. Especially when considering urban areas where most of the cables are laid underground. Thus in addition to cost, other challenges face the DSO and are mainly related to time, environmental impact as well as social difficulties with increased tariffs to cover investment costs (23).
Therefore, the main value that comes out of EV flexibility for the DSO lies in voltage regulation, phase balancing and congestion control. The latter will be considered for the analysis in this thesis due to the lack of information on the island’s low voltage grid configuration. So the market for the DSO in this case would be the savings it will generate from utilizing flexible resources rather than tending to traditional schemes (20). One study goes to show that power congestion can be solved through flexible sources in the order of 100-200 kW and for a duration of 1-4 hours with a yearly activation (24). The parker project estimates a potential revenue for DSO balancing services from EVs up to a monthly amount of 700 DKK, or around 95 euros, per transformer (25). It factors in
this amount the average price of the transformer (mean value of a 1000 or 1600 kVA transformer), its monthly depreciation as well as the potential for balancing services. However, it neglects the price of cables and the labor work for installing them, which could increase the revenue estimations significantly. This shows that the market is still not mature enough but could be built upon with other services.
Another advantage for utilizing EV flexibility at the distribution grid level comes from energy autonomy. In this case the value for the DSO comes from maximizing the use of locally produced energy as well as allowing for the trading of energy between local members of the community (20).
4.4 Role of the Aggregator
The need for aggregation mostly arises from the transition towards decentralized generation and integration of flexible resources, the aggregator can thus be defined as an “intermediary between electricity end-users, who provides distributed energy resources, and those power system participants who wish to exploit these services” (26). According to the USEF, the goal of the aggregator is to accumulate and maximize flexibility of prosumers in order to sell it to the BRP, DSO or TSO (8).
With a focus on DSO services as discussed in section 4.3.1, the role of an EV aggregator in particular becomes to manage the new bi-directionality of electricity. Thus, providing the DSO with flexibility services such as congestion management and voltage control.
Figure 21 highlights possible interactions between different market players on energy, monetary and service behaviors. The aggregator in this case could be viewed as a supervisor of the local electric market, managing its operations to maximize the profits of its participants. It can provide flexibility services to the DSOs, balancing responsibility parties (BRPs) and prosumers. Focusing on EVs in particular, the aggregator can pool flexibility from a group of EVs that are connected to the same DSO and provide this DSO with services such as congestion management. (27)
Figure 21: Possible Flexibility Market Overview (27)
4.5 Regulation Barriers
Unfortunately markets for aggregation are still not favorable in Denmark. Nonetheless, work has been done to create aggregator platforms and asses the future market players and how they will interact together. All of this is in anticipation of the new Market Model 2.0 shown in Figure 22. In this model the aggregator moves from being just a role to a market player supplying flexibility and electricity.
Figure 22: Danish Market Model 2.0 (1)
The main barriers preventing V2G aggregation on the Danish market are detailed in the Parker project and summarized in Figure 24. They can be placed down to the high electricity prices, the unfavorable tax system as well as the general lack of political will.
Specifically, the price of electricity in Denmark is still the highest in the EU at 0.2924 euros/kWh with more than 50% of this price being paid for taxes and VAT, as shown in Figure 23. The tax scheme in particular is a big demotivator for V2G adoption as it taxes at both ends of the service:
when charging and discharging. Add to that the high prices of V2G charging equipment, it somewhat discourages investors from participating in the flexibility markets. However, with correct government policies EV penetration on the power grid can increase significantly, further opening the market for investments in V2G services. (28)
Figure 23: Electricity prices for household consumers in EUR per kWh (29)
Figure 24: Barriers for V2G Services (25)
Political
• No incentives for V2G aggregation as well as EV purchases and charger installations
• Lack of framework on how the aggregator and DSO can interact
Economic
• High electricity prices and tax payments by consumers
• Need for DSO investment in meter structures
• Market restrictions does not allow aggregators to be profitable
Social
• Lack of subsidize for EV purchase
• Lack of proper communication about EV integrationgs
Technical
• Limited number of EVs on the market
• Lack of charging infrastructure in buildings and commercial settings
Environmental
• Low confidence on battery degredation
• Not all customers are environmentallly driven
Legislative
• Data security and large minimum amount of electricity to be traded on the market
• Outdated building regulations
5 Flexibility Market Platform
For modeling the flexibility market in this project two entities are considered: the DSO and the V2G aggregator. Figure 25 illustrates a possible scenario for this interaction where the DSO generates a flexibility demand on the market. The flexibility from this market could be provided from other flexible loads as well such as postponable devices, electric water heaters and thermostatically controlled loads. From the aggregator side, its role is to generate flexibility offers on this market, where all offers are combined with the DSO’s demand in order to generate a merit order list.
For this to become possible a flexibility market operator is required to govern such operations and transactions. After analyzing the merit order list, the DSO purchases flexibility from the market, which reimburses the corresponding V2G aggregator and dispatches the flexibility accordingly.
Figure 25: Flexibility Market Platform
Figure 29 illustrates another possible form of contact between a DSO and a V2G aggregator through bilateral contacts. This form is more straight forward and involves a DSO flexibility demand request directly to the aggregator. It then follows on a similar path as before but without a flexibility market as an intermediary.
Figure 26: Flexibility Bilateral Contracts
These are some possibilities of how DSOs and V2G aggregators can interact with flexibility transfer. They can be expanded on later to include more entities; however, for the purpose of aggregating flexibility from EV charger points it has been kept simple to illustrate the possible interactions between a DSO and a V2G aggregator. The platform can also include possible planning data that is generated by the aggregator in order for the DSO to study and plan its distribution network accordingly. Several other things will need to be tackled on the market such as the responsible party in case of imbalances, what happens in case the aggregator goes bankrupt, etc. However, such market framework details will not be analyzed further in this thesis.
