A machine learning based prediction model for the impact mechanical response of composite laminates considering microstructure sensitive transverse properties DOI Creative Commons
Zhang Yiben,

Feng Guangshuo,

Bo Liu

et al.

Polymer Composites, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 20, 2024

Abstract The accurate and efficient prediction of impact mechanical response is crucial for safety design composite structures. In this work, high‐fidelity representative volume elements (RVEs) with fiber, matrix fiber/matrix interface are established, in which random fiber distributions considered. A failure envelope under transverse loads proposed based on computational micromechanical RVEs, it implemented by ABAQUS VUMAT subroutines to predict the laminates loads. Based a dataset from macromechanical finite element simulations, an artificial neural network model established trained. It found that distribution introduced more obvious fluctuation tension/compression strength than shear strength. criteria showed better performance Hashin Tsai‐Wu especially combined compression An ANN 8 hidden layers can achieve acceptable coefficient determination (R 2 ) 0.98 loss functions mean absolute error (MAE) 71. For certain loading conditions, well trained machine learning predicted contact force history within 30 min, while FEA costs about 75 min same computer. speed increased over 60% conditions. hence shown method provides potential alternative evaluation resistance Highlights High‐fidelity micromechanics analysis performed uncover complex relationship between microstructure strengths laminates. dependent criterion shows high accuracy compared criteria. multi‐layer rapid achieved 0.98,

Language: Английский

Predicting natural aging effects on fatigue life of CFRP–aluminum adhesive joints using machine learning and accelerated aging data DOI
Sajjad Karimi,

A. Anvari

Journal of Adhesion Science and Technology, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 22

Published: Feb. 3, 2025

Language: Английский

Citations

3
Machine Learning for predicting strength properties of waste iron slag concrete DOI Creative Commons

Matiur Rahman Raju,

Syed Ishtiaq Ahmad, Md Mehedi Hasan

et al.

Heliyon, Journal Year: 2025, Volume and Issue: 11(3), P. e42133 - e42133

Published: Jan. 23, 2025

This study investigates the utilization of waste iron slag (WIS) as a sustainable alternative in concrete production to reduce environmental impact and preserve natural resources. The experimental investigation WIS-incorporated focused on compressive tensile strength with machine learning (ML) models for prediction. Among tested ML algorithms, Decision Tree (DT) XGBoost showed highest accuracy (R2 = 0.95135) predicting properties, while like SVM Symbolic Regression underperformed. Experimental results indicate that up 20 % WIS replacement maintains adequate strength, whereas higher proportions structural integrity. A ranking score index cost analysis confirmed technical economic feasibility using concrete. Cost demonstrated substantial savings 25 incorporation, confirming its feasibility. Integrating data predictions highlights WIS's potential applications, enabling optimized mix designs reduced reliance physical testing. Future work should address limitations, including dataset expansion exploration additional durability mechanical properties validate practicality construction further.

Language: Английский

Citations

1
Research on ultrasonic non-destructive detection method for defects in GFRP laminates based on machine learning DOI
Fan Ding,

Yingqiang Cai,

Wanli Tu

et al.

Ocean Engineering, Journal Year: 2025, Volume and Issue: 327, P. 120972 - 120972

Published: March 17, 2025

Language: Английский

Citations

1
Machine Learning-Based Strength Prediction of Nano-Reinforced Adhesive and Hybrid Joints Under Hygrothermal Conditions DOI
Sajjad Karimi,

Ali B.M. Ali,

Jianyong Yu

et al.

Materials Today Communications, Journal Year: 2024, Volume and Issue: unknown, P. 111046 - 111046

Published: Nov. 1, 2024

Language: Английский

Citations

8
Nanoparticle integration in adhesive and hybrid single lap joints: effect on strength and fatigue life under environmental aging DOI
Sajjad Karimi,

Aiham O. Altayeh,

Mahboobeh Kargar Samani

et al.

Journal of Adhesion Science and Technology, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 29

Published: Oct. 11, 2024

Examining the mechanical performance of CFRP and aluminum samples subjected to environmental aging is crucial. Additionally, it essential develop methods enhance their properties. This research investigates impact fullerene single-walled carbon nanotubes (SWCNT) on fatigue life static strength bonded bonded/bolted joints. The study focuses composite-to-composite (CTC) composite-to-aluminum (CTA) substrates under three-point bending, both before after hygrothermal aging. were divided into four categories: (1) neat specimens, (2) specimens with added fullerene, (3) SWCNT, (4) a combination 50% SWCNT fullerene. experimental results indicated that optimal nanoparticle ratio for joints differs from Adding nanoparticles adhesive increased SLJs, particularly in containing mixed particles SWCNT. In some cases, amplified effect conditions, enhancing further. integration use significantly improved joint strength, techniques yielding best results. These modified offer promising alternative traditional terms life. enhances understanding hybrid joints, especially dissimilar (composite metal), provides insights behavior various conditions. show potential optimizing composite structures, improving durability, reducing likelihood operational failures.

