Electric Vehicle Charger Infrastructure Planning: Demand Estimation, Coverage Optimization Over an Integrated Power Grid

Harshal D. Kaushik, Member, IEEE, Jingbo Wang, Graduate Student Member, IEEE, Roshni Anna Jacob, Member, IEEE, Jie Zhang, Senior Member, IEEE · 2025

Read Electric Vehicle Charger Infrastructure Planning: Demand Estimation, Coverage Optimization Over an Integrated Power Grid by Harshal D. Kaushik, Member, IEEE, Jingbo Wang, Graduate Student Member, IEEE, Roshni Anna Jacob, Member, IEEE, Jie Zhang, Senior Member, IEEE (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-23699.txt.

Introduction

Cities planning charger networks must please drivers and the power grid at the same time. Harshal D. Kaushik and 3 co-authors present a two-phase framework that plans from the grid's point of view. First it estimates charging demand and picks candidate sites using spatial points of interest. Then a maximum coverage optimization places chargers while keeping each new charger inside grid capacity limits.

Spatial studies often ignore grid limits, the authors note, while grid studies often guess demand from statistics. Their Dallas test case joins the existing charger network to an eighty-five-hundred-bus distribution model. The result is a deployment plan meant to be both useful and electrically safe.

Catalog record

CREATORHarshal D. Kaushik, Member, IEEE, Jingbo Wang, Graduate Student Member, IEEE, Roshni Anna Jacob, Member, IEEE, Jie Zhang, Senior Member, IEEE
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-23699