6 Database
6.1 Description of Data
The data, summarized in Table 1, is collected from 8 EV charger points within the Danish Island of Bornholm. These V2G chargers are of specific design in the Danish market with a charging and discharging limits of 10 kW. The EV fleet that utilizes such chargers are part of a social work, where care is provided to citizens at their homes. It operates in normal working hours from 8 in the morning until 4 in the afternoon on weekdays (30). Therefore, the main connection and disconnections for EVs are typically done within this time period and thus can be categorized under fleet charging.
The available data from the 8 EV chargers is of a minutely format for the months of January and July 2020. From this, a database is formed linking different types of information that will be used as a model for future integrations and estimations.
Table 1: Data Description
Data Description
Resolution 1 min
Intervals / Duration Months of January 2020 and July 2020
Location Danish Island of Bornholm
Number of Chargers 8
Table 2illustrates the different fields that are presented in the data. The readings are provided for each charger with a status field to show the relevancy of the entry with regards to the flexibility it might provide. For example, the Status field can take the one of the following options:
• A vehicle is connected to the charger but not charging
• A vehicle is connected to the charger and charging
• A vehicle is connected to the charger and providing grid services (i.e. charging and/or discharging)
• No vehicle is connected
• The charger is down for an error or for maintenance
For each of those minutely instances it is then possible to identify several important parameters that are essential for calculating the flexibility potential from these chargers. Mainly the current SoC of the traction battery and the current charging or discharging power. However, other fields, such as the current, will be utilized for data correction as will be seen in subsequent sections.
Field Units Description
Time Stamp MM/DD/YYYY HH:mm The day and time of the reading
Charger ID - The unique ID of the charger (Could be that two chargers exist in the same station; however, each has a unique ID) Charger Status - Status of the charger connection whether it is connected to
a vehicle or idle
SoC - The mean value of the state of charge of the connected vehicle
Provided
Power kW DC charging/discharging power provided by the charging station
Voltage Volts DC Voltage mean value at the EV side Current Amps DC current mean value at the EV side
6.1.1 Location of Chargers
There are currently more than the 8 chargers that are analyzed in this report spreading across the Danish Island of Bornholm. However, for the purpose of providing flexibility services to the DSOs it is important to identify the grid zone of operation and to what distribution transformer those chargers are connected to. For that purpose, Figure 27highlightsthe location of EV chargers on the island, which are mainly located in the three cities of Ronne, Hasle and Tejn.
Figure 27: Location of Chargers (31)
The 8 chargers in this case are all connected to the same distribution transformer, which allows for an accurate analysis of their aggregated potential for grid services. This is also essential for
estimating the potential of EV flexibility in satisfying the increase in peak demands as well as tackling load increases during the summer period. Especially as over 600,000 tourists visit the island during that season, which adds to the electric load from its 40,000 inhabitants (30).
6.2 Database Schematic
The idea behind defining a database is to better structure the data, save space and more importantly run fast applications especially as the data could be huge in numbers. A flexible structure is also desired in order to integrate, in the future, several forms of data from different aggregators or chargers. Based on the data described in the previous section, Figure 28provides a basic database structure that is drawn from this data. It shows the four main tables forming the database and how they are related to each other:
• Meter Readings: A table that contains data from the charger meters such as power, timestamps and SoC level of the connected vehicle. This table connects to two other tables, Chargers and Vehicles.
• Chargers: A table that contains data on the available charger models with important information such as their charging limits and incremental charging power. It is also connected with the Transformers table.
• Vehicles: A table that contains data on the available EV models with important information such as vehicle capacity, SoC levels and charging limits.
• Transformers: A table that holds information on the distribution transformers that supply the available chargers. Such data could include the rating of the transformer as well as its peak load. *This data is not included in this project as only 1 transformer is tackled.
More tables and relationships could be added in the future to expand on the analysis and integrate forecasting techniques. Some of those might include information on user behavior, transactions, charger models and market pricing. More on this is discussed in 10.
Figure 28: Basic Database Structure
•Contains data of DSO MVLV transformers
Transformers
•Contains data on the EV chargers
Chargers
•Contains data on charger meter in specific time steps
Meter Readings
•Contains data on the different EV models
Vehicles
The four tables are linked together as shown in Figure 29. Primary keys are provided for each table in the form of: TransformerID, ChargerID, MeterReadingID and VehicleID. While foreign keys maintain the relationships within these tables:
• TransformerID is a foreign key in the Chargers table
• ChargerID and VehicleID are both foreign keys in the Meter Readings table
A more detailed description of each field is provided inTable 3.It shows the symbol of each field that will be used as its identification in subsequent sections.
Figure 29: Database Table Layout
Table 3: Database Field Description
Table Field Symbol Description Units
Transformer TransformerID 𝑖𝑑𝑡𝑓𝑜 Unique identifier for the
distribution transformer. - Transformer Location 𝑔𝑝𝑠𝑡𝑓𝑜 GPS coordinates of the
distribution transformer. Degrees Transformer Rating 𝑃𝑟𝑎𝑡𝑖𝑛𝑔,𝑡𝑓𝑜 Power rating in kVA of the
transformer. kVA
*TransformerID Location
Rating PeakLoad
*ChargerID TransformerID MaxChargingPower MinChargingPower MaxDischargingPower
MinDishargingPower PowerIncrement
*MeterReadingID ChargerID VehicleID TransactionID
TimeStamp Status Power Current EnergyCharged EnergyDischarged
SoC
*VehicleID MinSoC MaxSoC MaxChargingPower MaxDischargingPower
Capacity CVSoC
