Abstract

In the past ten years, the electronics industry has seen a sustained demand to reduce thermal resistances while increasing volumetric heat removal. The limitations of traditional air-cooled systems have caused a major industry shift toward liquid cooling. To maintain beam quality, high output efficiency, and continuous uptime, high-powered lasers require cooling to manage the waste heat of 5-55 kW. Small modular nuclear reactors (SMRs) have a capacity of 350 MW or less per unit and operate at an efficiency of 33.4% to 45%. As a result, 55% to 66% of the heat generated must be dissipated. Within data centers, a single CPU rack can draw 140,000 W, and individual CPU chips can exceed 1,000 W/cm² in continuous workloads. This approaches the absolute limit of air cooling, with cooling systems consuming 40–60% of the total power draw. Consequently, facilities are transitioning to direct liquid cooling to increase efficiency. Battery cooling systems must maintain operating temperatures of 25°C and 40°C for maximum performance while managing a heat flux on the scale of 10 W/cm2. Military power amplifiers generate extreme power densities, with heat fluxes that generally exceed several hundred W/cm2. This thesis generates techniques for effective, direct, and accurate modeling of boiling phenomena without requiring finetuning of empirical coefficients. These techniques are grounded in fundamental principles of energy conservation, phase change, and numerical methods. They are compatible with conventional discretization techniques and support customization of numerical software. Techniques for customizing Ansys-Fluent are discussed, including specialized macros for computing normal vectors, volume fractions, cell temperatures near the interface, and distances from cell centers to the interface. The one-cell algorithm for sharp interface and mass transfer (OCASIMAT) directly calculates the temperature gradient using a probe method and linearly interpolates the exact temperature of the mixture cell. The large-coefficients method introduces a source term into the interface-cell temperature calculation, which forces the interface-cell temperature to equal the saturation temperature. The neighboring cells see the sharp interface temperature without solving additional differential equations. The interface area for mass transfer is calculated using the volume fraction gradient and the mixture cell’s normal vector. To maintain a sharp interface, if a mixture cell that is not located next to a vapor cell, its surface area is transferred to a neighboring mixture cell to maintain mass conservation and prevent deformation. In addition, adaptive mesh refinement algorithms were developed to ensure a low number of computational cells. A macro is implemented to update the Ansys-Fluent User-Defined Scalar (UDS) transport equation, assigning 0 to the liquid and vapor cells and 1 to the interface cells. The diffusion term in the scalar equation creates a smooth transition of value from 1 to 0 into the neighboring liquid and vapor regions; a field variable marks specific cells within this range for refinement. To accommodate modifications to Ansys Fluent, User-Defined Functions (UDFs) were customized within the software. UDFs were used to identify mixture and gradient cells, declare source terms in the energy equation, declare mass transfer in the VOF equation, calculate and declare a source term to fix the temperature of the mixture cells, and define the thermal diffusivity of the mixture cells. The simulations leveraged parallel processing with multiple cores, dividing the domain to handle the large number of computational cells. Additionally, parallelizing UDFs increased computational speed.  This work describes spherical bubble growth under constant mass flux and temperature-driven mass transfer, considering both regular and adaptive meshes. Lastly, the modeling of bubble dynamics during single- and multiple-bubble merging over a heated surface is presented. Theoretical equations for spherical bubble growth under adiabatic conditions and in superheated liquid, along with experimental images, are used to verify and validate the developed simulation framework. Results demonstrate the feasibility of the described methods for capturing the main mechanisms of heat and mass transfer, including the effects of vapor expansion due to high density ratios. The proposed methods enable accurate capture of thermal films near the interface with a relatively low number of computational cells for three-dimensional spherical bubbles. Spherical bubble growth with constant mass flux accurately simulated 3D spherical bubble growth, with a relative error of less than 3%, satisfying mass conservation. Spherical bubble growth with temperature-driven mass transfer reported temperature gradients with cells reporting 12% or less error, matching the theoretical calculation of the gradient 1.5 x 10^6 K/m. Mechanisms of heat transfer for bubble growth over a heated surface of a single bubble captured high temperature gradients near the contact line and cold liquid traveling toward the heated surface upon bubble departure, revealing an influence region extending 2.7 times the departure diameter and a local heat transfer coefficient that averages 15,000 W/m²-K within a 500 μm range of the interface. Wall shear and heat transfer coefficients provided evidence of trapped liquid where localized increases in both the wall shear and heat transfer occurred. The model predicted a heat transfer coefficient of 13,150 W/m²-K near bubble departure for three merged bubbles over a heated surface, revealing an overall influence region of 10.8 mm, 3.1 times the equivalent departure diameter. These findings demonstrate that heat transfer mechanisms during nucleate boiling can be accurately captured and quantified using the developed sharp interface simulation framework at significantly reduced computational cost. The concepts and methods developed in the research work provide an effective framework for accurately modeling boiling phenomena near the interface. The proposed algorithms and numerical techniques enable accurate three-dimensional modeling of boiling flows by customizing numerical software to be robust and compatible with standard discretization methods. This research work outlines a high-fidelity approach for acquiring the fundamental knowledge needed to develop sophisticated systems that exploit hydrodynamic, thermal, and phase-change phenomena at sharp interfaces.

Publication Date

7-22-2026

Document Type

Dissertation

Student Type

Graduate

Degree Name

Mechanical and Industrial Engineering (Ph.D)

Department, Program, or Center

Mechanical Engineering

College

Kate Gleason College of Engineering

Advisor

Isaac Perez-Raya

Advisor/Committee Member

Satish Kandlikar

Advisor/Committee Member

Jennifer O'Neil

Comments

This thesis has been embargoed. The full-text will be available on or around 8/17/2027.

Campus

RIT – Main Campus

Download

Share

COinS