Ultrasonic welding (UW) provides a rapid and efficient method for joining composite components by inducing resin flow through thermally driven diffusion and crystallization at the bonded interface. However, in the absence of a multiphysics modeling framework or a digital twin approach, current practice still depends on extensive trial-and-error testing to determine key welding parameters such as vibration amplitude, weld time, weld pressure, hold time, and downspeed. While in-situ thermal cameras can monitor surface temperatures, the internal temperature at the bonded interface is often significantly higher, introducing the risk of thermal degradation and inconsistent bond quality. To overcome these limitations, GEM developed a high-fidelity multiphysics model to establish a quantitative relationship between process parameters and the evolving temperature field within welded thermoplastic parts. The model integrates coupled mechanical, thermal, and acoustic physics to simulate high-frequency vibrations and static pressure, capture the generation and spatial distribution of heat, and represent the temperature-dependent viscoelastic response that governs bond formation. A validation test matrix was designed by systematically varying weld time and vibration amplitude. Through-thickness temperature distributions were measured using infrared thermal imaging, enabling direct comparison with model predictions. Upon validation, the model was applied for process tailoring, allowing precise control of temperature distribution to achieve target bond strength. This integrated modeling and validation approach demonstrated substantial benefits, including reduced design iterations, accelerated process optimization, and improved quality and performance of welded composite structures.

In this study, IM7/Cycom 5320-1 unidirectional prepreg has been utilized to manufacture 16-layer laminated composites: a symmetric cross-ply ([0 degrees/90 degrees]4s) and a quasi-isotropic ([45 degrees/90 degrees/-45 degrees/0 degrees]2s) configuration. Microwave and autoclave curing processes have been employed to manufacture the laminated composites. The manufactured composite cure was assessed using differential scanning calorimetry (DSC). The quality and porosity of the microwave-cured parts were juxtaposed to those of autoclave-cured parts through optical microscopy and micro-computed tomography (micro-CT) scanning. Mechanical characterization of the microwave-cured panels was conducted using uniaxial tensile and flexural tests, with results juxtaposed to autoclave-cured samples. Experimental characterization revealed that the microwave-cured parts exhibited nearly identical void content and mechanical performance as the autoclave-cured parts. A multiphysics modeling approach using COMSOL was developed to foresee the manufactured composites degree of cure and temperature distribution. The numerical results were validated against DSC analysis for the degree of cure, and the temperature profile was compared to the manufacturer-endorsed cure cycle. This multiphysics model was subsequently used for a parametric study to predict temperature distribution at 90%, 95%, 105%, and 110% of the actual microwave input power. Additionally, the model simulated temperature distribution at various laminate thicknesses (5, 3.75, 2.5, and 1.25 mm).

Microwave curing is a fast, energy-efficient, and a viable alternative to conventional thermal curing processes. It has been widely adopted for processing carbon fiber-reinforced polymer composites because the high electrical conductivity of carbon fibers enables strong microwave coupling. In this study, IM7/Cycom 5320-1 unidirectional prepreg has been used to fabricate 64-layer laminated composites. Symmetric cross-ply ([0 degrees/90 degrees]16S) and a quasi-isotropic ([45 degrees/90 degrees/-45 degrees/0 degrees]8S) layup have been investigated. A custom-built six-magnetron microwave applicator and an autoclave were employed to manufacture the composite panels. Degree of cure of the manufactured laminates was evaluated via differential scanning calorimetry. Interfacial bonding and porosity of the microwave-cured laminates were assessed and compared with autoclave-cured specimens using optical microscopy and micro-computed tomography. Mechanical performance of the microwave-cured composites has been juxtaposed to autoclave-cured composites in terms of the uniaxial tensile, three-point flexural, and open-hole tensile properties. Experimental results demonstrate that microwave-cured laminates possess void content, degree of cure, and mechanical properties as good as those of laminates manufactured by autoclave. The microwave curing cycle reduced the overall processing time by 23.3%, exhibited up to 16.75% higher tensile strength, and 23.20% higher flexural modulus as compared to the autoclave-cured laminates. Scanning electron microscopy has been used to investigate the surface topology of the manufactured specimens. Outcomes of this study highlight the potential of multi-magnetron microwave curing as a viable, energy-efficient alternative for fabricating high-performance carbon fiber-reinforced polymer composites. Polymer International published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

