Evaluation of a digital twin for structural damage localisation and sizing under model discrepancy

Abstract

This dataset contains the inputs and outputs from a reproducible simulation testbed developed to evaluate a digital twin for locating and sizing structural damage in an eight-storey structure under model-form discrepancy. It includes independent discrepancy-truth model outputs, ensemble Kalman inversion results, deterministic and Markov chain Monte Carlo baseline comparisons, identifiability analyses, uncertainty coverage and calibration measures, healthy-control and per-asset false-alarm results, damage recovery and robustness outputs, and cost-based inspection-decision results. All data are computationally generated and do not represent observations from actual structures, organisations, or human participants.