Datacenter Cooling Control Using Predicted Thermal Load
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Datacenters face significant power consumption challenges due to cooling requirements, with cooling systems reacting slowly to changes in thermal loads, leading to inefficient energy use and potential damage from temperature peaks or drops, and existing solutions do not effectively manage thermal inertia to maintain an optimal temperature range.
Innovation Solution
A method and system that predict thermal loads from a workload scheduler to proactively activate or deactivate cooling resources, considering thermal inertia, to maintain the datacenter temperature within an optimal range, using a main controller, control interface, scheduler application, and workload-based logic to adjust cooling device operations based on predicted thermal loads and user-defined thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Temperature
If the cooling system is activated immediately when temperature reaches a threshold, then temperature peaks are avoided, but the cooling system reacts slowly due to thermal inertia causing temperature oscillations and excessive cooling
Solution Approach 1:
The system performs preliminary action by predicting future thermal loads based on workload schedules and activating cooling resources in advance before temperature thresholds are reached. This proactive approach prevents temperature peaks while avoiding the need for aggressive reactive cooling that wastes energy.
Solution Approach 2:
The system implements feedback by continuously monitoring actual temperature measurements and comparing them with predicted temperatures, then adjusting cooling resource activation accordingly. This closed-loop control optimizes cooling energy consumption while maintaining temperature stability.
2Reliability
If the threshold limit is set significantly below critical temperature to avoid temperature peaks, then machine safety is improved, but more heat is conducted than necessary increasing energy consumption
Solution Approach 1:
The system sets an appropriate threshold limit while performing preliminary cooling activation based on predicted thermal loads. This ensures machines remain safe from temperature peaks without requiring the threshold to be set excessively low, thereby avoiding unnecessary energy consumption.
Solution Approach 2:
The system dynamically adjusts cooling activation timing and intensity based on predicted thermal load parameters rather than relying on a fixed low threshold. This optimizes the balance between machine safety and energy efficiency.
3Productivity
If all machines are activated at the same time, then productivity is improved, but temperature peaks occur requiring aggressive cooling
Solution Approach 1:
The system handles simultaneous machine activation by predicting the aggregate thermal load and activating cooling resources in advance. This allows high machine utilization without temperature peaks, as cooling is prepared beforehand rather than reacting to temperature rises.
4Use of energy by stationary object
If the cooling system is turned off when temperature falls below threshold, then energy consumption is reduced, but thermal inertia causes negative temperature peaks
Solution Approach 1:
The system predicts when thermal loads will decrease and deactivates cooling resources in advance before temperatures fall below thresholds. This prevents negative temperature peaks caused by thermal inertia while still reducing cooling energy consumption appropriately.
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
This approach reduces energy consumption by smoothing cooling resource usage, avoiding temperature peaks, and maintaining optimal conditions, thereby lowering operational costs and preventing damage from thermal extremes.
Implementation Method 1
The machines assembled in the datacenter produce a great amount of lost heat that has to be conducted from the machines
Implementation Method 2
The cooling systems of the datacenter have to face some initial inertia to change the temperature
Data Source
AI summary
A method for regulating the temperature of a datacenter within an optimum temperature range includes predicting, using a computing device, a thermal load from a workload scheduler containing information on machines assembled in the datacenter to be turned on and/or off during a particular time period, and the thermal load of the datacenter associated with the work of the machines within the particular time period; and controlling at least one cooling system of the datacenter based upon the predicted thermal load within the particular time period under consideration of the thermal inertia of the datacenter by at least one of activating, controlling, and deactivating cooling resources of the cooling system in advance to maintain the temperature of the datacenter within the optimum temperature range.