Accuracy assessment of predictive models for semirigid extended end‐plate connections

Taking advantage of semi‐rigid connections' inherent stiffness and strength can highly benefit the steel industry, since this can lead to efficient designs and consequently to lower construction costs and carbon emissions. This requires the ability to accurately predict the connection's moment‐rotation response. National standards and research studies proposed a number of predictive models to do so, including analytical, mechanical, and empirical models. This applies to the popular bolted extended end‐plate connections. A number of studies have indicated that such models have limitations and may provide inaccurate predictions. In this study, a recently compiled database, comprising more than 750 test specimens, is used to assess the accuracy of several models in predicting key response quantities related to stiffness, strength, and ductility. The mean and standard deviation of the prediction error is quantified while highlighting the advantages and drawbacks of each model. The results aim to shed light on the limitations of existing predictive models and offer recommendations for improved future models.