Table 13 Comparison between experimental and ML-based durability assessment approach.
Aspect | Experimental assessment | ML-based assessment |
|---|---|---|
Time Requirement | Long duration (6–24 h for RCPT and curing-dependent) | Instant prediction once the model is trained |
Resource demand | Requires costly test setup, power supply, and skilled operators | Low resource usage after data preparation |
Data input | Experimental measurements from lab testing | Input parameters easily assessed |
Accuracy & variability | Susceptible to human and procedural errors | Consistent, data-driven, and less operator-dependent |
Generalizability | Limited to tested samples and curing conditions | Can generalize across wide material combinations (if trained on diverse datasets) |
Environmental impact | Involves energy-intensive curing and testing | Promotes sustainability by reducing experimental repetitions |
Practical application | Direct measurement of durability indices | Predictive tool for design optimization and decision support |
