Data Center Transient Thermal Modeling for Power Outage Prediction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for predicting transient cooling performance in data centers are inadequate, particularly during utility power outages, as they fail to accurately model the transient temperature changes and cooling system restart dynamics, leading to potential overheating of sensitive electronics.

Innovation Solution

A flow network model is developed to estimate air and cooling fluid temperatures in data centers, incorporating energy balance and heat transfer equations for various components, including chillers, water storage tanks, cooling coils, and IT equipment, to predict temperature changes during transient events like power failures and system restarts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current prediction methods are used, then computational simplicity is maintained, but temperature prediction accuracy deteriorates during transient events

Engineering Contradiction:
Improvetemperature prediction accuracyVSAvoidmodel complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The data center cooling system is divided into discrete thermal zones (IT equipment racks, cooling equipment, water storage tanks, chilled water pipes) with energy balance equations applied to each segment. This segmentation allows accurate tracking of transient temperature changes in each component while maintaining a manageable computational structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The model transitions from static steady-state assumptions to dynamic transient analysis by incorporating time-dependent energy balance equations that capture the temporal evolution of temperatures during cooling failures and restart events. This enables accurate prediction of temperature changes as the system evolves through transient states.

Inventive Principle:
Principle #15Dynamics

2Reliability

If detailed energy balance equations are incorporated, then temperature prediction accuracy improves, but computational complexity increases

Engineering Contradiction:
Improveprediction reliability during power outagesVSAvoidsystem model complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The model pre-calculates and stores thermal mass parameters, heat transfer coefficients, and energy balance relationships for each component before transient events occur. During actual power outages, the system rapidly evaluates these pre-established relationships to predict temperature evolution, ensuring reliable predictions without requiring complex real-time computations.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary computational layer that bridges detailed physical models and practical prediction needs. Energy balance equations serve as intermediaries between the complex thermal dynamics of individual components and the overall system-level temperature predictions, enabling reliable forecasts while managing computational complexity through structured intermediate calculations.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Ease of operation

If cooling system restart dynamics are modeled, then operational guidance accuracy improves, but model complexity increases

Engineering Contradiction:
Improveoperational optimization capabilityVSAvoidrestart process modeling complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The model continuously tracks the thermal state of all components throughout the cooling failure and restart sequence, maintaining unbroken energy balance calculations from the moment cooling fails through the restart process. This continuous modeling provides uninterrupted operational guidance, showing how temperatures evolve and how restart timing affects overall system recovery.

Inventive Principle:
Principle #20Continuity of useful action

Solution Approach 2:

The model incorporates feedback mechanisms where predicted temperature changes from restart actions are fed back into the energy balance equations to refine subsequent predictions. This allows the system to evaluate the impact of different restart strategies and provide optimized operational guidance based on how actual system responses compare to predictions.

Inventive Principle:
Principle #23Feedback

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 enables more accurate predictions of temperature changes and cooling runtime, allowing data center operators to optimize design and operation, mitigate temperature fluctuations, and improve cooling efficiency, thereby ensuring the integrity of IT equipment during transient events.

Implementation Method 1

incorporating energy balance and heat transfer equations for various components, including chillers, water storage tanks, cooling coils, and IT equipment, to predict temperature changes during transient events

Methodology Applied
Scientific EffectHeat transfer: Conduction (thermal)

Implementation Method 2

A flow network model is developed to estimate air and cooling fluid temperatures in data centers, incorporating energy balance and heat transfer equations for various components

Methodology Applied
Scientific EffectEnergy balance: Thermal Energy Storage

Data Source

PatentEP3500078B1Method and system for predicting effect of a transient event on a data center
Publication Date: 2023.07.26 SCHNEIDER ELECTRIC IT CORP
  • EP3500078B1 patent drawingFigure 1
  • EP3500078B1 patent drawingFigure 2
  • EP3500078B1 patent drawingFigure 3

AI summary

A system and method is provided for predicting the effect of a transient event on a data center. According to one aspect, embodiments herein provide a method that comprises receiving input data related to a data center that includes at least one equipment rack and at least one cooling provider, the input data including data center architecture information, building data, and operating data, generating, by a computing device, a model based at least in part on the input data and on a set of energy balance and heat exchange equations for the data center that account for heat added by the at least one equipment rack and removed by the at least one cooling provider, and a thermal mass of the at least one equipment rack and a thermal mass of the at least one cooling provider, the model configured to predict at least one temperature in the data center during a transient event, and controlling a display device to display the at least one predicted temperature.