Head of NUS Mechanical Engineering & Executive Director of ESI | Expert in Sustainable AI Data Center Cooling | Keynote Speaker and Board Member
Innovation in Data Center Design & Operations Summary: Digital twins just graduated from design toy to operational backbone. At Thésée Data Center (Tier IV), Cadence’s Reality Digital Twin runs from day-zero design into day-two ops, giving facilities and IT a shared, physics-based “single source of truth.” • What’s new: A continuously updated, high-fidelity model integrates BMS and capacity tools to monitor airflow, temperatures and energy—then simulate “what-ifs” (failures, upgrades, layout changes) before touching hardware. • Operational gains: Lower PUE via targeted fixes, faster incident diagnosis, safer change management, and remote operations (e.g., France Télévisions) from central offices. • Design → Ops continuity: The same model built for planning is used daily in operations—closing the infamous gap between CFD at build time and tribal knowledge at runtime. • AI factory ready: Library support now includes NVIDIA DGX SuperPOD with GB200—teams can pre-validate space, power, liquid-cooling hydraulics/temperatures, and cost/performance trade-offs at rack-to-campus scale. • Why it matters: As AI drives step-changes in density and transient behavior, DCs need scenario planning for thermal, electrical, and controls—digital twins make this rigorous, auditable, and fast. Takeaway: The next wave of AI-class facilities will be model-first: design with physics, commission with data, operate with predictions. That’s how we square uptime, efficiency, and cost at scale. #DataCenters #DigitalTwin #AIFactories #LiquidCooling #ThermalManagement #CapacityPlanning #PUE #EnergyEfficiency #CFD #OperationsExcellence #NVIDIA #GB200 #Cadence #FacilitiesEngineering #Sustainability