Design of Power Supply Redundancy Architectures for High-Power-Density Computing and Reliability Analysis
DOI:
https://doi.org/10.70088/57awt591Keywords:
power redundancy, high-power-density computing, reliability engineeringAbstract
High-power-density computing systems-such as AI accelerators, high-performance GPUs, and dense server racks-demand unprecedented levels of electrical reliability while operating under strict thermal and spatial constraints. Power supply redundancy is no longer a mere availability enhancement but a foundational requirement for fault containment, graceful degradation, and sustained operational integrity. This research systematically designs, models, and validates three distinct redundancy architectures-N+1 parallel modular, hot-swappable dual-bus, and distributed hybrid-isolated by galvanic separation and coordinated via digital health monitoring. Each architecture is evaluated across five core reliability metrics: mean time to failure (MTTF), fault coverage ratio, failover latency, thermal derating margin, and system-level power density penalty. Experimental validation was conducted using 3.3 V/200 A DC-DC modules under accelerated stress testing (120 C ambient, 150% load cycling) over 20,000 hours. Results demonstrate that the distributed hybrid architecture achieves a 42% improvement in MTTF over conventional N+1 configurations while maintaining 93.1% power density efficiency relative to a non-redundant baseline. Failover latency remains below 100 μs for 94.9% of recorded fault events (median 62 μs, maximum 138 μs), and thermal derating margins remain above 9C across all tested architectures. Crucially, architectural trade-offs are quantified: increased control complexity correlates with higher initial cost but yields measurable gains in fault resilience and service lifetime. These findings provide a rigorous, measurement-grounded framework for selecting, sizing, and certifying power redundancy solutions tailored to next-generation compute infrastructure where reliability, density, and responsiveness are co-optimized rather than sequentially prioritized.References
H. Moradisizkoohi, *Efficient, High Power Density, Modular Wide Band-gap Based Converters for Medium Voltage Application*. Miami, FL: Florida Int. Univ., 2020.
Z. Hasan, “Redundancy for data centers,” ASHRAE J., vol. 51, no. 2, p. 52, 2009.
Y. Lu, Z. Tong, J. Yang, Z. Yu, M. Huang, and X. Mao, “Multi-chip multi-phase DC–DC converters for AI power: a ring, a chain, or a net, independent or master-slave?,” J. Semicond., vol. 46, no. 7, p. 070201, 2025.
F. Pittino, F. Beneventi, A. Bartolini, and L. Benini, “A scalable framework for online power modelling of high-performance computing nodes in production,” in Proc. Int. Conf. High Perform. Comput. Simul. (HPCS), Oradea, Romania, Jul. 2018, pp. 300–307.
J. H. Jeon, S. H. Jeon, and E. S. Lee, “Optimal design and control of high power density bi-directional CLLC converter with integrated transformer for wide-range EV battery charging,” IEEE Access, vol. 13, pp. 144662–144677, 2025.
N. Simpson, H. Felton, P. Munagala, and P. Mellor, “Direct winding thermal management schemes enabled by additive manufacturing,” in Proc. IEEE Energy Convers. Congr. Expo. (ECCE), Nashville, TN, Oct. 2023, pp. 4524–4531.
G. Rajendran et al., “Energy and exergy optimization of multilevel power converters in renewable energy systems: A review of AI-driven control strategies,” J. Energy Storage, vol. 181, p. 124299, 2026.
G. Refai-Ahmed et al., “CCECE 2021 invited paper: Holistic performance, reliability and thermal understanding of HPC real utilization on silicon architecture,” in Proc. IEEE Can. Conf. Electr. Comput. Eng. (CCECE), Niagara Falls, ON, Canada, Sep. 2021, pp. 1–8.
L. Smirnova, *Electromagnetic and Thermal Design of a Multilevel Converter with High Power Density and Reliability*. Moscow, Russia: Inst. of Power Eng., 2015.
R. Ravindran and A. M. Massoud, “State-of-the-art DC-DC converters for satellite applications: A comprehensive review,” Aerospace, vol. 12, no. 2, p. 97, 2025.
M. H. Ahmed, C. Fei, V. Li, F. C. Lee, and Q. Li, “Startup and control of high efficiency 48/1V sigma converter,” in Proc. IEEE Energy Convers. Congr. Expo. (ECCE), Cincinnati, OH, Oct. 2017, pp. 2010–2016.
S. Lee, *Topological Analysis, Design Optimization, Advanced Modulations, and Control of High-Power Density WBG Device-Based Current Source Inverter for Transportation Applications*. Madison, WI: Univ. Wisconsin-Madison, 2024.
S. Ci, N. Lin, Y. Zhou, H. Li, and Y. Yang, “A new digital power supply system for fog and edge computing,” in Proc. 14th Int. Wireless Commun. Mobile Comput. Conf. (IWCMC), Limassol, Cyprus, Jun. 2018, pp. 1513–1517.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Yichao Ma (Author)

This work is licensed under a Creative Commons Attribution 4.0 International License.





