Fire resistance of concrete columns: Meta-analysis of code-based methods

A dataset of 165 fire tests on rectangular concrete columns, spanning five decades, was compiled to evaluate six design methods from five international codes (Eurocode Methods A and B (ECA, ECB), Australian (AS 3600), Chinese (DBJ/T), American (ACI), and Indian (NBC)). A meta-analysis assessed their predictive reliability using historical and recent results. ECA and AS 3600 gave the most consistent and conservative predictions, with ECA performing best overall. For fire resistance ratings (FRR) < 240 min, about 70 % of ECA (2019) predictions were within ±20 % of test outcomes, though accuracy declined at longer durations. AS 3600 produced similar results due to its related formulation. ECB changed notably: 2019 method was unreliable for FRR >140 min, whereas the 2023 version aligned better with recent tests and reduced excessive conservatism. DBJ/T performed adequately on older specimens but overestimated newer ones, while ACI and NBC showed high variability and frequent unconservative predictions. Assessment of the robustness of the ECA (2023) Method with respect to design parameters indicated best performance for 250–300 mm columns, reinforcement ratios of 2.5–3.5 %, effective lengths of 3–5 m, covers of 35–65 mm, and concretes of 20–60 MPa, with reduced accuracy at extreme conditions.