To save content items to your account,
please confirm that you agree to abide by our usage policies.
If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your account.
Find out more about saving content to .
To save content items to your Kindle, first ensure no-reply@cambridge.org
is added to your Approved Personal Document E-mail List under your Personal Document Settings
on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part
of your Kindle email address below.
Find out more about saving to your Kindle.
Note you can select to save to either the @free.kindle.com or @kindle.com variations.
‘@free.kindle.com’ emails are free but can only be saved to your device when it is connected to wi-fi.
‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.
The accuracy, robustness and affordability of localisation are fundamental to autonomous robotic inspection within aircraft maintenance, repair and overhaul (MRO) hangars. Hangars typically have high ceilings and are predominantly steel-framed structures with metal cladding. Because of this, they are regarded as GPS-denied environments, characterised by significant multipath effects and strict operational constraints, which together form a unique challenging setting. The lack of comparative techno-economic benchmarks for localisation technologies in such environments remains a critical gap. Addressing this, the paper presents the first techno-economic analysis that benchmarks motion capture (MoCap), ultra-wideband (UWB) and a ceiling-mounted camera (CMC) system across three operational scenarios: robot localisation, asset monitoring and surface defect detection within a single-bay hangar. A two-stage optimisation framework for camera selection and placement is introduced, which couples market-based camera-lens selection with an optimisation solver, producing camera layouts that minimise hardware while meeting accuracy and coverage targets. The consolidated blueprints provide quantification of the required equipment and its performance: 15 global-shutter GigE cameras are adequate for drone localisation, 9 cameras meet the requirements for on-bay monitoring and 49 high-resolution cameras facilitate defect mapping of the upper airframe surfaces for midsize defects. Across these scenarios, the study reports indicative performance and cost envelopes: a MoCap installation delivers submillimeter localisation at an estimated £190k per bay, UWB delivers centimetre-level tracking for around £49k and the proposed CMC system layouts achieve task-specific coverage with costs in the £9k–£77k range. The analysis equips MRO planners with an actionable method to balance accuracy, coverage and budget, demonstrating that an optimised CMC system can deliver robust and cost-effective sensing for next-generation smart hangars.
Strategies for optimising air-fuel interaction are critical in supersonic combustion. This research alters the fuel injector design by adjusting the strut corner base angle, allowing the fuel to contact the air transversely. This computational analysis uses the Reynolds-averaged Navier-Stokes (RANS) equations in conjunction with the Shear Stress Transport (SST) k-omega turbulence model and the eddy dissipation turbulence chemistry model. The validation has been conducted for the present simulation with the experimental data, comparing the pressure, temperature and Schlieren images. The standard DLR scramjet combustor model consists of a single strut (fuel injector) injecting parallel to the air stream, but in this research, the design of the strut base is changed to angles 30, 45 and 60 degrees to inject the fuel in a new method. This slanted strut base aids fuel injection into the airstream and permits the mixture to generate swirls behind the strut base, resulting in better mixing and 35% greater turbulence. This modification improves the reaction process’s spontaneity and generates 37% higher temperatures, increasing mixing and combustion efficiency by about 37% and 23%, respectively.
Transition onset of high-speed boundary layers can move first downstream and then upstream with increasing nose-tip bluntness, which is called transition reversal. For the first time, our recent research reproduced the experimentally observed transition reversal by direct numerical simulation (DNS, Guo et al., J. Fluid Mech. vol. 1005, 2025, A5). As a continuation study, this work explores the effect of the form of free-stream disturbances, as the transition in the large-bluntness regime still remains poorly understood. The free-stream Mach number is 5 and the nose-tip radius 3 mm of the blunt plate exceeds the experimental reversal value. Three-dimensional broadband perturbation is carefully constructed through superimposition of planar fundamental waves in the free stream, which initiates the transition in DNS. For each Fourier component, the same perturbation strength is applied for slow/fast acoustic, vortical and entropic waves. All the cases present a ‘streak-turbulent spot’ two-stage transition scenario due to non-modal instabilities. The transition onset locations induced by entropic and slow/fast acoustic waves are close and significantly ahead of that by vortical waves. More evident impact of the disturbance form is manifested in the length of the transitional region, which is the shortest for entropic waves and the longest for vortical waves. Regarding the effect of the angle of incidence that mimics the tunnel environment, it alters the post-shock acoustic-wave structure and reduces the length of the transitional region. In the streaky stage, the form of free-stream disturbances changes the pronounced spanwise wavelengths on the blunt nose and the plate, where the two regions also differ from each other. In the turbulent-spot region, the shortest transitional region induced by the entropic wave is attributed to its largest mean spanwise spreading rate of the turbulent spot. From the perspective of energy budget, shear-induced dissipation dominates the heat transfer escalation in the transitional region. Overall, with significant leading-edge bluntness, the flight environment may tend to result in delayed transition onset compared with the tunnel counterpart.