This paper presents a high-fidelity fatigue damage modeling framework for composite structures with ply drops, incorporating several key advancements to capture localized fatigue behavior. The approach includes: (1) computation of local stress ratios at each fatigue cycle; (2) an R-ratio-dependent fatigue damage accumulation model; (3) implementation of a constant load diagram to construct S–N curves at arbitrary R-ratios; and (4) a cycle-jumping technique to account for the evolving rate of fatigue damage accumulation due to progressive stiffness redistribution. A combined experimental and numerical study was conducted on tapered composite beams subjected to mixed axial tension and vertical bending. A custom-designed fatigue test fixture was developed to capture displacement at the loading end, which was then used as a boundary condition in the fatigue life prediction model. To guide the selection of fatigue test peak loads, static failure analyses were first performed on representative tapered beams under constant axial tension and monotonic bending. Subsequent fatigue tests, conducted under constant axial tension and cyclic bending at two peak load levels, showed strong agreement between predicted and measured load–deflection responses, fatigue lives, and failure patterns, thereby demonstrating the accuracy and robustness of the developed framework.

Flow-particle interaction is commonly modeled by coupling computational fluid dynamics (CFD) and discrete element methods (DEM). However, the two prevalent methods for CFD-DEM coupling-unresolved and semi-resolved schemes-put specific requirements on the ratio of mesh size to particle diameter, hindering their applications to large-scale industrial processes with complex geometries. This paper introduces a unified unresolved/semi-resolved framework for effective flow-particle interaction simulations. A moving domain CFD using Arbitrary Lagrangian-Eulerian Variational Multiscale (ALE-VMS) and a DEM model are carefully integrated via two-way momentum exchange. A novel method is developed to enable smooth transitions between the semi-resolved and unresolved schemes on non-uniform unstructured meshes. We validate the proposed ALE-VMS-DEM framework by simulating a single-particle sedimentation process and a multi-particle solid-liquid fluidization process. Good accuracy has been achieved. We further apply it to simulate solid-liquid mixing in a stirred tank and powder dynamics during directed energy deposition processes. The simulation results show great agreement with experimental data, demonstrating the effectiveness of the proposed method in addressing large-scale, practical engineering problems with complex geometries.

Hybrid additive manufacturing (AM) and subtractive manufacturing (SM) processes utilize the combination of AM (e.g., LPBF and DED) and SM (e.g., milling and turning operations) to produce the final part. Due to the poor surface roughness resulting from the uneven melting of powders in AM, the subtractive process is a necessary finishing operation to improve the surface roughness of the AM part. The hybrid AM/SM technology combines the benefits of AM and SM processes to create complex geometry while introducing good surface finish and compressive stress to prevent crack initiation. However, the relationship between large process parameter space and the residual stress/distortion in the part is not well understood, which impedes the adoption of hybrid AM/SM to minimize the residual stress in the final product. To expedite the process optimization, we establish a pipeline for the sequential modeling of additive manufacturing (AM) and subtractive manufacturing (SM) processes. Key accomplishments achieved under this study include (1) development of thermal abstraction technique for the AM process to speed up the macroscale level heat transfer analysis based on the manufacturing factors including scanning vector, laser power, dwelling time, etc.; (2) development of the sequentially coupled thermal-mechanical model to predict the residual stress and distortion after AM process by passing the temperature history obtained from heat transfer analysis to the mechanical analysis at each time point; (3) validation of the thermal-mechanical model for AM using thin-wall structure from literature and cantilever beam structure from UNT’s experiments data; (4) conduction of the parametric study on the chamber temperature and part design in the AM process to demonstrate how the temperature gradient and supporting structure affect the residual stress and distortion; (5) exploration of macro and micro scale models to predict the bulk and surface residual stress after cutting; (6) applying the developed modeling framework to tailoring the hybrid AM/SM process. To support model verification and demonstration, we print cantilever beam structure with different supporting structure designs and cutting strategies to study how these factors affect the final part residual stress and distortion. The data collected in the printing and cutting process is used to examine the applicability of the developed simulation tool.

