Bird-Carreau Leidenfrost DeepONet: high-fidelity CFD dataset and trained operator-learning surrogates

Data, trained models and analysis code accompanying the manuscript "Vapour-film dynamics of Bird–Carreau Leidenfrost drops and operator-learning trajectory prediction" (submitted to International Journal of Heat and Mass Transfer). Contents: Dataset (master_dataset_with_Cr.csv): 397 high-fidelity Volume-of-Fluid + Hardt–Wondra phase-change simulations of shear-thinning (Bird–Carreau) Leidenfrost droplets, 4691 time-snapshot rows, giving the temporal trajectories of the vapour-neck radius (rn), minimum vapour-film thickness (hc) and vaporisation rate (qm) as functions of the dimensionless groups We, Oh, Ja and Cr. Per-case peak targets (peak_values_with_Cr.csv) for the peak-value predictor. Trained models: the DeepONet temporal predictor (Model B, branch {We,Oh,Ja,Cr}, trunk normalised time t*, 593,923 parameters, R^2 = 0.980/0.965/0.867 for qm/hc/rn on held-out cases), the MLP peak-value predictor (Model A), an auxiliary rn-only DeepONet, and all input/output scalers and training histories. Code: a portable load-and-predict example and the training/analysis notebooks (correlation analysis, DeepONet training, resolution-independence demonstration). Not included: the raw OpenFOAM solver decks for the 397 simulations (available from the corresponding author on request) and the manuscript source (distributed through the journal).