Electrical effects are known to play an important role in particle-laden flows, yet a holistic view of how they modulate turbulence remains elusive due to the complexity of multifield coupling. Here, we present a total of 119 direct numerical simulations of particle-laden turbulent channel flow that reveal a striking ability of electrical effects to induce turbulence relaminarisation and markedly alter wall drag. As expected, the transition from turbulence to laminar flow is accompanied by abrupt changes in the statistical properties of both the fluid and particulate phases. Nevertheless, with increasing electrical effects, the wall-normal profiles of the mean streamwise fluid velocity and mean local particle mass loading exhibit opposite trends in the turbulent and laminar regimes, arising from the competition between turbophoresis and electrostatic drift. We identify three distinct flow regimes resulting from the electrical effects: a drag-reduced turbulent regime, a drag-reduced laminar regime, and a drag-enhanced laminar regime. It is revealed that relaminarization originates from the complete suppression of the streak breakdown in the near-wall self-sustaining cycle, followed by the sequential inhibition of other subprocesses in the cycle. In the turbulent regime, increasing electrical effects induce opposing trends in Reynolds and particle stress contributions to drag, yielding a non-monotonic drag response. In laminar regimes, by contrast, the drag coefficient increases monotonically as the Reynolds stress vanishes and particle-induced stress becomes dominant.
We investigate the effect of inertial particles on Rayleigh-Bénard convection using weakly nonlinear stability analysis. An Euler–Euler/two-fluid formulation is used to describe the flow instabilities in particle-laden Rayleigh–Bénard convection. The weakly nonlinear results are presented near the critical point (bifurcation point) for water droplets in the dry air system. We show that supercritical bifurcation is the only type of bifurcation beyond the critical point in particle-laden Rayleigh–Bénard convection. Interaction of settling particles with the flow and the Reynolds stress or distortion terms emerges due to the nonlinear self-interaction of fundamental modes breaking down the top–bottom symmetry of the secondary flow structures. In addition to the distortion functions, the nonlinear interaction of fundamental modes generates higher harmonics, leading to the tendency of preferential concentration of uniformly distributed particles, which is completely absent in the linear stability analysis. Further, we show that in the presence of thermal energy coupling between the fluid and particles, the difference between the horizontally averaged heat flux at the hot and cold surfaces is equal to the net sensible heat flux advected by the particles. The difference between the heat fluxes at hot and cold surfaces increases with an increase in particle concentration.
Three-dimensional laminar flow over an inclined spinning disk is investigated at a Reynolds number of ${\textit{Re}} = 500$ and an angle of attack of $\alpha = 25^\circ$, for tip-speed ratios up to 3. Numerical simulations are performed to investigate the effect of spin on the aerodynamics and characterise the instabilities that occur. Increasing tip-speed ratio significantly increases both lift and drag monotonically. Several distinct wake regimes are observed, including vortex shedding in the non-spinning case, vortex-shedding suppression at moderate tip-speed ratios and a distinct corkscrew-like short-wavelength instability in the advancing tip vortex at higher tip-speed ratios. Vorticity generated by the spinning disk strengthens the advancing tip vortex, inducing a spanwise stretching in the trailing-edge vortex sheet. This helps to dissipate the vorticity, which in turn prevents roll up and suppresses vortex shedding. The short-wavelength instability shows qualitative and quantitative matches to the $(-2,0,1)$ principal mode of the elliptic instabilities seen in pairs of counter-rotating Batchelor vortices. The addition of vorticity from the disk rotation significantly alters the circulation and axial velocity in the tip vortices, giving rise to elliptic instability despite its absence in the non-spinning case. In select cases, lock-in between the frequency of the elliptic instability and twice the spin frequency is observed, indicating that disk rotation acts as an additional forcing for the elliptic instability. Additional simulations at different Reynolds numbers and angle of attacks are considered to examine the robustness of observed phenomena across different parameter combinations.