Language: Английский

Citations

7
Buckling analysis of perforated stiffened composite plates with interfacial debonding under hygrothermal and in-plane edge loadings DOI
Akshay Prakash Kalgutkar, Sauvik Banerjee

Journal of Composite Materials, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 17, 2025

Engineered composite structures often incorporate stiffeners to enhance the strength of perforated panels without significantly increasing their mass. However, interfacial debonding between stiffener and skin can compromise structural integrity under external loads, potentially leading failure. This study focuses on buckling behaviour stiffened laminated plates with debonding, subjected non-uniform edge loading environmental conditions. A computationally efficient reduced-order finite element (FE) formulation has been devised using 2D plate 1D beam elements minimise computational cost. The flange are modelled a 9-noded heterosis address shear-locking, while 3-noded isoparametric represents web ribs. To account for torsional behaviour, torsion correction factor is incorporated into ribs formulation. Interfacial simulated by introducing dummy node an independent displacement field flange, connected fictitious spring penetrated nodes, prevent nodal interpenetration. Displacement continuity enforced in bonded regions maintain compatibility fields. employs dynamic approach evaluate loads two boundary conditions, considering operational effects. Additionally, hygrothermal-dependent material properties considered effect hygrothermal elastic material. preliminary investigation identifies optimal pattern cutout geometry enhanced performance. In contrast prior research, this work examines various configurations stability determines configuration that improves capacity. analysis indicates circular panel incorporating SP-3 demonstrates 42.86% improvement resistance compared SP-2 design. Furthermore, reduces capacity CCSS 20.14%, especially at greater depths d s /b = 7 Moreover, larger sizes exacerbate reductions 2.70% due reference state. contrast, conditions results 38.44% reduction smaller SSCC panel, highlighting impact restraint Therefore, serves as foundation optimising designs ensure stability, durability, cost-effectiveness demanding scenarios.

Language: Английский

Citations

0
Thermal and mechanical performance analysis of basalt and glass fiber-reinforced polyurethane-epoxy composites DOI

Sabir Ali,

Jianxun Liu,

Innocent Chikira Msangi

et al.

Journal of Reinforced Plastics and Composites, Journal Year: 2025, Volume and Issue: unknown

Published: March 13, 2025

This study explores the tensile and bending behavior of fiber-reinforced polymer (FRP) composites under a wide range thermal conditions, focusing on epoxy- polyurethane-based matrices reinforced with glass (GFRP) basalt (BFRP) fibers. Comprehensive experiments were conducted at temperatures ranging from −25°C to 400°C evaluate mechanical performance failure mechanisms these materials. The results demonstrate that epoxy-based exhibited superior stability compared systems. Among tested materials, E-BFRP demonstrated best performance, maintaining 61% its strength 79.4% stress 400°C. E-GFRP moderate resistance, whereas composites, notably PGFRP-K, showed considerable degradation, reductions exceeding 72% losses surpassing 70% elevated temperatures. Sub-zero had negligible effects properties, but beyond 250°C induced resin decomposition, fiber pull-out, diminished fiber-matrix interaction, as confirmed through scanning electron microscopy (SEM). These findings underscore critical importance material selection matrix optimization for FRP intended use in high-temperature environments, providing valuable insights advancing structural designs.

Language: Английский

Citations

0
Integrating machine learning and digital twin for strength prediction of CFRP/aluminum adhesive joints under hygrothermal conditions DOI
Noor Hadi Aysa, Sajjad Karimi

Polymer Composites, Journal Year: 2025, Volume and Issue: unknown

Published: April 15, 2025

Abstract This study investigates the application of machine learning models integrated with a digital twin (DT) framework to predict and correlate performance carbon fibre‐reinforced polymer‐to‐aluminum adhesive joints subjected hygrothermal aging. By combining experimental methods techniques, research aims bridge gap between effects natural accelerated aging on joints. The were manufactured then left age naturally for period one 3 years. For aging, conditions four 50 days. Three‐point bending tests utilized evaluate To periods using data, five algorithms used: support vector regression (SVR), artificial neural network (ANN), linear regression, random forest (RF) XGBoost. scanning electron microscopy (SEM) analyses showed that moisture absorption caused substantial change in surface morphology aluminum adherends, including increased roughness crystalline formations. results indicated XGBoost has provided almost perfect predictions since MSE values equal 0 observed at all iterations, highlighting its accuracy reliability. In contrast, SVR demonstrated much lower accuracy, clear differences their predictions. integration approaches turns out be most efficient method real‐time adaptation model as well accurate prediction, enhancing durability reliability composite structures. Highlights Strength prediction by Machine twin. SEM revealed moisture‐induced changes morphology. high accuracy.

Language: Английский

Citations

0
Prediction of residual stresses in GFRP strips under wind-sand erosion by interpretable machine learning methods: feature engineering and SHAP analysis DOI
Wenhao Ren,

A Siha,

Changdong Zhou

et al.

Multiscale and Multidisciplinary Modeling Experiments and Design, Journal Year: 2025, Volume and Issue: 8(6)

Published: April 15, 2025

Language: Английский

Citations

0
High-velocity impact performance prediction of 3D angle-interlock woven composites based on machine learning approach DOI
Chao Zhang,

Tianhuan Chen,

Zeyu Bian

et al.

Mechanics of Advanced Materials and Structures, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 13

Published: April 23, 2025

Language: Английский

Citations

0