Indicator-Based Methodology for Assessing EV Charging Infrastructure Using Exploratory Data Analysis

Abstract

Electric vehicle (EV) charging infrastructure rollout is well under way in several power systems, namely North America, Japan, Europe, and China. In order to support EV charging infrastructures design and operation, little attempt has been made to develop indicator-based methods characterising such networks across different regions. This study defines an assessment methodology, composed by eight indicators, allowing a comparison among EV public charging infrastructures. The proposed indicators capture the following: energy demand from EVs, energy use intensity, charger's intensity distribution, the use time ratios, energy use ratios, the nearest neighbour distance between chargers and availability, the total service ratio, and the carbon intensity as an environmental impact indicator. We apply the methodology to a dataset from ElaadNL, a reference smart charging provider in The Netherlands, using open source geographic information system (GIS) and R software. The dataset reveals higher energy intensity in six urban areas and that 50% of energy supplied comes from 19.6% of chargers. Correlations of spatial density are strong and nearest neighbouring distances range from 1101 to 9462 m. Use time and energy use ratios are 11.21% and 3.56%. The average carbon intensity is 4.44 gCO 2eq /MJ. Finally, the indicators are used to assess the impact of relevant public policies on the EV charging infrastructure use and roll-out.Energies 2018, 11, 1869 2 of 18 incentives, and direct electric vehicle requirements. Charging infrastructure support to consumers is also a common characteristic of these markets.The main concerns addressed in the literature regarding the charging infrastructure deployment are related to cost, charging effectiveness, and ability to satisfy dynamic demand, as well as its overall environmental impact. According to the EAFO (European alternative fuels observatory) [3], the ratio of cars per charger vary widely from country to country, from 66 to 3.7 cars per charger in Iceland and Spain, respectively. Even though at first sight, factors such as the driven distance per year, and population size and density, do not change proportionally. There are no commonly accepted goals or standards for charging infrastructure density, either on a per-capita or per-vehicle basis. Different countries seem to follow different sizing and options for the public charging infrastructure development. This depends on many factors. There is no clear way to achieve an efficient deployment of EV charging infrastructure and the associated policy that will need to be addressed to help pave the way for electrification. A study on the emerging best practices for EV charging infrastructure [4] provides insights into the differences between present infrastructure roll outs. Using a multivariable regression of 350 metropolitan areas, the authors find that both level 2 and direct current (DC) charging are linked to EV acceptance, as are consumer purchase incentives. Yet the significant charging variability across hundreds of cities poin...

EV Idle Time Estimation on Charging Infrastructure, Comparing Supervised Machine Learning Regressions

“…The power histogram in Figure 2c shows that most sessions use on average of 3.8 kW of power supply. Figure 2d shows the complete range of idle time in the data set with extremely high values close to 300 h. However, the right skewed histogram shows a median value below 8 h. The authors of Reference [7], having analyzed the data set for the years 2014 and 2015, report that the mean availability of the public charging infrastructure depending on the time of the day may vary from 37.8 to 99.8%. The mean idle time assessed in their study represented about 64.1% of the connected time.…”

“…Power are calculated based on the meter readings per transaction. Following previous work done on the dataset [7] the attribute "Road Segment" was added. This was done by using OSM library in Python which identifies the location coordinates in the dataset and based on the google street map assigns a road segment to each charger.…”

“…Figure 1 shows how the charging sessions on the dataset analyzed are split across the road segments (left) and weekdays (right). The high share of 74.1% of charging sessions in the residential road segment is mostly explained by the higher number of residential segment chargers overall accounting for 67.8% of connectors followed by tertiary with 15.5% and secondary with 9.58% [7]. Furthermore, it is typically the public charging solution used by users when they are at home.…”

“…The use of real data has made possible the development of performance indicators studies for the charging infrastructure, which, despite scarce, capture important and key concepts to be taken in consideration [5][6][7][8]. Targeting the improvement of charging station deployment, a study [5] using real-world data, provides a new approach for municipalities, by treating the infrastructure as a complex network of charging stations and defining vulnerability in respect to the availability of its surrounding charging stations, within relevant walking distance.…”

“…In a study analyzing a dataset capturing 2014 and 2015 years [7] in the Netherlands, the authors developed eight quantitative performance metrics, providing a common methodology to compare and analyze different charging infrastructures. The authors used the following indicators: Energy demand from EVs, energy use intensity, charger's intensity distribution, the use time ratios, energy use ratios, the nearest neighbor distance between chargers and availability, the total service ratio, and the carbon intensity of the chargers themselves corresponding to the life cycle analysis of the materials involved.…”