Phase-Balanced Power Distribution for Data Center Loads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing power distribution systems in large computing facilities, such as data centers, face challenges in efficiently managing power consumption across multiple phases, leading to inefficiencies and potential waste of available power due to imbalanced electrical current consumption.
Innovation Solution
A power distribution system and method that includes monitoring and balancing electrical current consumption across multiple phases by adjusting power distribution to computing devices, using sensors and controllers to ensure balanced consumption within a threshold, and implementing strategies such as powering up, down, or reconfiguring connections to achieve phase balance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If power is distributed to multiple computing devices across multiple phases without monitoring, then the system is simple to operate, but the electrical current consumption becomes unbalanced leading to power waste
Solution Approach 1:
The system continuously monitors electrical current consumption across multiple phases and uses this feedback to detect imbalances. When an imbalance is detected (difference exceeds threshold), the system automatically redistributes computing devices between phases to restore balance, preventing power waste while maintaining operational simplicity through automation
Solution Approach 2:
The power distribution system performs self-balancing by automatically detecting phase imbalances and redistributing computing devices without human intervention. The system serves itself by monitoring its own state and correcting deviations from optimal operation, eliminating the need for manual power management while reducing energy waste
2Productivity
If the system monitors and balances electrical current consumption across phases, then power efficiency is improved, but the device complexity increases
Solution Approach 1:
The system implements continuous monitoring of electrical current consumption on each phase, comparing values against each other and a predefined threshold. This feedback mechanism enables automatic detection of imbalances and triggers redistribution actions only when necessary, improving power efficiency while minimizing the complexity overhead through event-driven operation rather than constant active management
Solution Approach 2:
The system dynamically changes the distribution parameter (which computing devices are connected to which phases) based on monitored electrical current consumption values. By adjusting the assignment of computing devices to different phases, the system optimizes power efficiency while using simple threshold-based logic to manage complexity
3Ease of operation
If computing devices are distributed across phases without balancing, then the system is easier to set up, but electrical current consumption becomes unbalanced
Solution Approach 1:
The system maintains ease of setup by allowing computing devices to be connected to any phase without manual configuration, then automatically self-corrects any resulting imbalances through monitoring and redistribution. This self-service approach preserves operational simplicity while preventing energy loss from unbalanced consumption patterns
Data Source
AI summary
A method for power distribution may include providing, via a power distribution unit, power to a plurality of computing devices, the computing devices including a respective processor and memory; monitoring electrical current consumption of the plurality of computing devices for a plurality of phases; determining if a difference between the electrical current consumption of a phase of the plurality of phases and the electrical current consumption of one or more other phases of the plurality of phases exceeds a threshold; and/or balancing the electrical current consumption of the plurality of phases, if the difference exceeds the threshold, such that the difference does not exceed the threshold.