The drawbacks of conventional measurement techniques motivate devising a non-destructive and automated optical method that can provide in situ heat flux quantification during boiling.Ĭurrent advances in deep learning and, in particular, convolutional neural networks (CNNs) have enabled automatic and scalable image analysis for, e.g., object detection 28, 29, 30, 31, classification 32, 33, 34, 35, 36, 37, and even image-based predictions 38, 39, 40, 41, 42, 43, 44. Despite the significance of gathering essential visual information, current measurement setups fail to synchronically analyze image data without extensive user involvement, which is not only time-consuming, but also introduces user bias. In other words, a bridge between measurements and visual information must be built to relate surface design inputs (e.g., surface morphology, material type, and liquid–vapor and liquid–solid interfaces) with boiling curves. However, these experimental methods are inefficiently connected with visual information, which is a huge downfall for providing a clear description of dynamic boiling physics. Therefore, researchers still heavily rely on experiments to measure the boiling heat flux via, e.g., thermocouples 24, electrical power input 25, 26, or infrared (IR) techniques 27. Although direct numerical simulation of the boiling process enables studying dissipated heat flux at local and global scales, the accuracy of these simulations is debatable 22. With numerical simulations, single to multi-bubble physics are investigated for detailed characterization of heat flux 21, 22, 23. However, the intrinsic complexity of the dynamic boiling phenomena has limited those theoretical studies to very simplified models 19, 20. Theoretical research on boiling mechanisms provided the foundations for heat flux estimation 15, 16, 17, 18. Quantification of boiling curves has been extensively studied in many theoretical, numerical, or experimental works.

These past findings suggest that inherent structural characteristics as well as intrinsic material properties significantly affect boiling performance, and therefore the boiling curve. With the goal of increasing the CHF limit and HTC, extensive prior works have investigated the effects of flow condition 6, 7, surface treatment and design 5, 8, 9, 10, 11, 12, 13, and bubble morphology on boiling curves 14. For instance, the efficacy of boiling heat transfer can be quantified by either the improvements in the critical heat flux (CHF) or heat transfer coefficient (HTC) 5, both of which are functions of the boiling heat curves. The energy per unit area (i.e., heat flux) measures how much thermal energy is relieved via boiling and is a critical factor in characterizing boiling heat transfer. Since boiling enables a system to maintain fairly constant surface temperatures during large thermal energy fluctuations, many modern high power density systems such as power plants, power electronics, laser diodes, and photovoltaics rely on boiling for thermal management 2, 3, 4. Our framework offers an automated, learning-based, alternative to conventional boiling heat transfer metrology.īoiling is a heat transfer mechanism that utilizes liquid-to-vapor phase transition to dissipate great amounts of heat with minimal temperature difference 1. By training on these features, our model learns physical boiling laws that statistically describe the manner in which bubbles nucleate, coalesce, and depart under boiling conditions, enabling in situ boiling curve prediction with a mean error of 6%.

The framework leverages cutting-edge deep learning models including convolutional neural networks and object detection algorithms to automatically extract both hierarchical and physics-based features. Here, we introduce a data-driven learning framework that correlates high-quality imaging on dynamic bubbles with associated boiling curves. Connecting boiling physics with bubble dynamics is an important, yet daunting challenge because of the intrinsically complex and high dimensional of bubble dynamics. Central to the boiling process is the development of bubbles. Boiling is arguably Nature’s most effective thermal management mechanism that cools submersed matter through bubble-induced advective transport.

Linking bubbles in mindly