A Faculty Member at the College of Engineering, University of Kirkuk, Publishes a Research Paper on Modeling and Analysis of Elliptical Concrete Columns Using Finite Element and Machine Learning Techniques

Dr. Diyar Nasih Qadir, a faculty member in the Civil Engineering Department at the College of Engineering, University of Kirkuk, has published a research paper in the British journal Structures, issued by Elsevier and ranked in the first quartile (Q1) with an impact factor (IF) of 3.9. The study focuses on analyzing the structural behavior of fiber-reinforced polymer (FRP) confined elliptical concrete columns using numerical modeling and machine learning techniques.

Elliptical sections are increasingly used in architectural designs, necessitating a comprehensive understanding of their structural performance under different loading conditions. In this research, ABAQUS finite element analysis (FEA) was employed to simulate the behavior of FRP-confined elliptical concrete columns, based on previously published experimental data. The accuracy of the numerical model was validated by comparing it with 45 experimental samples, showing a high correlation between numerical results and real-world data. Additionally, the study was expanded to include 40 more parametric models, bringing the total analyzed samples to 85.

The findings indicate that increasing the unconfined concrete strength reduces the efficiency of FRP confinement, as the containment effect becomes less effective, particularly in high-strength concrete columns unless sufficient FRP stiffness is provided. High-strength columns were also found to be more prone to failure post-peak stress without adequate confinement. Conversely, increasing the number of FRP layers significantly improved the stress and deformation capacity of elliptical concrete columns, enhancing their overall structural performance.

Since physical experiments require substantial time and cost, while numerical modeling depends on user expertise and computational settings, the research was further expanded to incorporate machine learning (ML) techniques for predicting the compressive strength of FRP-confined elliptical concrete columns. Four tree-based ML algorithms were used: Decision Tree, Random Forest, Gradient Boosting, and XGBoost. The results demonstrated that XGBoost was the most accurate, achieving R² = 0.95 for both compressive strength and axial deformation, highlighting the potential of artificial intelligence in predicting structural material behavior with high precision.