StrucIQ Beam Damage and Retrofit Dataset

Simulated response of simply supported reinforced concrete beams under staged damage, corrosion, sustained load and strengthening, built to develop and test methods that assess a beam's condition from measurements (deflection, natural frequency, crack width) and choose a retrofit. What it contains - A design-space population of 10,000 beams (span 2 to 6 m, IS 456:2000 designs, M20 to M40, Fe415 and Fe500, four-point and uniform loading, sampled material properties), each taken through five damage stages with a fibre beam model (OpenSeesPy). Per stage: residual and diagnostic deflections, the first three natural frequencies, static and vibration (breathing-crack) stiffness left, steel stress and strain, curvature, neutral axis, IS 456 Annex F crack widths; a final push gives the peak load and load-deflection curve.- A simulated measurement layer (frequency, deflection and crack-width readings with stated scatter and temperature effects) and two benchmark tasks with random and extrapolation splits.- Strengthening by mechanics for 1,000 beams: carbon fibre laminates, a bonded steel plate and a reinforced concrete jacket added to the damaged model (4,000 records).- A corrosion stage for 2,000 beams at 5, 10 and 15% steel mass loss (6,000 records), and a sustained-load stage for 1,000 beams over 50 years with creep and shrinkage (2,000 records).- A Eurocode 2 fatigue-life estimate of the tension steel for every reached stage.- A reference specimen Monte Carlo study (300 beams), a 10,000-beam retrofit decision population, blind reproduction of three published test programmes, and an independent STAAD.Pro check. Key results - A vibration test registers 48% of the static stiffness loss at the first damage stage, rising to 62% at the last (medians).- Corroded beams keep 91, 84 and 78% of their peak load at 5, 10 and 15% mass loss.- Under sustained service load, deflection grows 2.7 times over 50 years (median), 0.90 of the IS 456 Annex C estimate.- Benchmark task A1: reading vibration as stiffness errs by about 20 percentage points; a gradient-boosting baseline on the measurements errs by 2.9. Validation and limitations The fibre model reproduces the static and vibration stiffness and capacities of published tests (blind on Musiał et al. 2021; capacities of Ozturkoglu et al. 2024). The inputs of Pešić et al. 2015 were incomplete, so their measurements are used for comparison only. Elsewhere in the design space it is a simulation: flexure only (no shear failure or bond slip), corrosion without bond loss, creep and shrinkage only in the sustained-load stage, fatigue as a code estimate. A continuum cross-check on 40 beams agrees on frequency and gives larger cracked-state deflections; the README reports it in full. Failed runs and solver anomalies are kept and flagged. Every column is defined in `06_design_space_population/data_dictionary.csv`; `quickstart.ipynb` loads the main tables. Data under CC BY 4.0, code under MIT. Cite as: Pesaru, S. A. R., Patlola, S. R., & Murari, B. R. (2026). *StrucIQ Beam Damage and Retrofit Dataset* (Version 1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.23769960