Article
Engineering, Industrial
Gengxiang Chen, Yingguang Li, Xu Liu, Charyar Mehdi-Souzani, Qinglu Meng, Jing Zhou, Xiaozhong Hao
Summary: This paper proposes a physics-guided neural operator to directly predict the high-dimensional temperature history from the given cure cycle. By integrating domain knowledge into a time-resolution independent parameterised neural network, the mapping between cure cycles to temperature histories can be learned using a limited number of labelled data. Detailed experiments show that the proposed model can accurately predict the temperature histories and provide better process optimisation results.
JOURNAL OF MANUFACTURING SYSTEMS
(2023)
Article
Chemistry, Physical
Linnan Tu, Yingju Yang, Jing Liu
Summary: A machine learning model was proposed to predict the HER performance of single-atom chalcogenide catalysts, with the band gap of support materials identified as the most important descriptor. Sn@CoS and Ni@ZnS exhibit excellent catalytic activity towards HER, even outperforming current most efficient Pt catalysts.
INTERNATIONAL JOURNAL OF HYDROGEN ENERGY
(2022)
Article
Automation & Control Systems
Huan Liu, Shuo Liu, Zheng Liu, Nezih Mrad, Abbas S. Milani
Summary: Composite materials are crucial in the aerospace industry, but delamination poses a threat to their structural integrity. This article proposes data-driven methods to accurately quantify delamination area and address the problem of insufficient inspection data. Experimental results show that the proposed ensemble learning-based model outperforms other methods in terms of prediction accuracy and efficiency.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2021)
Article
Materials Science, Characterization & Testing
Guanyu Piao, Jorge Mateus, Jiaoyang Li, Ranjit Pachha, Parvinder Walia, Yiming Deng, Sunil Kishore Chakrapani
Summary: This article presents the use of phased array ultrasonic testing method to characterize the adhesive interface between thermoplastic composites. A deep learning algorithm is proposed to classify different adhesion conditions based on a set of physics-based damage indices extracted from the ultrasonic images. The experimental results show that support vector machine performs better than other machine learning algorithms, achieving a classification accuracy of over 95%.
NONDESTRUCTIVE TESTING AND EVALUATION
(2023)
Article
Computer Science, Artificial Intelligence
Noel P. Greis, Monica L. Nogueira, Sambit Bhattacharya, Catherine Spooner, Tony Schmitz
Summary: Physics-guided machine learning (PGML) offers a new approach to stability modeling during machining by leveraging experimental data and combining theoretical process modeling efforts. This research explores strategies for updating the machine learning model with experimental data, aiming to achieve a useful approximation of the true stability model while reducing the number of required measurements.
JOURNAL OF INTELLIGENT MANUFACTURING
(2023)
Article
Ecology
Mohannad Elhamod, Kelly M. Diamond, A. Murat Maga, Yasin Bakis, Henry L. Bart, Paula Mabee, Wasila Dahdul, Jeremy Leipzig, Jane Greenberg, Brian Avants, Anuj Karpatne
Summary: Species classification is a crucial task laying the groundwork for various applications involving species studies. The proposed hierarchy-guided neural network (HGNN) method outperforms traditional ConvNet models in terms of classification accuracy and robustness, especially in scenarios with limited training data.
METHODS IN ECOLOGY AND EVOLUTION
(2022)
Article
Engineering, Civil
Mohamadreza Sheibani, Ge Ou
Summary: This article proposes a framework for maximizing the information gain of earthquake damage through active learning, thus optimizing resource allocation. By selecting the most informative buildings for inspection during the reconnaissance process, accurate predictions of damage for the majority of buildings can be made within 1 week of damage inspections.
EARTHQUAKE SPECTRA
(2022)
Article
Acoustics
Faeez Masurkar, Saurabh Aggarwal, Zi Wen Tham, Lei Zhang, Feng Yang, Fangsen Cui
Summary: This research focuses on estimating the elastic constants of orthotropic laminates using ultrasonic guided waves and inverse machine learning models. The results show that this approach has the potential to accurately predict the elastic constants of a material and reduce computational time.
Review
Engineering, Multidisciplinary
Hongguang Yun, Rakiba Rayhana, Shashank Pant, Marc Genest, Zheng Liu
Summary: Nondestructive testing and evaluation (NDT&E) is commonly used in the industry for identifying damage and assessing material conditions. Ultrasonic testing (UT) is a popular technique, with nonlinear ultrasonic testing (NUT) offering advantages over conventional methods. Machine learning methods show promise for analyzing complex nonlinear ultrasonic signals.
