ORCID

0009-0005-9692-0614

Keywords

Alternative Fuels of Buses, Optimization, Transit, Machine Learning, Life Cycle Analysis, Smart and Sustainable Transportation

Abstract

Public transportation is undergoing a critical transition as agencies seek to balance environmental responsibility, operational efficiency, and economic viability. Alternative fuel buses (AFBs), including electric, compressed natural gas (CNG), propane, and biodiesel, offer opportunities to reduce emissions and advance sustainability, yet their adoption raises complex trade-offs. High upfront costs, infrastructure constraints, range limitations, and variable real-world performance complicate decisions for fleet managers and policymakers. This dissertation addresses these challenges by integrating real-world operational data, life-cycle modeling, and machine learning within a multi-objective optimization framework to evaluate and optimize AFB deployment. Using detailed telematics and GPS datasets from Central Florida's transit fleets, this research examines operational dynamics such as route type, seasonal variation, speed profiles, idling, and HVAC demand. Emissions and energy use are quantified through the MOVES simulator, while GREET and AFLEET models extend the analysis to life-cycle impacts, including greenhouse gases, regulated pollutants, externality costs, and economic outcomes. Machine learning methods, including predictive modeling, classification, and spatio-temporal analysis are applied to capture patterns in energy consumption and emissions, enabling fine-grained insights that traditional models cannot provide. The results highlight strong dependencies between bus technology, route type, and operating conditions. Electric buses demonstrate the lowest emission intensities but are highly sensitive to charging infrastructure and grid mix. Propane buses exhibit consistently low operational energy consumption and maintenance costs, while diesel and CNG buses reveal trade-offs between greenhouse gas and criteria pollutant emissions. Statistical models and predictive machine learning approaches confirm that roadway type (downtown, urban, limited access) is a critical driver of emissions variability, with downtown corridors producing the highest rates. To move beyond descriptive comparison, a multi-objective optimization tool was developed to evaluate fleet mixes under different cost and emission priorities. Scenarios reveal that diesel minimizes upfront investment, propane provides balance across cost categories, and electric buses dominate when emission reduction is prioritized. This tool equips agencies with a decision-support system to align fleet strategies with sustainability goals, budget realities, and equity considerations. By bridging machine learning, emissions modeling, and optimization, this dissertation contributes a novel, data-driven framework for sustainable transit planning. The findings not only advance academic understanding of AFB systems but also deliver practical pathways for agencies to reduce environmental impacts, control costs, and accelerate the transition to cleaner urban mobility.

Completion Date

2025

Semester

Fall

Committee Chair

Abou-Senna, Hatem

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Civil, Environmental and Construction Engineering

Format

PDF

Identifier

DP0029813

Document Type

Thesis

Campus Location

Orlando (Main) Campus

Download

Share

COinS