Adaptive Measurement Scheduling for Deflection-Based Corrosion Monitoring of Reinforced-Concrete Structures

Non-invasive estimation of reinforcement corrosion from the deformations of reinforced-concrete (RC) structures requires measurement intervals matched to the evolution of the corrosion process. Fixed intervals chosen from a discrete candidate set give no accuracy control between visits. We therefore formulate measurement scheduling as error-controlled adaptive step selection. At each measurement, an a priori estimate of the local prediction error, obtained from the disagreement between linear and quadratic extrapolations of the inferred corrosion-current density, limits the next interval, and a proportional–integral controller adjusts it from the realised prediction error under a user-defined tolerance. All schedules sample the same monthly numerical reference trajectory. For a reference RC beam, the adaptive schedules dominate the 6- and 12-month fixed-interval baselines, reduce the peak scheduling error 2.4- to 3.2-fold for tolerances of 0.5% to 2% relative to equally spaced schedules with equal numbers of measurements, and reproduce this numerical reference within 0.21% of the baseline current density using fewer than one fifth of the monthly measurements. The benefit arises from non-uniform deterioration and vanishes for uniform deterioration; measurement noise requires a tolerance of at least four times the noise level. The scheduler is verified numerically; experimental validation under realistic monitoring conditions is still required.