Enhancing Heat Sink Efficiency in MOSFETs using Physics Informed Neural Networks: A Systematic Study on Coolant Velocity Estimation
#Physics Informed Neural Networks #MOSFET cooling #Heat sink efficiency #Coolant velocity estimation #Power Electronic Building Blocks #Thermal management #Computational physics
📌 Key Takeaways
- Researchers developed PINNs methodology to determine optimal coolant velocity for MOSFETs
- MOSFETs are critical components in Power Electronic Building Blocks experiencing significant thermal loads
- The sequential training method decouples optimization of each layer, reducing complexity
- The methodology's predictions show good agreement with experimental results
📖 Full Retelling
Researchers Aniruddha Bora, Isabel K. Alvarez, Julie Chalfant, and Chryssostomos Chryssostomidis have developed a novel methodology using Physics Informed Neural Networks (PINNs) to determine optimal coolant velocity for multilayered metal-oxide-semiconductor field-effect transistors (MOSFETs), addressing a critical challenge in electronic thermal management published on arXiv on February 13, 2026. The research focuses on MOSFETs, which are integral components of Power Electronic Building Blocks (PEBBs) that bear the majority of thermal loads, making effective cooling essential to prevent overheating and potential burnout. The team specifically addresses the challenge of determining the required coolant velocity when given inlet and outlet temperatures and a specific heat flux in a multilayered MOSFET structure composed of aluminum, pyrolytic graphite sheets, and stainless steel pipes containing flowing water. The researchers propose an algorithm that employs sequential training of the MOSFET layers in PINNs, which mathematically decouples the optimization of each layer by treating parameters of other layers as constants during training, reducing the dimensionality of the optimization landscape and making it easier to find global minima while avoiding poor local ones. The team theoretically analyzed the convergence of their PINNs solution to analytical solutions and demonstrated that their methodology's predictions align well with experimental results, representing a significant advancement in computational engineering and physics for solving complex thermal management problems in power electronics.
🏷️ Themes
Thermal Management, Neural Computing, Power Electronics
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Neural and Evolutionary Computing arXiv:2602.20177 [Submitted on 13 Feb 2026] Title: Enhancing Heat Sink Efficiency in MOSFETs using Physics Informed Neural Networks: A Systematic Study on Coolant Velocity Estimation Authors: Aniruddha Bora , Isabel K. Alvarez , Julie Chalfant , Chryssostomos Chryssostomidis View a PDF of the paper titled Enhancing Heat Sink Efficiency in MOSFETs using Physics Informed Neural Networks: A Systematic Study on Coolant Velocity Estimation, by Aniruddha Bora and 3 other authors View PDF HTML Abstract: In this work, we present a methodology using Physics Informed Neural Networks to determine the required velocity of a coolant, given inlet and outlet temperatures for a given heat flux in a multilayered metal-oxide-semiconductor field-effect transistor . MOSFETs are integral components of Power Electronic Building Blocks and experiences the majority of the thermal load. Effective cooling of MOSFETs is therefore essential to prevent overheating and potential burnout. Determining the required velocity for the purpose of effective cooling is of importance but is an ill-posed inverse problem and difficult to solve using traditional methods. MOSFET consists of multiple layers with different thermal conductivities, including aluminum, pyrolytic graphite sheets , and stainless steel pipes containing flowing water. We propose an algorithm that employs sequential training of the MOSFET layers in PINNs. Mathematically, the sequential training method decouples the optimization of each layer by treating the parameters of other layers as constants during its training phase. This reduces the dimensionality of the optimization landscape, making it easier to find the global minimum for each layer's parameters and avoid poor local minima. Convergence of the PINNs solution to the analytical solution is theoretically analyzed. Finally we show the prediction of our proposed methodology to be in good agreement with experimental results. S...
Read full article at source