Analysis of Empirical Models for Predicting the Rupture Force in Four-Pile Caps.

Technical literature provides experimental campaign results for four-pile caps with significant value in structural engineering. However, failure modes and measurement of these specimens are particularly challenging due to their volumetric nature. These experimental records are frequently utilized to evaluate analytical models focusing on less conservative than normative models to predict rupture force. Moreover, recent studies emphasize the necessity of developing simplified predictive models. In this context, the objective of this study was to evaluate the feasibility of regression models for estimating the rupture force associated with strut failure. The evaluation was based on a commonly used database and employed analysis of variance (ANOVA) at a 5% significance level to identify critical variables. The regression models were developed with variable interactions incorporated into the equations in three forms: (i) without interaction, (ii) with linear interaction, and (iii) with quadratic interaction. An analysis of the developed regression models identified a model with satisfactory accuracy. This model achieved an average predicted force ratio of 1.00 (COV = 14%) for the database and 1.03 (COV = 16.68%) for extrapolated numeric models in finite elements with the concrete damaged and plasticity (CDP) constitutive model, initially calibrated with experimental tests. A methodology was proposed to assist in the initial design of four-pile caps. [ABSTRACT FROM AUTHOR]