Graduation Semester and Year
Fall 2025
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Mechanical Engineering
Department
Mechanical and Aerospace Engineering
First Advisor
Dr. Dereje Agonafer
Second Advisor
Dr. Miguel Amaya
Third Advisor
Dr. Aron Ameri
Fourth Advisor
Dr. Yogesh Fulpagare
Fifth Advisor
Dr. Saket Karajgikar
Abstract
High-power-density electronics in high-performance computing (HPC) and artificial intelligence (AI) systems demand highly efficient and reliable heat removal solutions that extend beyond conventional single-phase cooling. This dissertation presents an integrated experimental and numerical framework for analyzing and optimizing system-level pumped two-phase (P2P) cooling architectures spanning the chip-to-chiller thermal path. The research bridges component and system level studies, capturing fluid interactions, loop dynamics, reliability characteristics, and their influence on overall thermal performance metrics and availability. The first chapter focuses on numerical and experimental analyses of advanced air-cooled Triply Periodic Minimal Surface (TPMS) heatsinks. It elucidates how TPMS structures enhance thermal management by optimizing flow dynamics and surface area, demonstrating significant heat dissipation improvements validated across varied operating conditions. In the next chapter, the reliability and availability of pumped two-phase cooling loops are rigorously assessed through a system-level approach incorporating pump dynamics, fluid interactions, and component fault tolerance. Key performance indicators such as mean time between failures and system uptime are quantified, highlighting design strategies that enhance fault tolerance and maintenance efficiency. Dynamic transient thermal responses under time-varying heat loads are characterized experimentally and through coupled numerical-control simulations. These investigations inform the design of optimal control algorithms that reduce thermal overshoot, stabilize temperature setpoints, and enhance energy efficiency without compromising system robustness. Finally, a data-driven modeling approach is developed for two-phase cold plates to predict critical heat flux (CHF) thresholds. The methodology integrates experimental datasets with machine learning techniques to accurately forecast CHF under diverse operational scenarios, thus guiding safe and efficient cold plate design. Thus, this research establishes physics-based, validated modeling tools and insights that enable co-optimization of thermal performance, reliability, and availability in next-generation cooling infrastructures for HPC applications.
Disciplines
Heat Transfer, Combustion | Other Mechanical Engineering
License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Recommended Citation
Pundla, Sai Abhideep, "AN INTEGRATED EXPERIMENTAL AND NUMERICAL AND FRAMEWORK FOR SYSTEM-LEVEL PUMPED TWO-PHASE CHIP-TO-CHILLER ARCHITECTURE THERMAL, RELIABILITY, AND AVAILABILITY ANALYSIS" (2025). Mechanical and Aerospace Engineering Dissertations. 444.
https://mavmatrix.uta.edu/mechaerospace_dissertations/444