Date of Award

December 2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical Engineering

First Advisor

Hossein Salehfar

Abstract

The rise of distributed generation and the electrification of transportation are driving fundamental changes in the electric grid. The growing adoption of electric vehicles and solar rooftops introduces significant uncertainty in demand patterns, posing new challenges for the operation of distribution systems. Residential distribution systems are particularly vulnerable to the impacts of residential solar photovoltaics (PV) generation and electric vehicle charging, as they introduce significant variability and localized stress on the grid. To ensure reliable and efficient operation, utilities must account for the long-term growth of these technologies in their planning processes. While substantial progress has been made in assessing hosting capacity for community and utility-scale solar PVs, the effects of widespread residential solar PV and electric vehicle (EV) adoption on distribution system performance remain largely underexplored. This gap exists due to the absence of computational models that link distribution system operations with the underlying factors driving individual-level adoption of electric vehicles and rooftop solar PVs. The challenge lies in capturing uncertainty on two fronts: the adoption of distributed energy resources and electric vehicles, and the variability of solar-based generation and EV charging demand, as they depend on climate conditions and user behavior.

This dissertation explores the key socio-economic, perception and geographical factors driving EV and residential solar PV adoption in the US. Household traits and perception-based variables emerge as important indicators of potential electric vehicle and rooftop PV adoption. While multiple vehicles, homeownership, higher household income, gas prices, and winter temperatures drive EV adoption in both rural and urban areas, urban EV adoption is particularly higher among households with young or no children. In contrast, rural households with very high incomes, technological awareness, and high concerns about fuel cost are more likely to adopt EVs.

Social interaction or peer influence emerged as the strongest driver of residential PV adoption. It implies that peer influence plays an important role along with financial and technological factors in driving PV adoption. Similar to EV adoption, homeownership, moderate income, and higher education are common characteristics of PV adopters. Potential adopters are invested in PV systems’ ability to reduce energy costs, provide backup power, and benefit the environment, whereas non-adopters often view solar as politically driven and express low trust in installers. However, both EV and PV adoption are not uniform across the US. Therefore, the adoption datasets are typically imbalanced, with relatively few adopter records compared to non-adopters. To address this issue, this dissertation work introduces an ensemble learning approach to improve estimation accuracy and the reliability of adoption projection.

Assessing the hosting capacity of distribution systems requires the development of a detailed simulation model with the grid topology and variations in load demand. In this dissertation research, a distribution system model of a rural feeder is constructed using utility-provided network and consumer demand data. The model reflects typical demand patterns, with weekdays exhibiting morning and evening peaks. A stochastic hosting capacity analysis technique is developed in this dissertation that captures the impact of uncertain residential solar PV and EV adoption along with EV charging load and PV generation by assigning probability distributions to key input variables. The development process involves deriving adoption patterns from ensemble classification model outputs, while regional travel behavior, charging preferences, and weather conditions are used to build multiple EV charging and solar generation profiles. Each stochastic scenario simulates a unique set of conditions under operational constraints, and the outcomes are compared to evaluate how changes in these parameters influence hosting capacity across the distribution system. A regression analysis is further conducted to identify the relationship between adoption-related factors and the load and generation hosting capacity of the distribution grid.

The results indicate that reverse power flow and overvoltage are the primary constraints for PV hosting capacity, whereas load hosting capacity is limited by either thermal loading or undervoltage issues. Single-phase sections farther from the substation are more vulnerable to undervoltage issues due to high EV charging loads. The regression analysis implies hosting capacity of three-phase sections is directly influenced by feeder-level PV and EV adoption rates, whereas the hosting capacity of single-phase sections distant from the substation depends on local PV and EV adoption rates and charging parameters. Finally, this comprehensive approach enables a more realistic and data-driven evaluation of hosting capacity, taking into account the complex and uncertain nature of future distributed generation and electric vehicle integration in distribution systems.

Download

Share

COinS