Treatment with lime can increase marl strength but may also cause abrupt post-peak failure. Polypropylene (PP) fibers can bridge cracks and improve deformability, yet peak strength and total energy absorption alone do not show how energy is mobilized as failure develops. This study examines the post-peak response of lime-stabilized marl reinforced with PP fibers through a strain-resolved energy analysis. The experimental program covered 2–8
Reliable prediction of local scour around cylindrical bridge piers in cohesive beds remains difficult because hydraulic forcing and soil resistance interact nonlinearly. This study compares five data-driven models—Artificial Neural Network (ANN), Support Vector Machine (SVM), K-Nearest Neighbours (KNN), Gene Expression Programming (GEP) and QNET—for the dimensionless maximum scour depth dmax/D. The development set comprises 122 laboratory records taken from the cohesive-bed compilation of Choi and Choi (2022). Four predictors are used: water content Wc, clay content CP, pier Froude number Fr=V/√(gD) , and normalised critical shear strength τ _c/(ρ V^2) . All 15 non-empty subsets of these predictors are evaluated. Unlike Choi and Choi (2022), the comparison is multi-algorithm and does not introduce a new scour-scaling law. On the reported 70:30 holdout split, QNET achieved the highest test scores (R2=0.9958, RMSE = 0.0111, DC = 0.9836), followed by ANN (R2 = 0.9206, RMSE = 0.0303, DC = 0.9285). SVM and KNN were intermediate, and full-input GEP was weaker (R2 = 0.6791) despite an explicit formula. These ranks describe this partition only. The near-unity QNET result is not taken as proof of split-independent or field-scale generalisation. One-at-a-time deletion shows that predictor influence is model- and metric-dependent and does not support a universal single-variable ranking. Fr is repeatedly influential, but Wc, CP and τ _c/(ρ V^2) remain necessary. The models are screening tools within the laboratory ranges of the compiled matrix, complementary to erodibility-based design methods, not stand-alone design equations. Independent, preferably field-scale, testing is still required.
The present research has developed a computational approach that combines data-driven surrogate modeling and macro-scale analysis to provide a workflow for the analysis and trade-off exploration of micropile performance in liquefiable soils under seismic loading. A synthetic data set containing 2000 data points has been created using parametric relationships that define cyclic soil behavior. A multi-output neural network has been developed and trained to predict the excess pore pressure ratio, normalized shear modulus, and damping ratio, which are critical performance measures of liquefaction-sensitive soil response. These performance measures have been used in a macro-scale analysis to represent changing soil conditions and their influence on micropile performance. A multi-objective optimization process has been conducted to examine trade-offs between cost, displacement, and safety factor. The research has shown that stable performance of the surrogate model has been obtained for stiffness and damping approximations, but greater scatter has been observed in estimating the excess pore pressure ratio, particularly at intermediate response conditions. However, a new approach has been proposed that combines data-driven surrogate modeling and macro-scale analysis to demonstrate a computational workflow for the analysis and trade-off exploration of micropile performance in liquefiable soils under seismic loading. Although this research has not validated experimental and field data, it has shown that a new approach can be developed by combining data-driven surrogate modeling with macro-scale analysis to evaluate micropile performance in liquefiable soils subjected to seismic loading. This is a proof-of-concept study developed entirely on synthetic data: all quantitative results (approximation accuracies, design tables, and cost-displacement-safety trade-offs) are illustrative outputs of the computational workflow and are explicitly not intended for direct engineering use.
The construction industry generates 4 billion tonnes of Construction and Demolition Waste (CDW) worldwide. At the same time, infrastructure projects face a scarcity of natural resources. Managing CDW and balancing supply and demand are the two biggest challenges facing the infrastructure industry today and in the future. Recycling CDW and other industrial waste to produce low-carbon building blocks using emerging geopolymer technology can address these challenges. This study uniquely integrates precursor reactivity assessment, fresh and hardened property tests for the blocks, and links block performance to microstructural analysis, structural applications in masonry, and a comparative Life Cycle Assessment (LCA) within a single investigation. Two varieties of CDW, namely cement-based waste and brick powder, are used as recycled aggregates, along with cement, GGBS, and alkali activators as binders. Blocks with compressive strength (CS) ranging from 3.25 MPa to 7.06 MPa and flexural strength (FS) ranging from 1.9 MPa to 2.3 MPa were obtained. Their prism strength shows the blocks' enhanced load-bearing capability. The improved strength is attributed to the uniform distribution of particles in the geopolymer-OPC hybrid matrix and the development of polymeric reaction products, as observed by SEM–EDS and reflected in the XRD patterns. The LCA of the Geopolymer-OPC hybrid system Stabilized Recycled Aggregate Blocks (GSRAB) showed a 61
Project management faces multiple challenges due to the increasing complexity and demand for project-based work across various sectors. To address this issue, artificial intelligence (AI) is experiencing rapid growth and is expected to impact project management across various domains significantly. However, the lack of a comprehensive review hinders the clarification of future directions and implementation priorities. This paper aims to fill this gap by systematically analyzing related studies using quantitative and qualitative methods. Following a systematic screening process of records retrieved from Scopus and Web of Science, a final dataset of 1184 articles was included for bibliometric and qualitative analyses. The bibliometric analysis examined research trends, publication patterns, keyword co-occurrence, and keyword burst detection. Qualitative analysis indicated that the literature can be categorized into three main sectors: general project management, construction project management, and software project management. The analysis also identifies the main research areas within each sector, enabling the discussion of current research gaps and potential directions for future studies.
