D2C Cooling Control Using Server Power Draw for Fast Thermal Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional direct-to-chip (D2C) liquid cooling systems in data centers struggle with delayed responses to rapid changes in IT loads due to thermal resistances and physical distances, leading to inefficient coolant flow rates and excessive energy consumption.
Innovation Solution
A system and method that directly monitors power draws from IT devices to adjust coolant flow and supply temperature in real-time, bypassing delays by using a controller to dynamically manage coolant distribution units (CDUs) based on the highest power draw, ensuring microchip assemblies remain below their temperature thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional D2C liquid cooling systems are used to regulate microchip temperatures, then thermal management is achieved, but the response to rapid IT load changes is delayed due to thermal resistances and physical distances
Solution Approach 1:
The system proactively monitors power draws and predicts future thermal conditions before they manifest as temperature changes. By detecting increased power consumption early and preemptively adjusting coolant flow rates, the system prevents temperature excursions rather than reacting to them, eliminating the delay inherent in conventional reactive thermal management
Solution Approach 2:
The system implements a closed-loop feedback mechanism where power draw measurements from PDUs are continuously fed to the CDU controller, which dynamically adjusts coolant flow rates in real-time. This direct feedback loop based on electrical power consumption enables immediate response to load changes, bypassing the thermal inertia that causes delays in conventional temperature-based control systems
2Reliability
If conservative overcooling is applied to ensure microchips remain below temperature thresholds, then thermal safety is maintained, but energy consumption increases
Solution Approach 1:
The system dynamically adjusts coolant flow rates based on real-time power draw measurements and predicted thermal conditions. Instead of maintaining a static conservative flow rate that ensures thermal safety at all times, the system optimizes flow rates to match actual thermal demands, reducing energy consumption during low-load conditions while maintaining reliability during high-load periods
Solution Approach 2:
The system changes the operating parameters of the cooling system (coolant flow rate, supply temperature) based on measured power draws and predicted junction temperatures. By adjusting these parameters dynamically rather than maintaining fixed conservative values, the system achieves thermal safety while minimizing the energy required for cooling operations
3Reliability
If high coolant flow rates are maintained to handle peak power draws, then thermal management reliability is improved, but energy expenditure increases
Solution Approach 1:
The system applies partial cooling action matched to actual thermal demands rather than excessive cooling at all times. By monitoring power draws and adjusting coolant flow rates to match actual heat generation, the system provides sufficient cooling reliability when needed while avoiding the energy waste of maintaining high flow rates during low-load conditions
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
Enables immediate response to IT load changes, maintaining microchip assemblies within safe temperature ranges while minimizing energy expenditure by avoiding conservative overcooling, thus optimizing thermal management efficiency.
Implementation Method 1
regulating the junction temperature of each microchip assembly by circulating a liquid coolant through the microchip assemblies of the cluster
Implementation Method 2
circulating a liquid coolant through the microchip assemblies of the cluster via a coolant distribution unit (CDU)
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A direct-to-chip (D2C) cooling system maintains microchip assemblies within their optimal junction temperature ranges by circulating liquid coolant into contact with the microchip assemblies for heat transfer therefrom. A coolant distribution unit (CDU) controller directly monitors (via a power distribution unit supplying operating power to the servers) power drawn by each of a set of servers, switches, or other information technology (IT) devices and/or microchip assemblies disposed therein, determining on a continual basis which IT device is currently drawing the most power and directing the CDU of the D2C system to dynamically adjust the flow rate setpoint for the liquid coolant supply based on the current highest power draw.