Multi-Tier Cooling System With Machine Learning Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing server rack cooling systems lack flexibility in responding to real-time variations in IT load, requiring corresponding changes in the cooling layer, which limits their efficiency and adaptability to changing power consumption and IT requirements.
Innovation Solution
A multi-tier cooling system utilizing a machine learning model to regulate operations based on pre-created profiles and real-time data from sensors, including a vapor container, auxiliary condenser, fluid pump, and compressor, with multiple cooling tiers that can be triggered by various indicators such as pressure, fluid level, and renewable energy availability, allowing for efficient phase change cooling without direct perception of IT load variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the cooling system is tightly coupled with the IT load, then the cooling system can respond to IT load variations, but the system lacks flexibility and requires corresponding changes in the cooling layer for any IT load change
Solution Approach 1:
The cooling system is divided into multiple independent cooling tiers (first cooling tier with first condenser, second cooling tier with second condenser, etc.) that can operate independently or in combination. Each tier has its own condenser and control mechanisms, allowing selective activation based on cooling demands without requiring changes to the entire cooling layer structure.
Solution Approach 2:
The system dynamically selects and activates specific cooling tiers based on real-time cooling demands and environmental conditions. The controller can adjust which cooling tiers are active, allowing the system to adapt its cooling capacity without physical reconfiguration or structural changes to the cooling layer.
2Adaptability or versatility
If multiple cooling tiers are used to accommodate high variations in heat load, then the cooling capacity increases, but the device complexity increases
Solution Approach 1:
The cooling system is divided into multiple independent cooling tiers (first cooling tier with first condenser, second cooling tier with second condenser, etc.) that can operate independently or in combination. Each tier has its own condenser and control mechanisms, allowing selective activation based on cooling demands without requiring changes to the entire cooling layer structure.
Solution Approach 2:
Multiple cooling tiers share common components such as the vapor container, fluid pump, and controller. This multi-functionality allows the system to achieve high cooling capacity variation capability while reducing overall complexity, as components serve multiple tiers simultaneously.
3Measurement precision
If the cooling system uses real-time perception of IT load variations, then the cooling response is accurate, but the system complexity and control difficulty increase
Solution Approach 1:
The cooling system operates autonomously using pre-created IT load profiles and real-time sensor data from the environment. The controller selects and activates appropriate cooling tiers based on this information without requiring direct real-time perception or communication with the IT load, thereby achieving accurate cooling response while maintaining simpler control architecture.
4Adaptability or versatility
If the cooling system is designed to be flexible for changing IT requirements, then the adaptability improves, but the hardware design complexity increases
Solution Approach 1:
The cooling system is divided into multiple independent cooling tiers (first cooling tier with first condenser, second cooling tier with second condenser, etc.) that can operate independently or in combination. Each tier has its own condenser and control mechanisms, allowing selective activation based on cooling demands without requiring changes to the entire cooling layer structure.
Solution Approach 2:
The system dynamically selects and activates specific cooling tiers based on real-time cooling demands and environmental conditions. The controller can adjust which cooling tiers are active, allowing the system to adapt its cooling capacity without physical reconfiguration or structural changes to the cooling layer.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system achieves efficient thermal management with reduced cooling costs, accommodating high variations in heat load and cooling capacity, while simplifying design and operation, and providing redundancy and resource sharing for enhanced resilience.
Implementation Method 1
A multi-tier cooling system utilizes a machine learning model to regulate operations based on pre-created profiles and real-time data from sensors, including a vapor container, auxiliary condenser, fluid pump, and compressor, with multiple cooling tiers that can be triggered by various indicators such as pressure, fluid level, and renewable energy availability, allowing for efficient phase change cooling without direct perception of IT load variations.
Implementation Method 2
an auxiliary system that is positioned above an IT load. The auxiliary system includes a vapor container that receives vapor from the IT load, an auxiliary condenser that receives vapor from the vapor container via a compressor or a vapor valve, and condenses the vapor into liquid to be stored in a liquid container.
Implementation Method 3
The auxiliary system further includes a fluid pump on a cooling loop for cooling the auxiliary condenser
Implementation Method 4
an auxiliary condenser that receives vapor from the vapor container via a compressor or a vapor valve
Data Source
AI summary
The disclosed embodiments provide a cooling system with an auxiliary system that extends a main system. The auxiliary system includes a vapor container that receives vapor from the IT load, an auxiliary condenser that receives vapor from the vapor container via a compressor or a vapor valve, and condenses the vapor into liquid to be stored in a liquid container. The auxiliary system further includes a fluid pump on a cooling loop for cooling the auxiliary condenser, and a cooling controller that includes a machine learning model for regulating operations of the vapor valve, the fluid pump, and the first compressor based on a pre-created profile of the IT load and real-time information from at least one of many sources, including the vapor container and the liquid container. The auxiliary system includes multiple cooling tiers that can be partially trigger or completely trigger based on several indicators collected multiple sensors in the auxiliary system.