Reduced-Order Battery Thermal Modeling for Immersion Cooling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for computing temperature parameters in conjugate heat transfer systems, such as those in battery management systems, are computationally expensive and lack real-time accuracy, especially in complex flow patterns like immersion cooling applications, and machine learning models are sensitive to boundary condition changes.
Innovation Solution
A method combining physics-based and data-driven techniques using Proper Orthogonal Decomposition (POD) to compute reduced-order coefficients for fluid and battery temperatures, solving reduced basis equations to achieve faster and accurate temperature parameter estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional high-fidelity models like computational fluid dynamics (CFD) are used to model battery parameters, then accuracy of temperature parameter computation is improved, but computational cost and memory requirements increase significantly
Solution Approach 1:
The patent segments the computational problem by separating the complex CFD simulation into a training phase (offline) and a prediction phase (online). During training, high-fidelity CFD data is used to train machine learning models. During operation, these trained models provide fast predictions without requiring repeated CFD computations, thus reducing real-time computational cost while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-training machine learning models using extensive CFD simulation data before actual battery operation. This offline training phase captures the complex thermal-fluid dynamics relationships, enabling the model to provide accurate real-time predictions during battery operation without requiring computationally expensive CFD simulations at runtime.
2Productivity
If machine learning models are used for predicting battery parameters, then computational speed is improved, but performance degrades when extrapolating beyond training conditions or when boundary conditions change
Solution Approach 1:
The patent addresses extrapolation limitations by training the machine learning model on CFD simulation data that covers a wide range of operating conditions, including variations in boundary conditions, cooling flow rates, and battery thermal states. This extensive training enables the model to generalize better to unseen operating conditions while maintaining computational speed.
3Ease of manufacture
If conventional heat transfer modeling with empirical correlations is used, then computational simplicity is improved, but accuracy deteriorates in complex flow patterns like immersion cooling
Solution Approach 1:
The patent introduces machine learning models as an intermediary between simple empirical correlations and complex CFD simulations. These models are trained on high-fidelity CFD data that accurately captures complex immersion cooling flow patterns, then deployed to provide CFD-level accuracy with much lower computational cost, effectively bridging the gap between simplicity and precision.
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
Provides accurate and fast temperature parameter computation, achieving computational speed-ups of 8.6-13.8 times compared to full-order models, with improved performance in both interpolation and extrapolation scenarios.
Implementation Method 1
a first plurality of fluid steady state reduced basis equations are solved to compute a first plurality of fluid velocity reduced coefficients and a first plurality of fluid pressure reduced coefficients based on the plurality of input parameters and a plurality of reduced order matrices. a fluid velocity field and a fluid pressure field are computed based on the first plurality of fluid velocity reduced coefficients, the first plurality of fluid pressure reduced coefficients and the first reduced subspace
Implementation Method 2
Conjugate heat transfer is a type of heat transfer analysis between solids and fluid(s). Heat transfer in batteries of electric vehicles is an example of conjugate heat transfer, where heat dissipation in solid battery is transferred to fluids surrounding the solid battery/battery
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
To maintain a healthy and safe battery, the Battery Management System (BMS) require numerous sensors which is a costly process. Immersion cooling applications where the battery stack is cooled with a dielectric coolant fluid has complex flow patterns that are not modeled easily without high fidelity modeling. The present disclosure provides a solution by combining physics-based and data-driven techniques. Initially, a first reduced subspace and a second first reduced subspace are extracted from the high-fidelity or full-order model (FOM) data with the help of proper orthogonal decomposition (POD) in offline. Further, the dynamics of the reduced spaces (first reduced subspace and the second reduced subspace) are captured by computing the reduced coefficients. The reduced coefficients are computed by solving the reduced order equations. The reduced model is solved in the same manner as the high-fidelity or FOM model. The reduced-order solution is constructed once the reduced coefficients are obtained.