Simulation-Based Feasibility Assessment of a Hydrostatic Pressure-Difference Sensor for Bridge Deflection Monitoring

A hydrostatic pressure-difference (HPD) sensing chain for bridge dynamic-deflection measurement is assessed through finite-element (FE) simulation and re-analysis of two bridge-monitoring benchmarks. A virtual sensor model incorporating white plus 1/f noise, thermal drift and pipe Helmholtz dynamics is applied to a 30 m bridge subjected to moving truck and ambient excitation; an accelerometer double-integration baseline and a simulated GNSS receiver provide the references. The HPD chain attains a root-mean-square error of 2.0 mm and a Pearson correlation of 0.96. Peak picking recovers the first two modal frequencies accurately and reveals a spectral feature near the third reference mode, although quantitative interpretation of the latter is limited by its proximity to the processing-band boundary and by hydraulic-dynamic sensitivity. Independently, stochastic subspace identification recovers the first two modes with a sub-−1.5% bias, while the third exhibits a larger bias of approximately 10%. Fifteen months of hourly modal tracking from the KW51 railway-bridge dataset and raw Z24 progressive-damage records are re-analyzed; the underlying structural-change-detection methodology—damage detection at Z24, where the applied disturbances are documented damage scenarios, and structural-change (retrofit) detection at KW51, where the documented event is a strengthening intervention rather than damage—and not the HPD sensor is validated against these real datasets, since no field HPD deployment exists on either structure. A temperature-corrected Mahalanobis-distance detector, with training windows from 60 to 180 days, triggers on 100% of the post-retrofit days, with false-alarm rates ranging from 2.3% to 7.9% with the 90-day/6.6% operating point retained as the reference; four Z24 settlement scenarios (20–95 mm) show frequency shifts of the same real order of magnitude.