Himel Barua | Searcharxiv

arXiv2026

Toward Next-Generation AI Data Centers: Power Delivery Architecture Shifts, Emerging Technologies, and Challenges

The rapid growth of AI workloads is driving unprecedented increases in data center power demand, current transients, and thermal stress, exposing fundamental limitations in traditional 48 V rack architectures, low-voltage AC distribution, and line-frequency transformer interfaces. This paper reviews the three stages of architectural shifts required to support next-generation AI data centers and identifies three enabling technological building blocks: high-voltage conversion-ratio DC/DC converters, facility-level low-voltage DC distribution, and medium-voltage solid-state transformers. The advantages, technical challenges, and potential solutions associated with each building block are reviewed. Finally, future research directions and open challenges are discussed.

eess.SY↗

arXiv2021

Continuum and Molecular Modeling of Chemical Vapor Deposition over Nano-scale Substrates

Chemical vapor deposition (CVD) is a common industrial process that incorporates a complex combination of fluid flow, chemical reactions, and surface deposition. Understanding CVD processes requires rigorous and costly experimentation involving multiple spatial scales, from meters to nano-meters. Numerical modeling of deposition over macro-scale substrates has been conducted in literature and results show compliance with experimental data. For smaller scale substrates, where the corresponding Knudsen number is larger than zero, continuum modeling does not provide with accurate results that calls for implementation of molecular-level modeling techniques. In the current study the finite-volume method (FVM) and direct simulation Monte Carlo (DSMC) method have been coupled to model the reactor-scale flow with CVD around micro- and nano- scale fibers. CVD at fibers with round cross-section is modeled where fibers are oriented perpendicularly with respect to the feedstock gas flow. The DSMC method has been applied to modeling flow around the matrix of nano-scale circular individual fibers. Results show that for smaller diameters of individual fibers with the same filling ratio, the residence time of gas particles inside the fibrous media reduces, and, consequently, the amount of material surface deposition decrease.

physics.flu-dyn↗