This study investigates the effects of the restitution coefficient on the coupled vortex-induced vibration (VIV) and wall impact dynamics of a near-wall cylinder with a small gap ratio of 0.1. A two-dimensional fluid–structure interaction (FSI) model is established at Re = 150 for an elastically mounted cylinder undergoing pure transverse motion. Employing an overset grid scheme and a velocity reversal impact model, the structural vibration, impact loads, hydrodynamic forces, and wake evolution are systematically analyzed over a reduced velocity range of 2 ≤ Ur ≤ 14 under three restitution coefficients: b = 0, 0.5, and 1.0. The wall confinement substantially suppresses vortex development on the gap side, leading to two distinct wake regimes: a no-shedding (NS) mode for Ur = 2–5 and a single-sided shedding (SS) mode for Ur = 6–14. The NS–SS transition triggers a sharp increase in the root mean square vibration amplitude at Ur = 6 and forms a well-defined lock-in regime for Ur = 6–12. Continuous cylinder-wall collision occurs within the lock-in range, whereas collision is absent at low reduced velocities (Ur = 2–5) due to insufficient vibration amplitude and at high reduced velocities (Ur = 13–14) owing to the elevated time-averaged equilibrium position and attenuated vibration response. The mean collision force reaches its maximum at Ur = 10. The wall impact process is characterized by gap fluid compression, instantaneous velocity reversal, and inertia-dominated pressure and lift pulses, followed by post-collision high-frequency velocity oscillations and hysteresis. An increase in the restitution coefficient strengthens impact intensity, amplifies high-frequency dynamic responses, and prolongs the hysteresis stage, while exerting a negligible influence on the wake topology and low-frequency VIV characteristics. The studied FSI system presents a typical dual-timescale coupled behavior, where the global dynamic response is dominated by single-sided vortex excitation and weakly modulated by collision-induced nonlinearities.
With the increasing scale of wind farms, wake effects have become increasingly important for turbine safety and power generation efficiency. Accurate wake modeling for yaw control and layout optimization requires low computational cost, high prediction accuracy, and robust generalization across operating conditions. In this study, a physics-informed domain-adversarial neural network (PI-DANN) is proposed for wind turbine wake prediction. The incompressible Navier–Stokes equations and the vorticity equation are incorporated into the training objective to enforce physical consistency and improve the representation of wake flow structures. Meanwhile, a domain classifier coupled with a gradient reversal layer is introduced to extract condition-invariant flow features and enhance cross-condition generalization. The model is validated against light detection and ranging measurements and computational fluid dynamics simulations under multiple operating conditions. Compared with a fully connected neural network and a domain-adversarial neural network, PI-DANN reduces the prediction error by 15.42% and 2.71%, respectively. This work provides a practical wake modeling framework that supports intelligent control and digital operation of wind farms.
Nanofluids are often employed in cavity-based systems, which are common in industrial and electronic applications where efficient convective heat transfer is needed. Among these configurations, crescent-shaped cavities introduce complex flow behavior that can influence thermal performance. Therefore, the present study investigated the flow of a MoS2–SiC hybrid nanofluid inside a crescent-shaped cavity under the influence of a magnetic field. The cavity contains triangular heat sources, and hot and cold walls are used to control the temperature. MoS2 acts as an exothermic element to dissipate heat within the outer layer of the wall, whereas silicon carbide acts as an endothermic element to conduct heat around triangular structures. These nanomaterials combine to provide efficient heat transfer. A mathematical model is developed using momentum, energy, and concentration equations and solved via the finite element method through COMSOL Multiphysics. The effects of the Reynolds number (Re), Hartmann number (Ha), and Peclet number (Pe) on the velocity, temperature, and concentration are also studied. The results show that higher Re strengthens convective flow, leading to stronger vortex structures and enhancing heat transfer. In contrast, a larger value of Ha reduces the fluid velocity and consequently decreases thermal transport. Additionally, the quantitative results include the average Nusselt number (Nuavg), Sherwood number (Shavg) and kinetic energy. Their results revealed that both Nuavg and Shavg increased with increasing Re, resulting in increased heat and mass transfer. However, Pe has a small effect on Nu but significantly increases Sh, as it enhances the concentration gradient. The novelty lies in combining MoS2–SiC hybrid nanofluid with a crescent-shaped cavity and triangular heat-generating elements. The results demonstrate that cavity curvature and internal heat-source geometry can substantially modify the vortex structure and transport characteristics, which are useful for thermal-management applications.
A computational investigation of the hemodynamic mechanisms underlying saccular aneurysm (SA) initiation at the circle of Willis is presented. The study incorporates 3-dimensional patient-specific and idealized geometries, laminar flow, non-Newtonian blood rheology, fully coupled fluid–structure interaction, and physiologically realistic Windkessel-based boundary conditions. The developed methodology is validated against available experimental and analytical data. By integrating findings from previous studies, the work aims to provide a generalized assessment of aneurysm initiation and the associated hemodynamics. The significance of various parameters is evaluated by analyzing distinct SA at the anterior communicating and middle cerebral arteries. Vascular geometric abnormalities, such as absent or hypoplastic arterial segments, produced anatomical asymmetry and directed high mass flow rates toward the aneurysm ostium, leading to localized flow impingement and widespread wall shear stress regions. Both cases demonstrated elevated oscillatory shear index, large circumferential wall shear stress gradients, and persistently low aneurysm formation indicator, suggesting that these parameters function as supportive hemodynamic markers of aneurysm initiation. In contrast, vorticity and relative residence time exhibited case-dependent behavior. Overall, the findings suggest that a combination of vascular abnormalities and characteristic hemodynamic disturbances can be used to identify regions susceptible to aneurysm initiation.
Particle-based direct numerical simulation (DNS) is a promising approach to study microscale turbulence–cloud–aerosol interactions. Such DNS models require external forcing to sustain turbulent fluctuations in the momentum and thermodynamic fields. Although momentum forcing is common in particle-based DNS studies of atmospheric clouds, thermodynamic forcing has not gained enough attention. This study implements external forcing in the thermodynamic fields of temperature and vapor mixing ratio in addition to momentum forcing. The forcing terms are constructed in the Fourier space using a band of low wavenumbers. The turbulent flow of air is assumed to be homogeneous and isotropic. Aerosol particles and cloud droplets are tracked in Lagrangian fashion, with processes of condensation, evaporation, activation, and deactivation considered together. The impacts of thermodynamic forcing on entrainment of ambient air into the cloud and subsequent turbulent mixing are examined. It is found that thermodynamic forcing sustains fluctuations in the thermodynamic fields and reduces the mean temperature, water vapor mixing ratio, and supersaturation. Although thermodynamic forcing is applied at large scales only, it influences the energy at all scales by interacting with latent heat change. The probability distributions of forced thermodynamic fields are negatively skewed with kurtosis values larger than three. Further analysis of microphysical properties shows that thermodynamic forcing enhances evaporation and deactivation with broadening of particle size distributions. The activation and deactivation regimes are fluctuation-dominated if thermodynamic forcing is present as opposed to being mean-dominated only when thermodynamic forcing is not applied.