The Monin–Obukhov similarity theory (MOST) is a cornerstone of atmospheric science for describing turbulence in stable boundary layers. Extending MOST to stably stratified turbulent channel flows, however, is non-trivial due to confinement by solid walls. In this study, we investigate the applicability of MOST in closed channels and identify where and to what extent the theory remains valid. A key finding is that the ratio of the half-channel height to the Obukhov length serves as a governing parameter for identifying distinct flow regions and determining their corresponding mean velocity scaling. Hence, we propose a relation to estimate this ratio directly from the governing input parameters: the friction Reynolds and friction Richardson numbers ($\textit{Re}_{\tau }$ and $Ri_{\tau }$). The framework is tested against a series of direct numerical simulations across a range of $\textit{Re}_{\tau }$ and $Ri_{\tau }$. The reconstructed velocity profiles enable accurate prediction of the skin-friction coefficient crucial for quantifying pressure losses in stratified flows in engineering applications.
In this paper, we numerically investigate the orbit dynamics of three-dimensional symmetric Janus drops in shear flow using an improved ternary-fluids phase field method, focusing on how drop deformation and initial orientation affect the orbit drift of two configurations of Janus drops: dumbbell-shaped and near-spherical. We find that the motion of dumbbell-shaped drops eventually evolves into tumbling, while near-spherical drops attain stable spinning. We attribute this bifurcation in orbit drift to contrasting deformation dynamics and shape-dependent hydrodynamics of the two configurations. Specifically, the drift bifurcation is closely related to the aspect ratio of Janus drops at equilibrium, giving rise to two distinct mechanisms: (1) coupling between outer interface deformation and the surrounding flow field; and (2) interplay between inner interface deformation and vortices enclosed within the drop. In addition, we observe that for the dumbbell-shaped Janus drops with different aspect ratios, their tumbling dynamics resembles ellipsoids in shear flow. Moreover, the trajectories of the dumbbell-shaped Janus drops during orbit drift collapse onto a universal curve, independent of their initial orientations, and significant deformation and inertia accelerate the orbit transition. To quantitatively evaluate the effect of drop deformation on the orbit drift of the dumbbell-shaped Janus drops, we propose an effective aspect ratio model based on the drop shapes at equilibrium and at the maximum elongation. By incorporating the effective aspect ratio into Jeffery’s theory for solid particles, we accurately predict the rotation period and angular velocity of Janus drops in the tumbling regime and during the orbit drift, especially for drops with linear deformation. Moreover, the orbit parameter $C$ is found to vary exponentially with time for drops with linear deformation, while the time variation of $C$ transits from one exponential function to another for drops with nonlinear deformation.
The thermal interactions of liquid droplets impacting a moving substrate are investigated, combining theoretical modelling with experimental validation. An analytical model is developed to predict the time-evolving contact temperature and heat flux at the droplet–substrate interface. Accounting for the convective heat transport induced by the impacting drop, the model incorporates a finite thermal contact resistance, which is a critical parameter that was often neglected in earlier studies for drop impact. High-speed, spatially resolved infrared thermography is used to record the two-dimensional, transient temperature evolution at the droplet–substrate interface during drop impact on a rotating disc. Measured temperature maps are used for numerical simulations to reconstruct local interfacial heat fluxes. The model is validated for different droplet diameters, substrate velocities and thermal conditions. The findings demonstrate that the substrate velocity and droplet diameter have negligible influence on the thermal behaviour within the tested parameter space.
The deposition of droplets onto a swollen polymer network induces the formation of a wetting ridge at the contact line. Current models typically consider either viscoelastic effects or poroelastic effects, while polymeric gels often exhibit both properties. In this study, we investigate the growth of the wetting ridge using a comprehensive large-deformation theory that integrates both dissipative mechanisms – viscoelasticity and poroelasticity. In the purely poroelastic case, following an initial instantaneous incompressible deformation, the growth dynamics exhibits scale-free behaviour, independent of the elastocapillary length or system size. A boundary layer of solvent imbibition between the solid surface (in contact with the reservoir) and the region of minimal chemical potential is created. At later times, the ridge equilibrates on the diffusion time scale given by the elastocapillary length. When viscoelastic properties are incorporated, our findings show that, during the early stages (prior to the viscoelastic relaxation time scale), viscoelastic effects dominate the growth dynamics of the ridge and solvent transport is significantly suppressed. Beyond the relaxation time, the late-time dynamics closely resembles that of the purely poroelastic case. These findings are discussed in light of recent experiments, showing how our approach offers a new interpretation framework for wetting of polymer networks of increasing complexity.
