Date of Award
12-1-2025
Degree Name
Doctor of Philosophy
Department
Mechanical Engineering
First Advisor
Filip, Peter
Abstract
This thesis investigates automotive brake system improvements through computational fluid dynamics (CFD), experimental testing, additive manufacturing, and machine learning techniques. The primary objective is to develop realistic simulations and eco-friendly brake materials for enhanced performance and sustainability. Initially, a CFD model was developed using ANSYS CFX to analyze brake rotor interactions with aluminum NASA alloy material supplied by the company. Preliminary tribological assessment on a Universal Mechanical Tester (UMT) confirmed the material suitability. Multiple rotor geometries provided were modeled and simulated under FMVSS 135 hot stop conditions. Friction performance was validated through scaled-down UMT tests using commercial brake pads and an SUV baseline setup. Thermal properties and experimentally derived friction coefficients were integrated into simulations to rank rotor designs by airflow and heat dissipation efficiency. This approach achieved company approval, followed by full-scale die casting production and brake dynamometer testing.Second, 3D printing was utilized to prototype rotor designs for cost-effective evaluation of geometric optimizations informed by UMT and CFD results. Scaled-down experiments with printed rotors allowed assessment of heat dissipation effectiveness prior to committing to die manufacturing.Third, to promote sustainability, a design of experiments (DOE) was performed to optimize 3D printing parameters for recycling leftover rotor material. The objective was to match the tribological properties of virgin powder, enabling eco-friendly reuse.Finally, brake pad formulations containing over 60% recycled powder were developed for electric vehicles using a Taguchi L8 DOE combined with artificial neural network (ANN) and random forest (RF) modeling. A closed-loop process iteratively refined formulations and model predictions against physical testing on UMT. This methodology effectively integrates experimental data with predictive models to accelerate tribological material development.
Access
This dissertation is Open Access and may be downloaded by anyone.