Automatic fiber placement (AFP) is an advanced composite manufacturing technology with precise fiber steering and rapid in-situ forming capability. However, out-of-plane wrinkling caused by geometric incompatibility during curved-surface placement remains a major limitation. This study proposes an integrated optimization framework combining geometric contact mechanics, high-fidelity finite element simulation, and deep learning to investigate wrinkle suppression under curved-surface conditions. A finite element model incorporating the nonlinear shear behavior of thermoplastic prepregs and interfacial viscous contact was developed. A MeshGraphNets-based graph neural network was further established as a physics-aware surrogate model for the placement process. The results indicate that the roller yaw attitude reshapes the stress state in wrinkle-prone regions by regulating shear deformation and stress paths, while compaction force enhances interlaminar constraint and consolidation. Their synergistic effect suppresses fiber buckling and interrupts wrinkle formation. The surrogate model achieved a mean prediction error of 3.95% and improved computational efficiency by 1224 times, providing a rapid analysis method for wrinkle suppression in curved-surface AFP. This method broadens the wrinkle-free processing window and improves placement quality.

The time-/temperature-dependent viscoelasticity behaviors in polymer composites are due to the memory decay effects, inducing various interesting phenomena, including the shape memory effects, the morphology evolution in living of organisms, temperature-dependent fracture behaviors, and so forth. To model those behaviors, various constitutive models based convolutional integrals have been developed for capturing viscoelasticity. However, the more refined model introduced a greater number of parameters and made the parameter identification more intricate, hardening their applications. To solve this problem, we developed a viscoelastic constitutive artificial neural network (VCANN) for automated parameter identification. Following the recently developed physics-informed neural operator, the activation functions of the network architecture were chosen to incorporate the physical knowledge in the standard viscoelastic model within the continuum mechanics framework. This VCANN is equivalent to the viscoelastic model, mapping the input variables (time, temperature, and deformation histories) to the stress responses, and the network node weights are equivalent to the model parameters. Therefore, the parameter identification problems are equivalently transformed into the training problem of the VCANN. To validate the parameter identification by this VCANN, we used the finite element (FE) simulation method to generate datasets of the mechanical responses in polymer composites under different strain states (uniaxial tension and equibiaxial tension) and different loading history (monotonic stretching, relaxation, and step loading). Six training scenarios indicate that as the dataset incorporates more diverse material deformation features, the parameters identified by VCANN become more precise. The datasets should include both uniaxial and biaxial tensile data to achieve the decoupled identification of bulk modulus and shear modulus. Overall, this developed method could reduce the application threshold of the viscoelastic constitutive model by establishing a bridge between the results of experiments and the input of parameters in FE simulations.

High-fidelity Finite Element Analysis (FEA) is essential for predicting defects in composite textile forming, yet its prohibitive computational cost creates a fundamental bottleneck for iterative design. While deep learning surrogates like CNNs offer acceleration, they are inherently limited by fixed-grid representations and depend on massive datasets that are resource-intensive to generate. To address this, we propose a novel Graph Neural Network (GNN) simulator that learns the incremental, step-by-step evolution of the forming process rather than an end-to-end mapping. By encoding the textile’s woven connectivity into a graph topology, our model achieves robust predictive capability using an extremely scarce training dataset of only tens simple geometric primitives. Evaluations demonstrate that the proposed GNN accelerates simulations by over 600 times compared to FEA while accurately predicting global shear distribution ([[EQUATION]]). Critically, the model successfully generalizes to new scales and prepreg aspect ratios where traditional surrogates would fail. Furthermore, we validate the feasibility of a few-shot transfer learning strategy, enabling rapid deployment to entirely new, complex part families. With only a handful of additional samples, the base model is efficiently specialized for unseen geometry regimes, validating a practical “pre-train on primitives, adapt to products” workflow for composite manufacturing design loops.

