Study On Power Supply Redundancy Planning To Improve Data Canter Reliability
Abstract
The data center is a critical infrastructure that demands high reliability of power supply. Electrical system failures, including those caused by human error such as the incident in July 2024, can result in significant financial and reputational losses. This study aims to analyze and compare three redundancy configurations of data center electrical systems—N, N+1, and 2N—in order to enhance the operational reliability of the JTT Pasiranji Data Center. The research method was conducted through planning studies, dual-bus electrical system design analysis, and validation via a series of staged commissioning tests (CX 1 to CX 5), which included insulation testing, UPS and ATS switching time testing, generator performance testing, and failover tests under three failure scenarios. The results show that the 2N configuration provides an availability level of 99.982%, equivalent to the Uptime Institute Tier III standard, significantly surpassing N+1 (99.741%) and N (99.671%). Switching time tests confirmed that the double-conversion online UPS operates without switching gap (0 ms), ATS performs automatic transfer in an average of 8.2–8.5 seconds, and generators reach full load readiness within 11.3–11.8 seconds. Failover tests across all three failure scenarios verified that the 2N configuration maintained 100% server uptime without service interruption. This study also found that system reliability is not solely determined by technical design superiority, but is also influenced by human error, adherence to commissioning procedures, real-time monitoring, and preventive maintenance.