Coolant Distribution Unit Pump Control for Filter Clogging Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in liquid-cooled direct-to-chip cooling systems for data centers is the risk of filter clogging leading to catastrophic failures due to increased differential pressure, which current control methods exacerbate, necessitating a solution to ensure filter cleanliness and prevent system failures.
Innovation Solution
Implementing an additional alarm system to selectively shut down coolant distribution unit components when differential pressure exceeds thresholds, coupled with methods to control pump units and filter maintenance, thereby preventing failures and ensuring efficient cooling operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If filters with small pore diameter (25μ-50μ) are used to capture contaminants, then filter effectiveness is improved, but filter reliability deteriorates due to increased clogging and rupture risk
Solution Approach 1:
The system performs preliminary detection of differential pressure trends and predicts filter clogging before it occurs. The controller monitors differential pressure over time and predicts future clogging states, allowing proactive pump shutdown before the filter actually clogs or ruptures, thus maintaining both filter effectiveness and reliability
Solution Approach 2:
The system implements continuous feedback monitoring of differential pressure across the filter. The controller receives real-time differential pressure data, compares it against predicted values, and adjusts pump operation accordingly. This feedback loop enables early warning and preventive action, resolving the contradiction between using fine filters and maintaining system reliability
2Productivity
If pump speed is increased to meet IT load flow requirements, then cooling productivity is improved, but differential pressure across filters increases, accelerating clogging
Solution Approach 1:
The system dynamically adjusts pump speed based on predicted filter clogging status and current cooling demands. Rather than operating at fixed high speed, the controller modulates pump speed to balance cooling productivity with filter protection, shutting down or reducing pump speed when clogging is predicted, thus extending filter service life while maintaining adequate cooling performance
Solution Approach 2:
The controller periodically evaluates differential pressure readings and predicts future clogging states at regular intervals. This periodic assessment allows the system to adjust pump operation in cycles, reducing pump speed or shutting down temporarily to prevent clogging, then resuming normal operation when filters are clean, thereby balancing productivity and reliability
3Device complexity
If a single alarm threshold is used for filter differential pressure, then system simplicity is maintained, but early detection of filter contamination is insufficient
Solution Approach 1:
The system introduces a predictive parameter (predicted differential pressure) based on historical data and trends, in addition to the current differential pressure reading. By monitoring both actual and predicted values against thresholds, the system achieves early contamination detection without requiring complex additional hardware, thus maintaining simplicity while improving reliability
Solution Approach 2:
The system replaces purely mechanical/threshold-based alarm systems with an intelligent predictive algorithm running on the controller. Instead of using multiple physical sensors and complex mechanical switches, the controller uses software-based prediction of differential pressure trends, achieving early failure detection while keeping the physical system simple and maintaining reliability
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 solution effectively prevents catastrophic failures by detecting filter contamination early, reducing the risk of damage to cold plates and other electronic components, and maintaining system integrity through controlled pump operations.
Implementation Method 1
a pump selectively communicating with the heat exchanger by a valve
Implementation Method 2
a heat exchanger; a pump selectively communicating with the heat exchanger
Implementation Method 3
Cool liquid is piped to cold plates that sit directly next to components such as CPUs, GPUs, memory cards, etc. Small fluid channels can carry the cool liquid to each plate, where the liquid draws off the heat from the underlying components.
Implementation Method 4
CDUs are typically installed with filter(s) to capture a diameter of about 25μ (e.g., a white blood cell is 25μ) to 50μ (e.g., human hair ranges from 50 to 100μ) to protect cold plates.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A method of controlling a coolant distribution unit is disclosed. The method includes: starting, by a controller, a pump control mode; measuring, by a sensor, a differential pressure at a filter connected to a pump; determining, by the controller, whether the differential pressure is greater than a first predetermined value; upon determining that the differential pressure is greater than the first predetermined value, determining whether the differential pressure is greater than a second predetermined value, wherein the second predetermined value is greater than the first predetermined value; and upon determining that the differential pressure is less than or equal to the second predetermined value, generating a first output.