Experimental Study on Heat Transfer Performance of FKS-TPMS Heat Sink Designs and Time Series Prediction
As the demand for advanced cooling solutions increases with the rise in artificial intelligence and high-performance computing, efficient thermal management becomes critical, particularly for data centers and electronic systems. Triply Periodic Minimal Surface (TPMS) heat sinks have shown superior t...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-07-01
|
| Series: | Energies |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1996-1073/18/13/3459 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | As the demand for advanced cooling solutions increases with the rise in artificial intelligence and high-performance computing, efficient thermal management becomes critical, particularly for data centers and electronic systems. Triply Periodic Minimal Surface (TPMS) heat sinks have shown superior thermal performance over conventional designs by enhancing heat transfer efficiency. In this study, a novel Fischer–Koch-S (FKS) TPMS heat sink was experimentally tested with four porosity configurations, 0.6 (identified as P6), 0.7 (identified as P7), 0.8 (identified as P8), and a gradient porosity ranging from 0.6 to 0.8 (identified as P678) along the flow direction, under a mass flow rate range of 0.012 to 0.019 kg/s. Key thermal parameters including surface temperature, thermal resistance, heat transfer coefficient, and Nusselt number were analyzed and compared to the conventional straight-channel heat sink (SCHS) using numerical modeling. Among all configurations, the P6 design demonstrated the best performance, with surface temperature differences ranging from 13.1 to 14.2 °C at 0.019 kg/s and a 54.46% higher heat transfer coefficient compared to the P8 design at the lowest mass flow rate. Thermal resistance decreased consistently with an increasing mass flow rate, with P6 achieving a 31.8% reduction compared to P8 at 0.019 kg/s. The P678 gradient design offered improved temperature uniformity and performance at higher mass flow rates. Nusselt number ratios confirmed that low-porosity and gradient TPMS designs outperform the SCHS, with performance advantages increasing as the mass flow rate rises. To further enhance the experimental process, a deep learning model based on a Temporal Convolutional Network (TCN) was developed to predict steady-state surface temperatures using early-stage time-series data, to reduce test time and enable efficient validation. |
|---|---|
| ISSN: | 1996-1073 |