Large language models (LLMs) bear promise as a fast and accurate material modeling paradigm for evaluation, analysis and design. Their vast number of trainable parameters necessitates a wealth of data to achieve accuracy and mitigate overfitting. However, experimental measurements are often limited and costly to obtain in sufficient quantities for fine-tuning. To this end, here we present a physics-based training pipeline that tackles the pathology of data scarcity. The core enabler is a physics-based modeling framework that generates a multitude of synthetic data to align the LLM to a physically consistent initial state before fine-tuning. Our framework features a two-phase training strategy: utilizing the large-in-amount but less accurate synthetic data for supervised pretraining, and fine-tuning the phase-1 model with limited experimental data. We empirically demonstrate that supervised pretraining is vital to obtaining accurate fine-tuned LLMs, via the lens of learning polymer flammability metrics where cone calorimeter data are sparse.

Quenching is the most critical step in the sequence of heat-treating operations, aiming to preserve the solid solution formed at the solution heat-treating temperature by rapidly cooling the material to near room temperature. Currently, there is no reliable, performance-informed quenching process that can consistently reduce the high scrap rate of airframe aluminum forging parts, which often suffer from significant residual stress and distortion. This limitation stems from the complex interactions between temperature, phase transformations, and stress/strain behavior—each influenced by the evolving temperature distribution and microstructural state of the workpiece. Conventional modeling techniques for quenching processes typically lump these multiscale, multi-physics phenomena into a simplified heat transfer coefficient (HTC). However, determining the spatial and temporal variations of HTC through experiments is both prohibitively time-consuming and costly. To address this challenge and enable rapid process tailoring for reduced distortion, we have developed and validated a digital twin-based Quenching Laboratory Software (QLAB) tool. QLAB integrates a thermal multi-phase computational fluid dynamics (CFD) model, sequentially coupled with a Mechanical Threshold Stress (MTS) model and a precipitation model. The thermal CFD component captures turbulent flow, multi-phase transformations, and the complex heat transfer stages of quenching including vapor blanket formation, nucleate boiling, and convection—to accurately predict temperature evolution. The MTS-precipitation model quantifies the effects of microstructural precipitates on the material's mechanical response under thermal loading. QLAB has been thoroughly validated using representative aluminum airframe components, including aluminum bars with pockets and Lcorner parts. We demonstrate the tool's predictive accuracy by comparing its output against experimentally measured temperature and distortion fields. Finally, we apply the validated QLAB to conduct a virtual quenching test on a simplified aluminum airframe structure, showcasing its potential for performance-informed process optimization.

Quenching is a heat treatment process for the rapid cooling of a metallic workpiece in water, oil, or air to obtain certain desired material properties. It is the most critical step in the sequence of heat-treating operations to preserve the solid solution formed at the solution heat-treating temperature by rapidly cooling to near room temperature. Because of the complex interaction between temperature, phase-transformation, and stress/strain relation that depends on the temperature distribution and the microstructure of the workpiece, there is no performance-informed quenching process that can be applied reliably to reduce the high scrap rate of airframe aluminum forging parts with a significant amount of residual stress and distortion. Since large aluminum forging parts are increasingly used in aerospace structures to enable structural unitization, it is important to construct a digital twin modeling approach to mirror the physical quenching process for minimizing scrap rate, increasing production efficiency, and engineers and machine operators' handling of variances in forging operations. A high-fidelity modeling of the coupling of thermal, metallurgical, and mechanical interactions is a key component to creating a digital twin of the physical quenching process. A high-fidelity thermal multi-phase computational fluid dynamics (CFD) model is applied to simulate fluid dynamics and temperature fields in the quenchant tank. The developed immersogeometric modeling approach is used next for an efficient model generation of a 3D workpiece with various dipping orientations. Given the temperature and pressure profiles predicted from the CFD-based heat transfer module, residual stress and distortion prediction modules are developed by including temperature and pressure fields mapping and temperature and strain rate dependent property evolution via Abaqus' user-defined subroutines. Verification and demonstration studies are performed using aluminum coupons dipped into a quenching tank with different orientations. Time histories of the temperature and residual stress fields were predicted to explore the relationship between the process and performance.

