Dynamic Power Converter Control for Data Center Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computer data centers face increasing power consumption and cooling demands due to the growing number of servers, leading to inefficiencies in AC to DC power conversion, particularly when servers operate at partial loads, resulting in higher electrical losses and costs.
Innovation Solution
Implementing a dynamic power management system that detects power consumption and dynamically turns on and off AC to DC converters to maintain them at maximum efficiency points, using load balancing circuitry and monitoring subsystems to optimize power allocation and reduce overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple AC to DC converters are operated in parallel for redundancy and capacity, then power supply reliability and capacity are improved, but power conversion efficiency deteriorates when servers operate at partial loads
Solution Approach 1:
The system dynamically adjusts the number of active power converters based on real-time power consumption detection. The management subsystem monitors server power usage and dynamically enables or disables individual converters to match the load requirements, transitioning from static full-redundancy operation to dynamic adaptive operation. This resolves the contradiction by maintaining reliability through available standby converters while improving efficiency by activating only the necessary number of converters for the current load.
Solution Approach 2:
The invention changes the operational parameter of power converters from a fixed state (all converters always on) to a variable state (converters dynamically enabled/disabled). By detecting power consumption levels and adjusting the number of active converters accordingly, the system optimizes the balance between reliability and efficiency. When power consumption is low, fewer converters are activated; when power consumption increases, more converters are enabled to meet the demand.
2Power
If all AC to DC converters are kept enabled for maximum capacity, then power supply capacity is improved, but electrical losses increase due to operation away from maximum efficiency points
Solution Approach 1:
The system implements dynamic power converter management where the number of active converters is continuously adjusted based on detected power consumption. The management subsystem monitors real-time power usage and dynamically enables or disables individual converters to operate near their maximum efficiency points. This dynamic adjustment maintains adequate power supply capacity while minimizing electrical losses by avoiding operation of excessive converters at inefficient partial loads.
Solution Approach 2:
The power management system automatically detects power consumption levels and self-adjusts the number of active converters without external intervention. The management subsystem continuously monitors power usage and autonomously enables or disables converters based on the detected load, allowing the system to self-optimize its efficiency while maintaining adequate capacity.
3Reliability
If redundant power supplies are deployed on a per-system basis with N+1 redundancy, then system reliability is improved, but overall power efficiency deteriorates due to multiple converters operating at low individual loads
Solution Approach 1:
The invention merges the management of multiple power converters across different server systems into a unified dynamic management approach. Instead of each server independently managing its own redundant converters (N+1 per system), the system consolidates power conversion resources and dynamically allocates them based on aggregate power consumption. This merging allows converters to be shared across multiple systems, enabling them to operate at higher individual loads and improving overall efficiency while maintaining reliability.
Solution Approach 2:
The system transitions from static per-system N+1 redundancy to dynamic aggregate power management. The management subsystem monitors total power consumption across all servers and dynamically adjusts the number of active converters based on the aggregate load rather than individual system requirements. This dynamic approach allows converters to operate more efficiently by consolidating load while maintaining adequate redundancy for system reliability.
Data Source
AI summary
Dynamically managing power consumption in a computer system having at least two parallel power converters in order to improve efficiency. A maximum power capacity for each of the power converters is determined and then power consumption of the computer system is monitored. If the power consumption of the computer system can be provided by less than all of the parallel power converters then one or more of the power converters is turned off, such that a reduced number of parallel power converters remains turned on. A reduced maximum power capacity of the reduced number of parallel power converters is determined and a power cap value is set for the computer system that is less than or equal to the reduced maximum power capacity. The computer system is throttled at the power cap to prevent power consumption of the computer system from exceeding the power cap value.