Title:
Study on the Behavior of Shear-Critical UHPC Beams Using Machine Learning and Finite Element Analysis
Author(s):
Amjad Diab
Publication:
Web Session
Volume:
ws_S23_AmjadDiab.pdf
Issue:
Appears on pages(s):
Keywords:
DOI:
Date:
4/2/2023
Abstract:
The work presented herein investigates the feasability of employing machine learning-based constitutive models for the tension response of UHPC within a finite element analysis (FEA) framework in order to determine the behavior of shear-critical UHPC beam elements. A machine learning (ML) model was developed to analyze the complez relationships between the UHPC mix design and the material tensile behavior in terms of cracking and post-cracking characteristics. The ML-generated results were reasonably accurate in predicting the tensil properties of UHPC, with an R2 of 0.92 for the ultimate tensile strength, based on 491 data points. The ML-generated UHPC properties were then employed within a FEA methodology that uses a secant-stiffness, smeared, rotating-crack formulation to simulate the response of shear-critical UHPC beams reported in the literature. The results were accurate in depicting the behavior of elements in terms of stiffness, shear capacity, and cracking pattern.