Thermal-response trend screening for inspection prioritisation of prestressed concrete continuous box-girder bridges

Routine bridge monitoring archives already contain temperature, strain, and deflection records, but year-scale changes in thermal response are mixed with creep drift, annual cycles, and seasonal variation. This study develops Semi-Diurnal Differential Analysis (SDDA) for inspection screening from such records. SDDA uses a 12-h difference to retain the day–night thermal excitation while attenuating slower components, estimates monthly effective thermal-response coefficients α̂, and applies Seasonal Mann–Kendall testing to the month-by-year matrix. Monte Carlo simulations check false-positive control, trend recovery, and detection power under field-scale noise. The method is applied to two prestressed concrete continuous box-girder bridges in northeastern China with 8–9 years of monitoring data. In the primary Bridge A case, the S1 side-span strain channel shows a decline in α̂ of 2.77%/yr (pperm≤0.0001, cumulative ∼18%), whereas the S3 side-span reference channel has no statistically detectable monotonic trend and the main-span channel is retained as a supplementary heterogeneous signal. Bridge B provides a shorter exploratory watch-list application. SDDA translates screening results into an inspection-priority ranking by sensor, section, and support region for field inspection, sensor cross-checking, and maintenance-record review, without assigning the structural cause.