The viscoelastic characteristics and complex multiphase microstructures of highly filled polymers render their fracture behaviors time- and scale-dependent. As a result, the prediction of their behaviors and the analysis of their multiscale fracture mechanisms are formidable engineering challenges. In this study, a multiscale fracture model for highly filled polymers was established by coupling a viscoelastic phase field model in the form of the generalized Maxwell framework with the cohesive zone model. Numerical implementation was achieved through a dual-layer mesh structure associated with cohesive elements. The simulation results showed favorable agreement with the experimental results obtained for notched and center-holed propellant specimens. The maximum load prediction error of the developed model remained within 7.0%, while the prediction error for the displacement corresponding to the maximum load remained within 7.5%. At the mesoscale, the fracture process of solid propellants was divided into three distinct stages: particle dewetting (Stage I), crack initiation (Stage II), and matrix tearing (Stage III). Dewetting of relatively large ammonium perchlorate particles induced stress concentration, thereby triggering crack initiation and propagation (perpendicular to the loading direction under uniaxial loading, and along the +/- 45 degrees directions under biaxial loading). Smaller high melting explosive particles caused the crack to propagate along a curvilinear path. During the relaxation stage, viscoelastic hysteresis of the polymer induced the enlargement of dewetting-induced voids, consequently leading to stress reduction and crack propagation. The model established in this study provides a methodological reference for other highly filled polymeric materials.

Induction welding of thermoplastic composites offers substantial application potential, yet uncertainty in the heat generation mechanisms at the welding interface severely hinders uniform temperature field control. This study presents a quantitative methodology to decouple the heat generation mechanisms in CF/PEEK laminates with varying ply angles. Four stacking configurations ([0/30]6, [0/45]6, [0/60]6, and [0/90]6) were characterized for their electrical and thermal properties. A microscale heat generation model was established to distinguish the contributions of fiber heating, contact resistance heating, and dielectric hysteresis. Results reveal that junction heating exceeds fiber heating by 106-107 times. The ratio of dielectric hysteresis to contact resistance heating increases monotonically from 2.0512 to 2.8241 as the layup angle increases from 30° to 90°, with dielectric hysteresis dominating (67.2%-73.9%). This trend is attributed to the significant reduction in interlaminar contact resistivity with increasing angle, which diminishes contact resistance heating efficiency and relatively enhances the dominant role of dielectric hysteresis. An angle-dependent compensation coefficient method is proposed to enable quantitative conversion of heat generation efficiency across different layups. Furthermore, a transient three-dimensional finite element model incorporating layup angle is developed and validated, overcoming the conventional limitation to orthogonal layups. This work provides a robust theoretical and simulation framework for process optimization and temperature homogenization in induction welding of thermoplastic composites.

The ignition characteristics of amorphous boron (B) particles in carbon dioxide (CO2) were experimentally investigated using a reflected shock tube over a wide temperature range from 2100 K to 3400 K. The shock tube was equipped with one spectrometer and two monochromators, enabling a comprehensive exploration of the influence of particle size, temperature, pressure and concentration of CO2 on the ignition delay time (tign). Notably, it was found that above 2850 K, tign became more dependent on the temperature, which might be attributed to the pyrolysis of CO2. Meanwhile, tign became more sensitive to the pressure with the decrease of temperature. Specifically, the temporal emission spectra of BO2 from B and pure B2O3 were detected at 2600 K in both CO2 diluted with argon (Ar) and pure Ar atmospheres. Different from the B oxidation in the pure Ar atmosphere, the normalized BO2 spectral signals and their first-order derivatives indicated two ignition stages of B particles in CO2. That is, the first ignition happens in the oxide layer, and the subsequent reaction is on the boron surface. These two stages involved three main reactions: (1) pyrolysis of B2O3 to BO2, (2) reaction of B2O3 and CO2, and (3) reaction of B and CO2. Finally, a theoretical analysis based on valence electron configurations and chemical bonding was performed to explain the bilateral diffusion in stage (1) at the atomic scale.

Wrinkle defects frequently arise during automated fiber placement of thermoplastic composites due to inadequate process design, with prepreg buckling instability as a defining feature. In this paper, a defect prediction framework integrating multiple deformation mechanisms was constructed to simulate wrinkle formation during automated fiber placement. A hypoelastic constitutive model was developed to capture the mechanical response of unidirectional thermoplastic prepreg. The formation mechanisms of wrinkles under varying temperatures and steering radii were investigated, and the predictive capability of the model was validated experimentally. Wrinkle defects were attributed to mechanical instabilities driven by the interplay of deformation mechanisms, evolving through a dynamic process of shear compensation followed by buckling instability. Experimental and simulation results demonstrated that, as the temperature increased, in-plane shear deformation became the dominant mechanism, dissipating part of the compressive strain energy and thereby mitigating axial buckling. Wrinkle formation was fundamentally driven by axial compression induced by placement trajectory curvature. With reduced steering radii, bending progressively became the prevailing mechanism, while interlaminar bonding was insufficient to restrict fiber motion or accommodate it through shear. The localized accumulation of compressive strain energy subsequently triggered prepreg buckling, resulting in the formation of wrinkle defects.