The two-dimensional (2-D) evolution of perturbed long weakly nonlinear surface plane, ring and hybrid waves, consisting, to leading order, of a part of a ring and two tangent plane waves, is modelled numerically within the scope of the 2-D Boussinesq–Peregrine system. Numerical runs are initiated and interpreted using the reduced 2-D cylindrical Korteweg–de Vries (cKdV)-type and Kadomtsev–Petviashvili II (KPII) equations. The cKdV-type equation leads to two different models, the KdV$\theta$, where $\theta$ stands for a polar angle, and cKdV equations, depending on whether we use the general or singular (i.e. the envelope of the general) solution of the associated nonlinear first-order differential equation. The KdV$\theta$ equation is also derived directly from the 2-D Boussinesq–Peregrine system and used to analytically describe the intermediate 2-D asymptotics of line solitons subject to sufficiently long transverse perturbations of finite strength, while the cKdV equation is used to initiate outward- and inward-propagating ring waves with localised and periodic perturbations. Both of these equations, together with the KPII equation, are used to model the evolution of hybrid waves, where we show, in particular, that large localised waves (lumps) can appear as transient (emerging and then disappearing) states in the evolution of inward-propagating waves, contributing to the possible mechanisms for the generation of rogue waves. Detailed comparisons are made between the key features of the non-stationary 2-D modelling and relevant predictions of the reduced equations.
In this work, we study the reaction-controlled dual bubbles ripening on a heterogeneous substrate with high surface wettability hysteresis, where the bubbles evolve with constant contact radius but varied contact angle. We first theoretically derived the governing kinetic equation of bubble curvature radius $R_B$, based on which we surprisingly found three possible ripening processes under six different conditions, i.e. the classical Ostwald ripening (the bubble with the larger curvature radius $R_B$ exhibits an increase in $R_B$, while the bubble with the smaller curvature radius $R_B$ experiences a decrease in $R_B$), the reversed ripening (converse to Ostwald ripening), and the consistent ripening ($R_B$ of both bubbles increases or reduces consistently). Further analyses from the aspects of chemical potential and free energy lead to an interesting finding that the $R_B$ of two bubbles finally reach egalitarianism, independently of different ripening processes. Numerical results obtained from two-phase lattice Boltzmann modelling demonstrate excellent agreement with theoretical predictions, specifically concerning the kinetic equation, the various ripening processes, and the egalitarianism of bubble radii $R_B$ after ripening completion.
This study investigates the dynamics of free-surface turbulence (FST) using direct numerical simulations (DNS). We focus on the energy exchange between the deformed free-surface and underlying turbulence, examining the influence of high Reynolds (${\textit{Re}}$) and Weber (${\textit{We}}$) numbers at low to moderate Froude (${\textit{Fr}}$) numbers. The two-fluid DNS of FST at the simulated conditions is able to incorporate air entrainment effects in a statistical steady state. Results reveal that a high ${\textit{We}}$ number primarily affects entrained bubble shapes (sphericity), while ${\textit{Fr}}$ significantly alters free-surface deformation, two-dimensional compressibility and turbulent kinetic energy (TKE) modulation. Vortical structures are mainly oriented parallel to the interface. At lower ${\textit{Fr}}$, kinetic energy is redistributed between horizontal and vertical components, aligning with rapid distortion theory, whereas higher ${\textit{Fr}}$ preserves isotropy near the surface. Evidence of a reverse or dual energy cascade is verified through third-order structure functions, with upscale transfer near the integral length scale, and enhanced vertical kinetic energy in upwelling eddies. Phase-based discrete wavelet transforms of TKE show weaker decay at the smallest scales near the interface, suggesting contributions from gravitational energy conversion and reduced dissipation. The wavelet energy spectra also exhibits different scaling laws across the wavenumber range, with a $-3$ slope within the inertial subrange. These findings highlight scale- and proximity-dependent effects on two-phase TKE transport, with implications for subgrid modelling.
In this experimental study, we investigate, for the first time, the structure and evolution of the near wake of a circular cylinder in a flowing soap film at the onset of vortex shedding. The study primarily focuses on the changes occurring within the recirculation bubble, along with the evolution of vortex shedding. A significantly large recirculation bubble forms behind the cylinder in the soap film environment, characterized by small-scale vortices along its edges, an observation that starkly contrasts with its three-dimensional counterparts. These small-scale vortices driven by the Kelvin–Helmholtz instability, further induce a transverse deflection of the recirculation bubble, leading to an intermittent generation of the wake vortices. The instantaneous velocity field in the wake is examined, highlighting the clear evidence of intermittency in vortex formation. The frequency and wavelength of the chain of small-scale vortices on the recirculation bubble is evaluated, and a functional relationship with the flow Reynolds number is determined. We believe this observation to be novel, potentially revealing a new pathway for understanding the two-dimensional transition in bluff-body wakes.