Article
Construction & Building Technology
Faramarz Bagherzadeh, Torkan Shafighfard
Summary: This study provides a correlation between the structural performance and mechanical properties of carbon nano-tubes reinforced cementitious composites through efficient predictive Machine Learning models. Random Forest and Gradient Boosting Machine were implemented for predicting the properties, and the results show that the GBM model performs better.
CASE STUDIES IN CONSTRUCTION MATERIALS
(2022)
Article
Mechanics
Bruno Loureiro, Cedric Gerbelot, Maria Refinetti, Gabriele Sicuro, Florent Krzakala
Summary: This manuscript develops a quantitative and rigorous theory for studying the fluctuations in an ensemble of generalized linear models trained on high-dimensional correlated features. The results can be applied to various classification and regression tasks and help understand the impact of ensembling on test error as well as the roots of the "double-descent" phenomenon.
JOURNAL OF STATISTICAL MECHANICS-THEORY AND EXPERIMENT
(2023)
Article
Chemistry, Multidisciplinary
Thijs Stuyver, Connor W. W. Coley
Summary: Bioorthogonal click chemistry is an essential tool for biochemists. This study presents a computational workflow to discover new bioorthogonal click reactions with promising properties. By sampling a small portion of the search space, a machine learning model is developed and able to accurately predict activation and reaction energies. The screened search space identifies a diverse pool of candidate reactions for future experimental development.
CHEMISTRY-A EUROPEAN JOURNAL
(2023)
Article
Geochemistry & Geophysics
Kyubo Noh, David Pardo, Carlos Torres-Verdin
Summary: Deep learning inversion is a promising method for real-time interpretation of logging-while-drilling (LWD) resistivity measurements. We develop a method to enhance the robustness of DL inversion methods in the presence of noisy LWD resistivity measurements by generating training data sets and constructing DL architectures.
GEOPHYSICAL JOURNAL INTERNATIONAL
(2023)
Review
Geosciences, Multidisciplinary
Adoubi Vincent De Paul Adombi, Romain Chesnaux, Marie-Amelie Boucher
Summary: With recent technological advances, hydrogeologists now have access to large amounts of real-time data, but traditional modelling tools face challenges. Artificial intelligence and machine learning may be the future of hydrogeological research and applications. Theory-guided machine learning methods can help overcome limitations of machine learning, reducing model opacity and enhancing convergence and generalizability.
HYDROGEOLOGY JOURNAL
(2021)
Article
Engineering, Electrical & Electronic
James Lee, Jun-Ha Yun, Jae-Hun Shim, Suk-Ju Kang
Summary: In this study, we propose a novel method that uses clustering algorithm to detect multitouch coordinates, aiming to solve the big-finger problem. The method utilizes pixel intensity-weighted K-means clustering to prevent biased coordinate detection and employs an efficient CNN-based multitouch classification network to accurately classify the number of touches in adjacent-touch data. Experimental results demonstrate that our algorithm achieves the most accurate touch coordinate detection among all touch scenarios.
IEEE SENSORS JOURNAL
(2023)
Article
Mechanics
R. Sourki, B. Crawford, R. Vaziri, A. S. Milani
Summary: This study demonstrates that the loading/unloading regimes are often overlooked during the design and simulation of woven fabric structures and forming processes. Experimental results show that local bending/reverse-bending can occur during a typical forming process, impacting critical mechanical properties of the fabric. The study also highlights the importance of considering multiple parameters in predicting the cyclic bending response of woven fabrics.
COMPOSITE STRUCTURES
(2022)
Article
Forestry
Shaikh Atikur Rahman, Mahmud Ashraf, Mahbube Subhani, Johannes Reiner
Summary: This study compares the performance of four continuum damage models for simulating the behavior of timber materials and validates the reliability of the MAT-261 model in simulating timber fracture behavior.
EUROPEAN JOURNAL OF WOOD AND WOOD PRODUCTS
(2022)
Article
Materials Science, Multidisciplinary
Yun-Fei Fu, Johannes Reiner
Summary: This paper presents a systemic calibration methodology for efficiently simulating progressive damage evolution in four different pultruded GFRP composites. The results show that the best set of input parameters can yield accurate crack length predictions, and the incorporation of bi-linear softening laws improves simulation results.
INTERNATIONAL JOURNAL OF DAMAGE MECHANICS
(2022)
Review
Forestry
John Paul Cabral, Bidur Kafle, Mahbube Subhani, Johannes Reiner, Mahmud Ashraf
Summary: Timber densification is a process that enhances the structural properties of timber by altering its cellular structure through compression, chemical impregnation, or a combination of both. The density and mechanical properties of the timber can be modified by adjusting parameters such as compression ratio and temperature. The current processes, effects, and potential future directions of timber densification are discussed in this paper.