Solid propellants are particle-filled polymeric materials exhibiting nonlinear viscoelastic properties. In this study, viscoelastic compression tests are conducted on solid propellants. The compression nominal stress-strain curves exhibit a J shaped. Relaxation time and viscous stress increase with increasing deformation. The Tension-compression asymmetry in viscoelastic behavior is analyzed. Compared with tension, compression induces higher stress, longer relaxation time, and a larger viscous part. The mechanisms of nonlinear relaxation and Tension-compression asymmetry are analyzed through free volume theory and mesoscale simulations. At the microscopic scale, the limited free volume hinders the rearrangement of molecular networks and chain segments under large deformations. Less free volume and more coiled chain segments under compression lead to higher stress and longer relaxation time. At the mesoscopic scale, the patterns of interface debonding and damage evolution differ under tension and compression. The strength disparity between interfaces and particles leads to distinct tension and compression modulus. The micro-mesoscale coupling mechanisms result in Tension-compression asymmetry in the macroscopic viscoelastic mechanical behavior. The methodology and findings provide insights for multiscale investigations of other particle-filled composites.

This study addresses the unclear thermo-mechanical aging mechanisms in novel glycidyl azide polymer (GAP)-based propellant by conducting thermal aging and thermo-mechanical coupled aging experiments, studying the impacts of thermo-mechanical interactions on the storage aging performance from the perspectives of macroscopic mechanical properties, mesoscopic morphology, and microscopic structures. The experimental results demonstrate that thermal aging mechanisms primarily involve binder matrix degradation and plasticizer decomposition. Since the pre-strain level does not exceed the debonding strain, no significant differences in storage performance are observed between thermal aging and thermo-strain coupled aging. However, the thermo-stress coupled aging significantly affects the mechanical properties. In addition to thermal aging mechanisms, creep effects induce permanent deformation and surface area expansion. Mesoscopic simulations are then conducted, demonstrating that an increased permanent deformation reduces interfacial strength, while a surface area expansion accelerates plasticizer volatilization, leading to the hardening of the binder matrix. The combined effect of these factors yields a significant decline in elongation at break.

Thermal protection materials (TPMs) serve as a critical barrier ensuring the safe flight of spacecraft. Achieving simultaneous optimization of load-bearing capacity and ablation resistance in TPMs while maintaining lightweight characteristics poses a significant challenge. To address this, the present study proposes an interface engineering strategy that constructs an interface-ceramizable aramid nanofiber (ANF) preform reinforced phenolic aerogel composites. The ANF not only provides essential mechanical support but also endows the material with structural stability under high-temperature ablative environments. Benefiting from the design of the interface layer, the ANF and phenolic aerogel are tightly bonded; during the pyrolysis process, their linear shrinkage rates are highly compatible, effectively suppressing the formation of microcracks. Meanwhile, the surface functional layer can be in-situ sintered into high-temperature-stable phases (e.g., SiC, ZrC), significantly improving ablation resistance. The resulting composites exhibit a density as low as 0.327 g/cm3 and a compressive modulus as high as 285.7 MPa, allowing flexible adaptation to various application requirements. Notably, after exposure to a butane flame for 120 s, the linear and mass ablation rates reach 0.0113 mm/s and 0.00134 g/s, respectively, with the back temperature only 100.9 °C. This strategy offers a fundamental solution and a scalable fabrication route for next-generation high-performance thermal protection systems for spacecraft.

