Table 5 Sensitivity analysis results for the influence of variation of the input features on inhibitor concentration
From: Surrogate modelling of corrosion inhibition finite element simulations using machine learning
Features | Baseline value | Pertubation | Inh. Conc. (mol/m3) | Inh. Conc. (mol/m3) | Inh. Conc. (mol/m3) | Normalized sensitivity | Normalized sensitivity |
|---|---|---|---|---|---|---|---|
−p% | Baseline | +p% | −p% | +p% | |||
Initial PVC (%) | 13.5 | ±9% | 3.79 | 4.90 | 6.34 | 2.04 | 2.64 |
Defect width, W2 (μm) | 500 | ±10% | 4.54 | 4.40 | 4.25 | 0.32 | 0.32 |
Primer thickness, H2 (μm) | 24 | ±8% | 9.46 | 10.38 | 11.31 | 1.07 | 1.07 |
Defect depth, H3 (μm) | 50 | ±8% | 7.24 | 7.31 | 7.37 | 0.11 | 0.11 |
Water layer thickness, δ (μm) | 1000 | ±10% | 3.49 | 3.22 | 3.83 | 0.85 | 1.90 |
