Data Center Cooling Performance Prediction During Transient Events

The system performs preliminary computational modeling and simulation to predict cooling performance during transient events before they occur. By using CFD and potential flow models to pre-calculate temperature and airflow patterns under various failure scenarios, the system enables proactive planning and mitigation strategies without requiring complex real-time monitoring during actual events.

The patent introduces computational models (CFD, potential flow models) as intermediaries between the physical data center system and the management software. These models serve as virtual replicas that can be simulated repeatedly under different conditions, providing prediction capabilities without directly monitoring the actual physical system during transient events, thus reducing operational complexity.

The computational domain is segmented into multiple zones and time intervals. The transient event period is divided into discrete time steps, and the physical space is divided into computational cells. This segmentation allows the complex CFD simulations to be performed in manageable increments, balancing prediction accuracy with computational efficiency by focusing resources on critical areas and time periods.

The system applies computational models selectively rather than uniformly across the entire data center. Potential flow models are used for regions where detailed accuracy is critical, while simplified models are applied to less critical areas. This partial application of complex computations only where necessary reduces overall computational time and resource consumption while maintaining required prediction precision.