JOURNAL OF WOOD SCIENCE
(2022)
Article
Materials Science, Ceramics
Johannes Reiner, Darren Narain, Peng Zhang, Emmanuel A. Flores-Johnson, Ondrej Muransky
Summary: This study quantifies the damage resistance of cross-ply C/C composites using compact tension tests at room temperature. The analysis of different specimen sizes shows that baseline and large scaled-up samples yield consistent fracture energy values, while the scaled-down version shows unwanted failure. A microscopic cross-sectional analysis explains the relatively low fracture energy values of C/C composites compared to carbon fibre reinforced polymers.
CERAMICS INTERNATIONAL
(2023)
Article
Materials Science, Multidisciplinary
Udesh M. H. U. Kankanamge, Johannes Reiner, Xingjun Ma, Santiago Corujeira Gallo, Wei Xu
Summary: With the increasing use of CubeSats in space exploration, the demand for reliable high-temperature shape memory alloys (HTSMA) continues to grow. This study uses a data-driven approach to identify suitable NiTiHf alloys for actuator applications in space. Of the machine learning models evaluated, the K-nearest neighbouring model offers reliable and accurate prediction in developing NiTiHf alloys with balanced functional properties.
JOURNAL OF MATERIALS SCIENCE
(2022)
Article
Mechanics
Reza Sourki, Behnaz Khatir, Saeed Shaikhzadeh Najar, Reza Vaziri, Abbas S. Milani
Summary: This study investigates the bending behavior of unconsolidated fabrics under large deformation. The interlaced architecture and dissipative behavior of yarns in fabrics affect the effective bending rigidity and hysteresis effects. In addition, the bending rigidity interacts with in-plane trellising, suggesting limitations of classical laminate theory in capturing shear-bending coupling in dry fabrics.
COMPOSITE STRUCTURES
(2023)
Article
Engineering, Mechanical
Johannes Reiner, Sergio Orellana Pizarro, Kenny Hadi, Darren Narain, Peng Zhang, Matt Jennings, Mahbube Subhani
Summary: This study investigates the mechanical properties of thin quasi-isotropic [90°-45°-0°-45°] beech veneer laminates. The results show that beech veneer laminates can be tested and analyzed similarly to fiber-reinforced laminates, with consistent values of strength and damage resistance. However, the modulus, tensile strength, open-hole strength, and translaminar fracture energy of beech veneer laminates are one order of magnitude lower compared to fiber-reinforced polymer composites.
ENGINEERING FAILURE ANALYSIS
(2023)
Article
Materials Science, Composites
Johannes Reiner, Nhu H. T. Nguyen
Summary: This study proposes a Discrete Element Method (DEM) for analyzing the progressive failure of IM7/8552 carbon fiber reinforced polymer laminates at the macroscale. The DEM model is calibrated using experimental results from over-height compact tension (OCT) tests and validated with various open-hole tension (OHT) test specimens. The results show that DEM can incorporate damage evolution leading to realistic macroscopic damage patterns, but with a computational cost 100 times higher than finite element (FE) simulations.
JOURNAL OF COMPOSITE MATERIALS
(2023)
Article
Polymer Science
Mathew Wynn, Navid Zobeiry
Summary: The morphology and performance of semicrystalline thermoplastic composites are affected by processing parameters, such as temperature history. The final morphology is determined by the competition between spherulite growth in resin-rich areas and transcrystallinity growth from fiber surfaces. This study used a polarized microscope equipped with a heating and cooling controlled stage and a probabilistic machine learning approach, Gaussian Process Regression (GPR), to study the growth of crystals in low volume fraction PEEK-carbon fiber composites. GPR revealed that growth kinetics of spherulites follows the established Lauritzen-Hoffman equation, while transcrystallinity growth deviates from the theory. The competition between diffusion and secondary nucleation at growth front of spherulites and transcrystalline regions was deconvoluted using a combined GPR model and Lauritzen-Hoffman equation.
Article
Mechanics
Yun-Fei Fu, Johannes Reiner
Summary: This study explores the application of genetic algorithms for the objective and automated calibration of damage models in composite materials. The method is demonstrated to be generally applicable and robust through three case studies involving carbon and glass fiber-reinforced laminates. The load-displacement curves of fracture tests are used to optimize the input parameters of the damage models. The optimized parameters produce accurate and physically meaningful results, as validated in independent load cases and through good correlation with experimental observations.