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Electric Vehicle Charger Infrastructure Planning: Demand Estimation, Coverage Optimization Over an Integrated Power Grid Harshal D. Kaushik, Member, IEEE, Jingbo Wang, Graduate Student Member, IEEE, Roshni Anna Jacob, Member, IEEE, Jie Zhang, Senior Member, IEEE Abstract For electrifying the transportation sector, deploying a strategically planned and efficient charging infrastructure is essential. This paper presents a two-phase approach for electric vehicle (EV) charger deployment that integrates spatial point-of-interest analysis and maximum coverage optimization over an integrated spatial power grid . Spatial-focused studies in the literature often overlook electrical grid constraints, while grid-focused work frequently considers statistically modeled EV charging demand. To address these gaps, a new framework is proposed that combines spatial network planning with electrical grid considerations. This study approaches EV charger planning from a perspective of the distribution grid, starting with an estimation of EV charging demand and the identification of optimal candidate locations. It ensures that the capacity limits of newly established chargers are maintained within the limits of the power grid. This framework is applied in a test case for the Dallas area, integrating the existing EV charger network with an 8500-bus distribution system for comprehensive planning. I Introduction The widespread adoption of electric vehicles (EVs) depends significantly on the expansion of the charging infrastructure network. Establishing charging station infrastructure practically involves three key stages: first, estimating the EV charging demand within the target area; second, identifying optimal locations for new chargers; and third – often overlooked – ensuring that the charger capacities do not exceed limits that could cause power grid violations. A common approach in the literature for estimating the EV charging demand involves modeling EV battery usage and simulating energy consumption under various traffic scenarios. However, obtaining realistic data on actual EV charging demands in specific areas is often challenging [ 1 ] . Some studies predict charging demands based on vehicle ownership, travel behavior, and household travel surveys [ 2 , 3 , 4 , 5 , 6 , 7 , 8 ] . The main issue with these simulation-based approaches is their computational expense, which makes them difficult to scale. Given these scalability challenges, data-driven approaches have gained popularity in recent years. These methods typically divide the area into discrete spaces, extract driver’s patterns from the travel mobility data, and estimate public charging demands for each cell [ 8 , 7 , 9 ] . Some studies consider the proximity of demand from neighboring regions [ 10 , 11 , 9 ] , while others argue that the charging demand depends on the specific characteristics of each region [ 12 ] . In these approaches, each discrete region is considered as a node, and nodes are ranked based on their unique features. Flow-capturing location model is a major tool in the literature that faces scalability issues for larger networks [ 13 , 14 , 15 , 16 , 17 , 18 , 19 ] . From the power grid perspective, placing the EV charging stations introduces significant power demands on the grid, which can fluctuate due to unpredictable charging needs, affecting the distribution of active and reactive power. To ensure reliability and safety, voltage and current levels at any bus must remain within certain ranges to prevent network overloads [ 20 , 21 , 22 ] . There are several gaps in the current literature. The primary challenge is that spatial-focused approaches often overlook power grid considerations and power grid-centric methods overlook spatial aspects. Secondly, estimating EV charging demand is not only complex but also difficult to scale up. To address this, recent works such as [ 9 , 12 , 10 ] have explored data-driven approaches for EV charging demand prediction using urban informatics and mobility data. Building on this, a multidisciplinary two-phase approach is developed in this work. The contributions of this work are as follows: • A data-driven point-of-interest methodology is employed to predict EV charging demand. The target area is divided into a grid, and EV charging demand is estimated for representative cells. Data is scraped from the Google Maps dataset to predict charging demand. • A maximum coverage model is formulated to identify optimal charging locations, ensuring that all demand values are met. The capacity of each charging station is determined. • OpenDSS simulations and power flow calculations are performed to validate the capability of the distribution network to accommodate the additional EV charger loads. We conduct power flow analyses both before and after the integration of new chargers The rest of the paper is arranged as follows. Section II discusses the prediction model, maximum coverage model, and algorithm outline. Section III presents the numerical implementation on a case involving the Dallas Fort Worth (DFW) area and concluding remarks. II Methodology This section outlines the methodology for establishing new charging stations. We begin by presenting the prediction model, used to determine charger demand in Phase 1. Phase 2 involves solving the maximum coverage model over an integrated power grid. Finally, we ensure that the newly established chargers do not violate any power grid metrics. II-A Phase I : Estimation of EV charging demand A primary objective here is to identify the EV charging demand within a specified area. We start by analyzing the relationship between EV charger distribution and nearby amenities, defined as points-of-interest (POIs).This analysis employs a data-driven approach, leveraging machine learning models to identify key patterns and needs. First we finalize the region and then chargers within this region are identified through Google Maps, excluding stations with zero or unknown port counts. In constructing the dataset, the necessary data is extracted, cleaned, and further balanced using sampling techniques. All potential EV charging station candidates are selected for the training process. The EV charger capacity at each potential location is predicted using the extreme gradient boosting (XGB) [ 23 ] model. After the model is fitted to the training data, its performance is assessed using two key metrics: Mean Squared Error (MSE) and R^{2} score. Further details with the features and the dataset used, please see Section III . II-B Phase II : Maximum coverage problem formulation For any given demand point, there are two possible choices. The first option is to go to a nearby charging station where an additional charging port can be added at a lower cost per port. The second option is for the demand point itself to become a charging station, thereby meeting both its own demand and that of nearby points. To achieve this, we define four sets of decision variables. Decision variables: For convenience, we define \mathcal{J} as the set of all potential locations of establishing a charger and existing chargers, where \mathcal{J}:=\mathcal{J}_{1}\cup\mathcal{J}_{2} . Here, \mathcal{J}_{1} is the set of all candidate locations for establishing new chargers, and \mathcal{J}_{2} is the set of all existing chargers. \mathcal{P} is the set of all buses within the 8500 power grid [ 24 ] . A binary variable y_{j} is 1 when a charging station is established at a new site for j\in\mathcal{J} . Superscript “1” ( y^{1}_{j} ) denotes a newly opened station, while “2” stands for an existing one ( y^{2}_{j} ). Decision variable, x^{1}_{i,j} , represents the amount of EV demand traveling from demand point i to charging station

Read the complete text