The curing process of nitrate ester plasticized polyether (NEPE) propellants involves complex interactions between heat generation, heat transfer, and mechanical evolution, which fundamentally determine the structural integrity of solid rocket motors. In particular, the exothermic nature of the curing reaction leads to non–uniform temperature fields, giving rise to heterogeneous curing behavior and residual stress accumulation. In this study, a fully coupled thermo–chemo–viscoelastic finite–element model is developed to simulate the curing–cooling process and validated by experiments. Simulations were then performed to quantify the effects of temperature–pressure coupling on residual stress evolution during the curing process. A four–factor Box–Behnken design combined with Response Surface Methodology (RSM) is employed to evaluate the influence of curing temperature, curing pressure, cooling rate, and depressurization rate. Analysis of variance shows that curing temperature and pressure dominate stress formation, exhibiting a pronounced synergistic amplification, whereas rate–controlled factors play a secondary role. Optimization yields an optimal parameter combination—50.0 °C, 3.33 MPa, 0.064 °C/min, and 0.007 MPa/min—reducing maximum residual stress to 0.018 MPa, a 53.6% decrease compared with atmospheric curing. The findings provide a quantitative framework for optimizing pressurized curing and improving structural reliability in polymer–based energetic materials.

Viscoelastic materials such as polymer-modified asphalt, rubberized concrete, and structural adhesives are widely used in civil engineering for damping and deformation-recovery properties. However, their time-dependent and heterogeneous features hinder long-term performance prediction and structural integrity assessment, necessitating an efficient, accurate multiscale analysis. This study proposes a multiscale finite element method (MsFEM) for viscoelastic materials, integrating the generalized Maxwell model (GMM) and the initial stress method. It converts time-domain convolution constitutive relations into incremental elastic problems, retains a constant stiffness matrix to boost efficiency, and uses mesoscale heterogeneity-representing basis functions for coarse-fine mesh coupling. Validated via axial rod and cantilever beam examples, the method shows high accuracy against analytical solutions, serving as a robust tool for predicting material long-term performance with applications in structural health monitoring, life cycle assessment, and sustainable design.

Inverse design of solid propellant grains seeks geometries that reproduce prescribed internal-ballistic pressure–time profiles to enable rapid design iteration under complex performance requirements. Most surrogate-based methods rely on fixed-topology templates, limiting their ability to represent diverse burnback evolutions, while free-form topology optimization often lacks engineering interpretability and compatibility with template-based workflows. This study proposes a unified multi-topology inverse-design framework based on geometric-element-level parameterization within a shared design space. Grain geometries are represented by activatable geometric elements, enabling controlled topology switching while preserving compatibility with conventional parametric representations and engineering interpretability. A physics-consistent multi-topology dataset is constructed through a coupled forward-analysis pipeline integrating level-set burnback simulation with a zero-dimensional internal-ballistic model. A learning-based inverse strategy performs topology identification and topology-conditioned geometric parameter inversion directly from pressure–time inputs. Numerical results demonstrate reliable recovery of topology and key geometric parameters across multiple topologies and burnback scenarios. For the three representative pressure–time targets, the relative errors in total impulse are 0.26%, 1.89%, and 1.73%, with cosine similarities of 0.9853, 0.9938, and 0.9932, respectively. Uncertainty and noise analyses indicate stable topology identification, with variability concentrated near ignition peaks and stage transitions. The proposed framework provides an engineering-compatible pathway for multi-topology inverse grain design.

A conservative UGKP deposition-resuspension boundary scheme is proposed for high-Stokes-number alumina particle/droplet impingement in aluminized solid rocket motors. It preserves particle-phase conservation and alleviates the near-wall statistical bias caused by traditional trap boundaries. A mean-free-path-based wall-resolution requirement is identified for grid-independent near-wall deposition-resuspension statistics, and an algebraic wall-adjacent subcell is introduced to recover these statistics on coarser meshes. Experimental cross-validation in a converging two-phase-flow facility reveals the macroscopic deposition-reflection transition mechanism. Numerical results show that near-wall collisional relaxation reduces incident-velocity statistics and suppresses wall-mass-flux fluctuations, leading to a statistically distributed deposition criterion. Discrete deposition increases the effective wall roughness, thereby elevating the critical transition Sommerfeld number under macroscopic nonequilibrium impingement to 20-25. An equivalent contact heat-transfer coefficient conditioned on the deposition state is calibrated against measured heat flux, providing a semi-empirical closure for heat transfer under nonequilibrium particle scouring.