COMPOSITE STRUCTURES
(2023)
Article
Automation & Control Systems
Ehsan Haghighat, Sahar Abouali, Reza Vaziri
Summary: Constitutive models are fundamental in modeling physical processes by connecting conservation laws with system kinematics. However, characterizing these models can be challenging, especially in nonlinear regimes. We believe that theory-based parametric elastoplastic models are still the most efficient and predictive.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Engineering, Manufacturing
Caleb Schoenholz, Shuangshan Li, Kyle Bainbridge, Vy Huynh, Alex Gray, Navid Zobeiry
Summary: This paper presents an in-situ inspection method to evaluate the physicochemical properties of release coating and the surface condition of large production tools. The proposed method utilizes global mapping, sparse sensing, and machine learning to quickly identify the condition of release coating or contamination on tool surfaces. The results show significant chemical changes in aerospace-graded release coatings after multiple autoclave processing cycles.
JOURNAL OF MANUFACTURING AND MATERIALS PROCESSING
(2023)
Article
Mechanics
Johannes Reiner, Nathaniel Linden, Reza Vaziri, Navid Zobeiry, Boris Kramer
Summary: This study combines finite element analysis, machine learning, and Markov Chain Monte Carlo to estimate the probability density of input parameters for progressive damage simulation in fiber-reinforced composites. By conducting numerous FEA simulations with randomly varying input parameters and using synthetic data to train a neural network, a highly efficient surrogate model is developed. The application of Markov Chain Monte Carlo algorithms, along with statistical test data from experiments, enables Bayesian parameter estimation and determination of virtual design allowables.
COMPOSITE STRUCTURES
(2023)
Article
Engineering, Industrial
Austin Lee, Mathew Wynn, Liam Quigley, Marco Salviato, Navid Zobeiry
Summary: This study investigated the effect of additive manufacturing parameters on the thermal history and crystalline morphology of PEEK using a combined experimental and numerical approach. It was found that the high melting temperature of PEEK resulted in fast melt cooling rates and short annealing times, leading to relatively low degree of crystallinity and small crystalline morphology during printing.
ADVANCES IN INDUSTRIAL AND MANUFACTURING ENGINEERING
(2022)
Article
Mechanics
Rawan Aqel, Patrick Severson, Rani Elhajjar
Summary: A novel core splice joint configuration for composite sandwich structures is studied and proposed to improve the strength and toughness. Experimental and numerical efforts show that this configuration can significantly increase the ultimate strength by 13% to 51% and the toughness by 2% to 35%.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Xianheng Wang, Cong Chen, Jinsong Zhang, Xinming Qiu
Summary: In this paper, a new form-finding method based on spatial elastica model (FMSE) is proposed for elastic gridshells. The method integrates the deformations of elastic rods into the overall deformation of the gridshell, and solves a set of transcendental equations using the quasi-Newton method to ensure the deformation satisfies the given boundary conditions. The method is validated through experiments and expected to have potential applications in the investigations of elastic gridshells.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Hao Huang, Zitong Guo, Zhongde Shan, Zheng Sun, Jianhua Liu, Dong Wang, Wang Wang, Jiale Liu, Chenchen Tan
Summary: The conventional evaluation of 3D braided composites' mechanical properties through numerical and experimental methodologies hinders material application due to the expenses, time constraints, and laborious efforts involved. This study establishes a multi-scale finite element model and a surrogate model for predicting the elastic properties of 3D4D rotary braided composites with voids. By optimizing a neural network model, the results are validated and provide valuable insights into the microstructure and properties of these composites.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Xinyu Li, Hao Zhang, Haiyang Yang, Junrong Luo, Zhongmin Xiao, Hongshuai Lei
Summary: Due to their excellent mechanical properties and design flexibility, fluted-core composite sandwich structures have gained significant attention in aerospace and rail transit applications. This study investigated the free-vibration characteristics and optimized design of composite fluted-core sandwich cylinders through theoretical models and experimental tests.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Chao Li, Chunzheng Duan, Xiaodong Tian, Chao Wang
Summary: A mechanistic model considering the bottom edge cutting effect and the anisotropic characteristics of the material is proposed in this paper to accurately predict cutting forces. The model was validated through a series of milling experiments and can be used to predict the cutting force of various parts of the cutter and any feed direction.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Camila Sanches Schimidt, Leopoldo Pisanelli Rodrigues de Oliveira, Carlos De Marqui Jr
Summary: This work investigates the vibro-acoustic performance of graded piezoelectric metamaterial plates. The study shows that piezoelectric metamaterial plates with reconfigurable properties can provide enhanced vibration and sound power attenuation.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Jun Ke, Li-jie Liu, Zhen-yu Wu, Zhong-ping Le, Luo Bao, Dong-wei Luo
