Composite concrete Filled Tubular Steel (CFT) members, which have excellent deformability due to the well-known confined and constrained interaction between steel tube and concrete, have largely been utilized as bridge piers or columns in high-rise buildings, resulting in increased strength and decreased column size. This study examined the experimental performance of steel tube columns filled with reactive powder concrete (RPC) under axial compression. Three sets of columns were used in the experiment, each with variations in shape (square, rectangular, and circular), length-to-diameter ratio, and compressive strength of the RPC. The first set consisted of five columns, while the second and third sets each had seven columns with three different lengths (750 mm, 600 mm, and 450 mm) and two different compressive strengths (54 and 92 MPa). A new numerical model was developed to calculate the ultimate failure load of the columns by considering factors such as the yield strength of steel, the compressive strength of concrete, the column shape, and the ratio of concrete to steel. This model was validated by comparing the results obtained from the experiments to those predicted by the model, as well as by designing equations from various codes. The results showed that the proposed numerical model accurately predicted the ultimate failure load for columns filled with different types of concrete, especially for RPC, while maintaining conservatism compared to the ACI, AISC, and EN codes equations. Doi: 10.28991/CEJ-2023-09-06-04 Full Text: PDF

This study explores the potential of a composite binder comprising cement bypass dust (CBD) and spent fluid catalytic cracking (FCC) catalyst for sustainable pavement base stabilization. Various CBD/FCC ratios (30:70, 50:50, 70:30) and binder contents (4%, 6%, 8%, 10%) were evaluated through laboratory testing. The 50:50 CBD/FCC mixture demonstrated optimal performance, achieving an unconfined compressive strength (UCS) of 15.6 MPa at 28 days with 10% binder content. The mix exhibited improved stiffness (E50 modulus up to 13,922 MPa) and resistance to degradation under wetting–drying cycles, attributable to synergistic cementitious and pozzolanic reactions. Microstructural analysis revealed a denser matrix, validating the enhanced performance. These findings suggest CBD and FCC, as promising materials for sustainable pavement construction, align with circular economy principles.

The structural stability of pipe pile foundations under seismic loading stands as a critical concern, demanding an accurate assessment of the maximum settlement. Traditionally, this task has been addressed through complex numerical modeling, accounting for the complicated interaction between soil and pile structures. Although significant progress has been made in machine learning, there remains a critical demand for data-driven models that can predict these parameters without depending on numerical simulations. This study aims to bridge the disparity between conventional analytical approaches and modern data-driven methodologies, with the objective of improving the precision and efficiency of settlement predictions. The results carry substantial implications for the marine engineering field, providing valuable perspectives to optimize the design and performance of pipe pile foundations in marine environments. This approach notably reduces the dependence on numerical simulations, enhancing the efficiency and accuracy of the prediction process. Thus, this study integrates Random Forest (RF) models to estimate the maximum pile settlement under seismic loading conditions, significantly supporting the reliability of the previously proposed methodology. The models presented in this research are established using seven key input variables, including the corrected SPT test blow count (N₁)₆₀, pile length (L), soil Young’s modulus (E), soil relative density (Dr), friction angle (ϕ), soil unit weight (γ), and peak ground acceleration (PGA). The findings of this study confirm the high precision and generalizability of the developed data-driven RF approach for seismic settlement prediction compared to traditional simulation methods, establishing it as an efficient and viable alternative.

The seismic response of tapered piles was investigated using a predictive framework based on radial basis function (RBF) interpolation. A database of 3,600 nonlinear three-dimensional simulations was generated to represent soil–pile interaction under seismic loading. The simulations incorporated nonlinear soil behavior, realistic tapered pile geometry, interface interaction, and dynamic ground motion input. Key parameters, including pile slenderness ratio, taper angle, soil unit weight, and peak ground acceleration, were systematically varied to cover a broad range of design scenarios. Maximum shaft friction, maximum bending moment, and lateral pile-head displacement were extracted and used to develop three predictive models. The models demonstrated high accuracy, with coefficients of determination of 0.990 for shaft friction, 0.997 for bending moment, and 0.991 for displacement, and root-mean-square errors of 0.766 MPa, 0.821 MN·m, and 0.156 mm, respectively. Gaussian kernels provided the best performance, and 95% prediction intervals showed strong calibration. Feature-importance analysis revealed that soil unit weight governs shaft friction, slenderness ratio controls bending moment, and lateral displacement results from coupled soil–geometry effects. The proposed framework closely reproduces nonlinear simulation results while significantly reducing computational cost, enabling rapid seismic evaluation and preliminary design of tapered pile foundations.

Machine learning models for the shear contribution of externally bonded fibre-reinforced polymer (FRP) to reinforced concrete beams are validated almost exclusively on a random hold-out split of a single self-compiled database. Because such databases are assembled from a few dozen experimental programmes, each contributing several near-identical specimens, a random split leaves siblings of nearly every test specimen in the training set. This study provides a cross-database assessment of that practice. Two independently compiled databases (442 and 455 specimens) were harmonised into a provenance-traced dataset of 828 records from 167 programmes spanning 1992–2024, and six learners were evaluated under five regimes of increasing severity. For gradient boosting the coefficient of determination fell from 0.67 and 0.69 under random splitting to 0.39 and 0.22 under study-grouped validation, and to 0.21 under cross-database transfer. A control in which the same physical specimens were scored under each compilation in turn changed little (0.35 versus 0.37), showing that clustering by programme, not curation disagreement, is responsible. A hierarchical errors-invariables model attributes 38% of the residual variance to programme identity and delivers calibrated predictive intervals (91.8% coverage at a nominal 90%), which span a factor of about twelve. Design provisions remain less accurate than the fitted models, but the reported margin roughly halves under honest validation. A reporting protocol is proposed, requiring publication of the specimen-to-study mapping, validation grouped on it, disclosure of out-of-range evaluation cases, and reporting of interval coverage and a characteristic value.