A multi-stage precipitation model is formulated to predict the microstructural evolution and explain the high performance of additive friction stir deposited aluminum alloy 7050 (AA 7050) for hole repair. The first stage is the heating process due to the high-temperature thermomechanical process of the stir. In this process, small eta precipitates dissolve as they lose their stability with increasing temperature, and this causes the volume fraction of eta precipitates to decrease and the concentration of Mg and Zn in the matrix to increase. The second stage is the cooling process at the end of the repair where material feeding ends and the tool is lifted away. Heterogeneous nucleation of eta precipitates may occur and as the temperature cools below 250 degrees C, Guinier-Preston (GP) zones start to form. The final stage is the natural aging process, where the eta ' precipitate starts to grow. The volume fraction and precipitate radius are predicted for each type of precipitate. Furthermore, the fine eta ' precipitates and GP zones with a decent volume fraction improve the material strength and fatigue life. (c) 2024 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license

Temperature uniformity inside the autoclave and in the manufactured thermoset composite part is the key to enhancing curing performance. The present study incorporated experimental setup, multiphysics and computational fluid dynamics (CFD) models to provide insight into the gas flow pattern and temperature distribution inside the autoclave besides temperature and curing evolution in composite parts during the manufacturing process. The ultimate goal is to help improve the uniformity of the degree of curing and produce robust composite parts. The used numerical models were validated by experiment and the predicted results were in agreement with experimental efforts with a difference of 5%. A maximum temperature difference of 3 K was observed for panel 1, 1 K for panel 2, and 4 K for panel 3 during the curing stages. The variation was more pronounced during the post-curing stage, especially at the center. Correspondingly, less degree of curing (DOC) occurred at the center. The DOC variation between the center and the sides at the base of the panel is 2.5%. Variation in DOC across the thickness of the composite panels was also observed due to thermal spikes and uneven heat transfer. The DOC variation between upper surface and midplane is about 11%. Regions of high turbulence intensity lead to better heat transfer while regions of low speed such as confined regions under the tooling table had the poorest heat transfer. The study provided some practical implications to reduce the non-uniformity in temperature distribution and improve part curing.

This paper presents a high-fidelity fatigue damage modeling approach for composite structures with ply drops by including the local stress ratio calculation, an R-ratio dependent fatigue damage accumulation, a constant load diagram for the determination of an S-N curve at an arbitrary R-ratio, and a cycle jumping technique to account for the change of fatigue damage accumulation rate resulting from the stiffness redistribution. A combined experimental and numerical study of tapered composite beams was performed using the mixed axial tension and vertical bending. A reliable test fixture was fabricated to track the motion at the loading end that was used as the boundary conditions for the fatigue life prediction model. To assist in the design of the peak load for the fatigue test, static failure prediction was performed first using a representative tapered beam subjected to a constant axial tension along with monotonic bending. Fatigue tests under constant axial tension and cyclic bending at two different peak load levels were conducted and the test data and failure patterns were compared with the model predictions.

Large language models (LLMs) bear promise as a fast and accurate material modeling paradigm for evaluation, analysis, and design. Their vast number of trainable parameters necessitates a wealth of data to achieve accuracy and mitigate overfitting. However, experimental measurements are often limited and costly to obtain in sufficient quantities for finetuning. To this end, we present a physics-based training pipeline that tackles the pathology of data scarcity. The core enabler is a physics-based modeling framework that generates a multitude of synthetic data to align the LLM to a physically consistent initial state before finetuning. Our framework features a two-phase training strategy: (1) utilizing the large-in-amount while less accurate synthetic data for supervised pretraining, and (2) finetuning the phase-1 model with limited experimental data. We empirically demonstrate that supervised pretraining is vital to obtaining accurate finetuned LLMs, via the lens of learning polymer flammability metrics where cone calorimeter data is sparse.

A digital twin (DT), with the components of a physics-based model, a data-driven model, and a machine learning (ML) enabled efficient surrogate, behaves as a virtual twin of the real-world physical process. In terms of Laser Powder Bed Fusion (L-PBF) based additive manufacturing (AM), a DT can predict the current and future states of the melt pool and the resulting defects corresponding to the input laser parameters, evolve itself by assimilating in-situ sensor data, and optimize the laser parameters to mitigate defect formation. In this paper, we present a deep neural operator enabled computational framework of the DT for closed-loop feedback control of the L-PBF process. This is accomplished by building a high-fidelity computational model to accurately represent the melt pool states, an efficient surrogate model to approximate the melt pool solution field, followed by an physics-based procedure to extract information from the computed melt pool simulation that can further be correlated to the defect quantities of interest (e.g., surface roughness). In particular, we leverage the data generated from the high-fidelity physics-based model and train a series of Fourier neural operator (FNO) based ML models to effectively learn the relation between the input laser parameters and the corresponding full temperature field of the melt pool. Subsequently, a set of physics-informed variables such as the melt pool dimensions and the peak temperature can be extracted to compute the resulting defects. An optimization algorithm is then exercised to control laser input and minimize defects. On the other hand, the constructed DT can also evolve with the physical twin via offline finetuning and online material calibration. Finally, a probabilistic framework is adopted for uncertainty quantification. The developed DT is envisioned to guide the AM process and facilitate high-quality manufacturing.

