Title:
Machine Learning Approach to Rapidly Evaluate Curling of Concrete Pavement
Author(s):
Sangyoung Han, Taemin Heo, Chul Min Yeum, Kukjoo Kim, Jongkwon Choi, Mang Tia
Publication:
IJCSM
Volume:
19
Issue:
1
Appears on pages(s):
Keywords:
Concrete pavement, Machine learning, Finite element, TETD, FWD
DOI:
10.1186/s40069-024-00703-x
Date:
1/31/2025
Abstract:
The current KDS 14 design method yields reasonable accuracy with acceptable safety in assessing the unbalanced moment-carrying capacity of slab–column connections. However, the model requires considerable computational effort owing to the effects of various design parameters, particularly the gravity-load effect. This study proposes a method to simplify the KDS 14 model to evaluate the unbalanced moment-carrying capacity of slab–column connections. In the proposed method, the gravity-load effect is decoupled from equations used for evaluating unbalanced moment-carrying capacity components. Subsequently, the total unbalanced moment-carrying capacity is determined by establishing an interaction between the gravity shear ratio and unbalanced moment components without considering the gravity-load effect. For practical design purposes, final simplified design equations are proposed. The reliability of the simplified method is validated based on a comparison with the current KDS 14 design code using a comprehensive database encompassing interior, exterior, and corner slab–column connections. Furthermore, a parametric study based on the proposed simplified approach, current design codes, combined with finite-element (FE) analysis is performed to elucidate the effects of constituents on the unbalanced moment-carrying capacity of corner slab–column connections. The results show that the proposed simplified model and KDS design method are strongly correlated with the experimental and FE results for a range of design parameters. Meanwhile, the ACI 318 model consistently provides a lower limit for strength prediction, thus yielding overly conservative and safe results compared with the test and FE results in most cases.