Data Center Cooling Infrastructure Management During Ramp-Up
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face inefficiencies in thermal management during the ramp-up phase, leading to high Power Usage Effectiveness (PUE) due to underutilized cooling infrastructure designed for full occupancy, resulting in energy wastage and hot spots.
Innovation Solution
The implementation of a system and method that uses Computational Fluid Dynamics (CFD) tools and a PUE predictor model to optimize data center configuration and operation by mapping design and operational parameters with predefined partitions and rack placements, dynamically maintaining PUE within an efficient range during the ramp-up period through efficient cooling infrastructure management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cooling infrastructure is designed for full occupancy, then cooling capacity is sufficient for maximum load, but energy efficiency deteriorates during ramp-up phase due to underutilization
Solution Approach 1:
The data center is divided into multiple zones or sections that can be independently controlled and optimized. During ramp-up phase, only the occupied zones are actively cooled with appropriate capacity, while other zones remain in standby or minimal cooling mode. This segmentation allows the cooling infrastructure to match actual heat load distribution, preventing energy waste from cooling empty spaces while maintaining sufficient capacity for future full occupancy.
2Temperature
If cooling capacity is increased to address hot spots, then temperature uniformity improves, but cooling efficiency deteriorates due to overcooling non-problem areas
Solution Approach 1:
Different cooling strategies and capacities are applied to different locations within the data center based on local heat generation patterns. Hot spots receive targeted cooling with higher capacity, while areas with lower heat loads receive proportionally reduced cooling. This localized approach maintains temperature uniformity across the facility without the energy penalty of uniform overcooling, as cooling resources are precisely matched to actual thermal demands in each zone.
3Temperature
If supply temperature of cooling units is decreased to cool problem areas, then temperature control in hot spots improves, but overall cooling efficiency deteriorates
Solution Approach 1:
Supply temperature is optimized locally for each zone rather than using a single uniform temperature for the entire data center. Hot spots receive cooler air at lower supply temperatures to effectively remove excess heat, while other zones receive air at higher, more energy-efficient supply temperatures appropriate to their lower heat loads. This localized temperature optimization resolves hot spots without the energy penalty of system-wide overcooling.
4Temperature
If cooling capacity is increased near problem areas, then hot spot temperature control improves, but energy consumption increases due to concentrated cooling power
Solution Approach 1:
The cooling system is segmented into multiple independent cooling zones or modules, each with capacity matched to the local heat load. Instead of one large centralized cooling system operating at partial efficiency, multiple smaller cooling units are distributed throughout the data center, with each unit sized appropriately for its local zone. This segmentation allows hot spots to receive adequate cooling capacity locally while avoiding the energy waste of a oversized centralized system cooling empty areas.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and system is disclosed for maintaining Power Usage Effectiveness (PUE) of a new data center constant or within narrow range around efficient level during ramping up stage of the data center. The method comprises of capturing a plurality of design and operational parameters of the data center, computing an efficient design for the data center at full occupancy, and maintaining the Power Usage Effectiveness constant or within narrow range around efficient level at a current occupancy during a ramp up period of the data center.