Answers are grounded in this paper and cite their sources.
Estimating sill geometry and flow properties is essential for the preliminary design of stilling basins with compact sill-controlled hydraulic jumps. This study combines dimensional analysis, empirical regression, and physics-informed learning using 151 laboratory records from a horizontal rectangular channel, spanning inflow Froude numbers of 2.80–9.89 and compactness ratios of 0.893–2.217. Compactness, γ, is defined as the ratio of the calculated classical-jump roller length to the sill distance from the jump toe. Empirical relationships for relative sill height, S, relative sill position, X, and conjugate-depth ratio, Y, yield calibration R2 values of 0.980, 0.998, and 0.979, respectively. A cascaded physics-informed neural network with multi-head attention (C-PINN-MHA) provides complementary inference conditioned on γ. Under stratified five-fold cross-validation with globally normalized C7 loss scaling, R2 averaged 0.965 ± 0.005 for S and 0.982 ± 0.004 for Y, while the mean violation rate for the adopted one-sided specific-force criterion was 1.3 ± 1.6%. These results represent internal estimates of predictive performance under the stated normalization protocol. No violations of the same criterion occur on the fixed 37-record test set; however, satisfying this criterion alone does not establish the full momentum balance or hydraulic admissibility. Cumulative ablation indicates that the asymmetric feasibility penalty reduces violations, whereas subsequent architectural additions do not consistently improve accuracy. Because γ incorporates sill-position information, X is interpreted as a conditional reconstruction rather than an independent prediction. The empirical equations offer simple preliminary estimates, while the network makes hydraulic constraints explicit within its training objective.
No takes yet. Share an insight, caveat, or question.
Bouriche et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: