Modified Normalization Approach for Scaling of 1 g Model Tests of Laterally Loaded Rigid Piles in Sand

Abstract Determining the full-scale response of piles from the response of a reduced-scale model is challenging. This is mainly attributed to the inherent complexity in the nonlinear soil–pile interaction. One of the challenges is maintaining equivalent relative soil–pile stiffness between the model and the prototype. This paper discusses existing scaling and normalization methods and presents experimental evaluation of the effect of inequivalent relative soil–pile stiffness between the model and the prototype on the accuracy of predicting the normalized lateral-load rigid pile response. In this study, small-scale (model) and large-scale (prototype) testing setups were used to physically test and compare the response of two reduced-scale pile models with different pile stiffnesses (i.e., diameter to wall thickness ratios), and consequently, different relative soil–pile stiffness. The lab test models were prepared under similar conditions and tested under monotonic and cyclic lateral loading. The measured responses are presented in nondimensional form (i.e., normalized responses) for comparison. Using a commonly used normalization approach for laterally loaded rigid piles, the results showed that the root-mean square error (RMSE) exceeded 25% when comparing the normalized lateral responses of the two tested model piles. This emphasizes that having inequivalent relative soil–pile stiffness significantly impacts the consistency of the normalized responses. To correct for such inconsistency, this paper develops a modified normalization approach for scaling 1 g model tests (where g is gravitational acceleration) of laterally loaded rigid piles in sand to accurately predict the normalized lateral pile responses within the service-level range of behavior (i.e., at small strains). The proposed modification is developed and validated experimentally in the laboratory using static load tests. The applicability of this approach is also extended to cyclic loading. Moreover, the validity of the developed approach is further demonstrated using field measurements of piles reported in the literature and numerical simulations of full-scale prototypes. The validation results showed a significant enhancement when comparing the normalized responses of these piles and facilitate accurate predictions of the full-scale responses.