The Autoclave processing is commonly used in manufacturing high-performance fibre-reinforced thermoset composite components in the aerospace industry. Variations in the cure cycle, sometimes even apparently minor deviations from the prescribed cure cycle, can harm the laminate properties. Given the costly and time-consuming autoclave manufacturing process, there is a strong need to cure the maximum number of parts in the shortest possible time without compromising quality. In order to achieve high-rate automated manufacturing with the optimized autoclave process, it is important to construct a digital twin modelling approach to mirror the physical composite curing process in the virtual domain based on the integration of high-fidelity multi-physics models. The resulting digital twin includes a thermal CFD model, a thermo-chemo-mechanical module, and an efficient and accurate block coupling between these two modules. The customized Abaqus driven by local and spatial variation of the turbulence-induced heat transfer coefficient (HTC) imposed through one-way coupling determines the thermo-mechanical response in composite parts. Using the developed digital twin tool (SMARTCLAVE), HTC's spatial and temporal variation can be generated digitally without invoking an expensive and time-consuming experimental approach. The predicted local boundary conditions are used in SMARTCLAVE to determine the cure kinetics, temperature distribution, and thermal-mechanical response that drives the residual stress and distortion of composite parts after curing. The accuracy of the digital twin for autoclaving is demonstrated first using a benchmark problem followed by the capability demonstration with a single-part L-beam assembly. The benefits of using the digital twin tool are illustrated via the optimal placement of multiple parts in an autoclave to balance the throughput and quality.

Many aerospace applications involve complex multiphysics in compressible flow regimes that are challenging to model and analyze. Fluid–structure interaction (FSI) simulations offer a promising approach to effectively examine these complex systems. In this work, a fully coupled FSI formulation for compressible flows is summarized. The formulation is developed based on an augmented Lagrangian approach and is capable of handling problems that involve nonmatching fluid–structure interface discretizations. The fluid is modeled with a stabilized finite element method for the Navier–Stokes equations of compressible flows and is coupled to the structure formulated using isogeometric Kirchhoff–Love shells. To solve the fully coupled system, a block-iterative approach is used. To demonstrate the framework’s effectiveness for modeling industrial-scale applications, the FSI methodology is applied to the NASA Common Research Model (CRM) aircraft to study buffeting phenomena by performing an aircraft pitching simulation based on a prescribed time-dependent angle of attack.

The damage to fastener holes in aerospace aluminum structures presents significant challenges for aircraft durability, and conventional bushing methods for repairing oversized holes often fall short due to the lack of metallurgical bonding and limited edge distance availability. This study investigates additive friction stir deposition, a non-melting additive process, as a viable alternative for the structural repair of aerospace fastener holes. The repair process, demonstrated on AA7050 (Al-Zn-Mg-Cu-Zr) hole structures, involves filling oversized holes with new material and machining to restore the original hole size. The repaired hole coupons are defect-free and exhibit good fatigue performance under fully reversed tension-compression loading (R = -1). At a nominal stress amplitude of 123.5 MPa, the average number of cycles to failure is 12,666 for unrepaired baseline coupons and 17,372 for effectively repaired coupons. Restoring complex geometries without compromising fatigue performance has been difficult in aerospace applications; this study marks the first demonstration of additive repair that consistently outperforms the unrepaired baseline coupons. Notably, the result is achieved through a low-energy, cost-effective solution without the need for post-repair heat treatment. Except for a few outliers, the post-repair fatigue performance generally remains inferior to that of undamaged, pristine coupons, likely due to precipitate evolution in AA7050 caused by the thermomechanical processing nature of additive friction stir deposition. This evolution weakens the repair region and the adjacent base material, leading to faster crack initiation and growth compared to the properly aged base material, AA7050-T7451.