Several factors, such as the payload capacity of a geotechnical centrifuge, performance of an in-flight shake table and a maximum achievable centrifugal acceleration, limit the size of a structure that can be modelled to study soil−foundation−structure interaction (SFSI) effects using centrifuge modelling. The two-stage scaling method is an innovative technique to model larger prototypes in currently available centrifuge facilities. In this study, two sets of dynamic centrifuge experiments were conducted to evaluate the applicability of the method for SFSI studies. Two physical models, representing a target structure placed on dry sandy soil, were designed based on the scaling factors of conventional centrifuge modelling at 50g and the two-stage scaling method at 25g. The models were excited under different realistic earthquake motions. The results showed good agreements for the estimation of flexible-base natural frequencies of the physical models. The performance of the method for estimation of seismic amplification factors, kinematic incoherence parameters and foundation impedance functions was fair. Poor and fair agreements were observed for the estimation of seismic soil settlement recorded beneath the physical models and free field, respectively. Overall, the method can be used for SFSI studies in the centrifuge modelling while considering its strengths and weaknesses.
