Data driven models for accurate prediction of nucleate boiling on oxidised surfaces.

沸腾流动的数值预测是计算流体力学（CFD）的一个具有挑战性的领域。沸腾流在广泛的行业中普遍存在，从核热工液压到咖啡机设计。然而，沸腾的CFD预测往往严重依赖于经验相关性。这些相关性解释了在比计算网格更小的尺度上发生的物理效应。在“核沸腾”中尤其如此——加热表面几何形状的微尺度缺陷“播种”相变导致气泡的形成。更复杂的是，已知在气泡和受热表面之间形成了一层微尺度层。准确预测通过该微层的传热（和相变）对气泡动力学的预测至关重要，因此对工业具有重要的宏观传热率。

Numerical prediction of boiling flow is a challenging area of Computational Fluid Dynamics (CFD). Boiling flows are prevalent in a broad range of industries, ranging from e.g., nuclear thermal hydraulics to coffee machine design. However, CFD prediction of boiling is often heavily reliant on empirical correlations. These correlations account for physical effects occurring on scales smaller than the computational mesh. This is especially true of "nucleate boiling" - where micro-scale imperfections in the heated surface geometry "seed" phase change leading to the formation of bubbles. To further complicate matters, a micro-scale layer is known to form between the bubble and the heated surface. The accurate prediction of heat transfer through (and phase change of) this micro-layer is crucial to the prediction of bubble dynamics, and hence macro-scale heat transfer rates of importance to industry.