Risk-aware buckling design of functionally graded porous beams via machinelearning surrogates and reliability analysis.
Saved in:
| Title: | Risk-aware buckling design of functionally graded porous beams via machinelearning surrogates and reliability analysis. |
|---|---|
| Authors: | Pitta, Satyasaibaba1 satyasaibabapitta@gmail.com, Ginka, Ranga Janardhana2, Banavathu, Balakrishna1 |
| Source: | International Journal of Automotive & Mechanical Engineering. Mar2026, Vol. 23 Issue 1, p13367-13392. 26p. |
| Subjects: | Structural reliability, Machine learning, Engineering reliability theory, Mechanical buckling, Monte Carlo method, Shear (Mechanics), Design techniques, Girders |
| Abstract: | Functionally graded porous beams offer high stiffness-to-weight ratios, but their buckling strength is sensitive to induced porosity variability. Designers, therefore, need tools that are both fast and explicitly risk-aware. This study develops and validates an interpretable methodology that combines higher-order shear deformation theory, machine learning surrogates, and structural reliability analysis to support buckling design of functionally graded porous beams. Deterministic buckling responses are first generated using a higher-order shear deformation theory for two boundary conditions (simply supported and clamped-clamped), two slenderness ratios (L/h=10 and 40), geometric controls (taper and width), porosity indices 0 |
| Copyright of International Journal of Automotive & Mechanical Engineering is the property of Universiti Malaysia Pahang, Faculty of Mechanical Engineering and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
| Abstract: | Functionally graded porous beams offer high stiffness-to-weight ratios, but their buckling strength is sensitive to induced porosity variability. Designers, therefore, need tools that are both fast and explicitly risk-aware. This study develops and validates an interpretable methodology that combines higher-order shear deformation theory, machine learning surrogates, and structural reliability analysis to support buckling design of functionally graded porous beams. Deterministic buckling responses are first generated using a higher-order shear deformation theory for two boundary conditions (simply supported and clamped-clamped), two slenderness ratios (L/h=10 and 40), geometric controls (taper and width), porosity indices 0<a<0.3, and two porosity patterns (Regular and Uneven). A gradient-boosted tree surrogate is then trained on the log-transformed dimensionless critical buckling load using five-fold crossvalidation. Manufacturing variability in porosity is propagated through the surrogate via Monte Carlo simulation, and a companion First-order Reliability Method (FORM) in the porosity dimension provides efficient estimates of failure probability. Risk is summarized through domain-wise mean buckling capacity and Conditional Value-at-Risk (CVaR) at the 5% level (CVaR5%), from which risk-return frontiers and "knee" designs distinguish risk-neutral from risk-averse choices. The surrogate generalizes strongly (out-of-fold log coefficient of determination, R2log=0.94; mean absolute error, MAE=2.7; mean absolute percentage error, MAPE=9.5%), while FORM tracks Monte Carlo with near-perfect concordance (average R2=0.9997; MAE(Pf)=0.0005; MAPE=0.33%), with small, interpretable deviations confined to slender, simply supported beams with uneven porosity. Across all regions, clamped-clamped, regular, and thick beams systematically dominate both mean and lower tail performance. Knee designs typically incur only 1-2% loss in mean capacity while retaining 95-99% of the best attainable CvaR, providing a reliable basis for compact, risk-aware buckling design guidance. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 22298649 |
| DOI: | 10.15282/ijame.23.1.2026.16.1014 |