This study presents a numerical investigation of mode transition and combustion characteristics in a dual-mode scramjet engine, employing a compressible flamelet/progress variable (CFPV) model coupled with a self-adaptive turbulence eddy simulation (SATES) approach. The flamelet tables were generated for a methane/ethylene/air mixture using the Gas Research Institute 3.0 chemical mechanism. A mixture-fraction-dependent lookup table was incorporated to rescale the source term of progress variable, thereby accounting for the pressure effects. Steady Reynolds-averaged Navier–Stokes simulations against experimental measurement from a strut and a cavity model combustor confirm that the present CFPV model yields more accurate predictions of wall pressure, flame topology, and heat release distribution than the conventional uncorrected flamelet/progress variable model. Detailed SATES simulations were performed at flight Mach numbers of 5.8 and 6.5 to resolve the inherent mode transition characteristics. The transition process from the ramjet to scramjet mode was well captured as the inflow Mach number and overall equivalence ratio increase, which was accompanied by a shift in the flame stabilization mechanism from a jet-wake stabilized to a cavity-assisted jet-wake flame mode. Further statistical analyses distinguish variations in turbulent combustion regimes and premixed/diffusion flame proportions under different operating conditions, and transient shock-induced auto-ignition events inside the jet-wake region are reasonably reproduced by the proposed numerical approach.

Nitrous oxide (N2O) and ethanol (C2H5OH) are green propellants that have great potential in propulsion systems. This study reports the experimental measurements of ignition delay times (IDTs) for N2O/C2H5OH mixtures utilizing a shock tube. The influences of the pressure, equivalence ratio, and temperature on IDTs are evaluated within the pressure of 2.5 to 7.5 atm, equivalence ratio of 0.5 to 2.0 and temperature of 1300 to 1914 K. The measured IDTs exhibit an Arrhenius-type temperature dependence and decrease with elevated pressures. Conversely, transitioning from fuel-lean to fuel-rich conditions significantly prolongs the IDTs, with the equivalence ratio exerting a more pronounced influence than pressure. Furthermore, these IDTs are substantially longer than those observed in conventional O2 atmospheres. Several established kinetic models are evaluated, culminating in the development and validation of a revised Shrestha mechanism that demonstrates substantially improved predictive accuracy. Subsequent sensitivity analyses reveal that the global reactivity is mainly driven by the highly endothermic unimolecular decomposition N2O (+ M) = N2 + O (+ M) and the reaction N2O + H = N2 + OH, while ethanol decomposition acts as principal ignition inhibitor. Rate of production and reaction pathway analyses further demonstrate that the drastically extended ignition delay stems from the absence of highly efficient chain-branching steps. Under fuel-rich conditions, intermediate fragments aggressively scavenge the limited radical pool, forcing a thermally-driven induction period. Furthermore, integrated reaction flux analysis elucidates the kinetic origins of high-pressure discrepancies, revealing that elevated pressures competitively dissipate the active radical flux. This work provides essential experimental data and a validated chemical kinetic model to support further engineering applications of N2O/C2H5OH mixtures.

Induction welding of carbon fiber reinforced thermoplastic (CFRTP) composites has become one of the most promising welding techniques, attributed to its efficiency, adaptability, and noncontact advantages. However, the uneven temperature field within the welding region severely restricts the weld strength. In this paper, a focused induction welding method (FIWM) was proposed, devoted to controlling magnetic field energy and regulating the temperature field. A three-dimensional electro-magnetic-thermal Multiphysics model was developed to analyze temperature distribution, and the model was validated by induction welding experiments. To guide the magnetic field distribution and reduce the temperature gradient, a cylindrical permanent magnet is introduced into FIWM as a magnetic flux concentrator (MFC). The surrogate models for equilibrium temperature, temperature gradient, and effective area were established to rapidly optimize multiple process parameters. Thus, the optimal parameters were found out by the particle swarm optimization (PSO) method. The results show that the temperature gradient decreased from 123.89 degrees C to 49.68 degrees C, and the effective area increased from 41.0 to1900.3 mm2. Finally, the strength of the welded joint was assessed by single lap shear tests, and the result showed that the mechanical properties could reach 36.09 +/- 1.44 MPa. Therefore, FIWM is an efficient approach to regulate the temperature field and improve the mechanical performance in CFRTP induction welding.