Summary: Compared with other green natural fibers, ramie has higher mechanical properties and lower cost. In this study, ramie and glass fiber are made into composite circular tubes. The results show that the hybrid circular tube with ramie and glass fiber has improved torsional mechanical properties and reduced weight and cost. The failure mechanisms are affected by the loading direction and the content of each fiber.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Natalia Pingaro, Gabriele Milani
Summary: This paper proposes an enhanced analytical model for predicting the behavior of FRCM samples tested under standard tensile tests. The model takes into account the interaction between fibers and matrix through the interface, and assumes different material properties at different phases. By solving a second order linear differential equation, an analytical solution can be obtained. The model is validated with experimental data and shows good predictability.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Jialiang Fan, Anastasios P. Vassilopoulos, Veronique Michaud
Summary: This article investigates the effects of voids, joint geometry, and test conditions on the fracture performance of thick adhesive Double Cantilever Beam (DCB) joints. It concludes that grooved DCB joints with low void content tested at low displacement rates showed stable crack propagation without significant crack path deviation.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Auwalu I. Mohammed, Kaarthikeyan Raghupathy, Osvaldo De Victoria Garcia Baltazar, Lawson Onokpasah, Roger Carvalho, Anders Mogensen, Farzaneh Hassani, James Njuguna
Summary: This study investigates the performance of composite pressure vessels under damaged and undamaged conditions, providing insights into their reliability and residual strength capabilities. The results demonstrate that the damage profile and its effect on compressive strength are similar between damaged and non-damaged cylinders. When subjected to quasi-static compression, the polyethylene liner absorbs enough elastic strain energy to recover without plastic deformation. Additionally, quasi-static compression has little to no influence on the axial strength of the cylinders. The damage characterization reveals fiber breakage, delamination, local buckling, and brooming failure. This study has direct implications for the safety design tolerances, manufacturing strategies, and operational failure conditions of composite overwrapped pressure vessels (COPVs).
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Muhammad Irfan Shirazi, Samir Khatir, Djilali Boutchicha, Magd Abdel Wahab
Summary: Structural health monitoring is important to ensure the safety of components and structures. This study proposes a method using finite element models and 1D-CNN network to extract and classify vibration responses for crack detection. The results show that the proposed approach is effective in real-time damage detection.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Maryam Mirsalehi, Kiarash Kianpour, Sharif Shahbeyk, Mohammad Bakhshi
Summary: This study comprehensively investigates the one-way response of 3D-woven sandwich panels (3DWSPs) and their interfering parameters, providing interpretation of elastic and failure results, failure maps, and reliable theoretical models for linear elastic response and observed failure mechanisms.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Yiming Zhao, Zhonggang Wang, Zhigang Yang, Bin Qin
Summary: The paper proposes a Ritz and statistical energy analysis (Ritz SEA) hybrid method for calculating rectangular plate acoustic vibration coupling in the mid-frequency range. This method combines the fast convergence and ability to handle arbitrary boundary conditions of the Ritz method with the power flow equation of the statistical energy analysis method. The results show that this approach effectively filters out random fluctuations in mid-frequency domains while demonstrating exceptional stability and precision.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Joao Henrique Fonseca, Woojung Jang, Dosuck Han, Naksoo Kim, Hyungyil Lee
Summary: This study addresses the enhancement of an injection-molded fiber-reinforced plastic / metal hybrid automotive structure and its plastic injection molding process through the integration of the finite element method, artificial intelligence, and evolutionary search methods. Experimental validation of finite element models, the generation of a database through orthogonal array and Latin hypercube methods, and the training of artificial neural networks are conducted. The genetic optimization algorithm is then applied to identify optimal process parameters. The results show significant reduction in product warpage and manufacturing time while maintaining structural strength, contributing to the advancement of composite automotive structures with superior quality.
COMPOSITE STRUCTURES
(2024)
Article
Mechanics
Alessandro Vescovini, Carina Xiaochen Li, Javier Paz Mendez, Bo Cheng Jin, Andrea Manes, Chiara Bisagni
Summary: This paper presents a study on six single-stringer specimens manufactured using the card-sliding technique with non-crimp fabrics and adopting a Double-Double (DD) stacking sequence. The specimens were tested under compression loading conditions to investigate post-buckling and failure in aerospace structures. Experimental results and numerical simulations were compared to analyze the behavior and failure modes of the specimens. The study found promising evidence of a viable solution to optimize aeronautical structures and enhance resistance to skin-stringer separation, particularly with the use of tapered flanges.
COMPOSITE STRUCTURES
(2024)