Slip effects on solid boundaries are common in complex fluids. Boundary depletion layers in polymer solutions can create apparent slip effects, which can in turn significantly impact the dynamics of moving bodies. Motivated by microswimmer locomotion in such environments, we derive a series of slip slender-body theories for filamentous bodies experiencing slip-like boundary conditions. Using Navier’s slip model, we derive three slip slender-body theories, linking the body’s velocity to the distribution of hydrodynamic forces. The models are shown to be consistent with each other and with existing numerical computations. As the slip length increases, we show that the drag parallel to the body decreases towards zero while the perpendicular drag remains finite, in a manner which we quantify. This reduction in drag ratio is shown to be inversely related to microswimmer mobility in two simple swimmer models. This increase could help rationalise empirically observed enhanced swimming in complex fluids.
Regression and classification are closely related, as shown in this chapter, which discusses methods used to map a linear regression function into a probablity function by either logistic function (for binary classification) or softmax function (for multi-class classification). According to this probablity function, an unlabeled sample can be assigned to one of the classes. The optimal model parameters in this method can be obtained based on the training set so that either the likelihood or the posterior probability of these parameters are maximized.
This paper investigates the flow past a flexible splitter plate attached to the rear of a fixed circular cylinder at low Reynolds number 150. A systematic exploration of the plate length ($L/D$), flexibility coefficient ($S^{*}$) and mass ratio ($m^{*}$) reveals new laws and phenomena. The large-amplitude vibration of the structure is attributed to a resonance phenomenon induced by fluid–structure interaction. The modal decomposition indicates that resonance arises from the coupling between the first and second structural modes, where the excitation of the second structural mode plays a critical role. Due to the combined effects of added mass and periodic stiffness variations, the two modes become synchronised, oscillating at the same frequency while maintaining fixed phase difference $\pi /2$. This further results in the resonant frequency being locked at half of the second natural frequency, which is approximately three times the first natural frequency. A reduction in plate length and an increase in mass ratio are both associated with a narrower resonant locking range, while a higher mass ratio also shifts this range towards lower frequencies. A symmetry-breaking bifurcation is observed for cases with $L/D\leqslant 3.5$, whereas for $L/D=4.0$, the flow remains in a steady state with a stationary splitter plate prior to the onset of resonance. For cases with a short flexible plate and a high mass ratio, the shortened resonance interval causes the plate to return to the symmetry-breaking stage after resonance, gradually approaching an equilibrium position determined by the flow field characteristics at high flexibility coefficients.
This chapter offers a comprehensive overview of large language models (LLMs), examining their theoretical foundations, core mechanisms, and broad-ranging implications. We begin by situating LLMs within the domain of natural language processing (NLP), tracing the evolution of language modeling from early statistical approaches to modern deep learning methods.</p>The focus then shifts to the transformative impact of the Transformer architecture, introduced in the seminal paper Attention Is All You Need. By leveraging self-attention and parallel computation, Transformers have enabled unprecedented scalability and efficiency in training large models.</p>We explore the pivotal role of transfer learning in NLP, emphasizing how pretraining on large text corpora followed by task-specific fine-tuning allows LLMs to generalize across a wide range of linguistic tasks. The chapter also discusses reinforcement learning with human feedback (RLHF)—a crucial technique for refining model outputs to better align with human preferences and values.</p>Key theoretical developments are introduced, including scaling laws, which describe how model performance improves predictably with increased data, parameters, and compute resources, and emergence, the surprising appearance of complex behaviors in sufficiently large models.</p>Beyond technical aspects, the chapter engages with deeper conceptual questions: Do LLMs genuinely "understand" language? Could advanced AI systems one day exhibit a form of consciousness, however rudimentary or speculative? These discussions draw from perspectives in cognitive science, philosophy of mind, and AI safety.</p>Finally, we explore future directions in the field, including the application of Transformer architectures beyond NLP, and the development of generative methods that extend beyond Transformer-based models, signaling a dynamic and rapidly evolving landscape in artificial intelligence.
This chapter is concerned with the constrained optimization problem which plays an important role in ML as many ML algorithms are essentially to maximize or minimize a given objective funcion with either equality or inequality constraints. Such kind of constrained optimization problems can be reformulated in terms of the Lagrangian function including an extra tern for the constraints weighted by their Lagrange multipliers as well as the original function. The chapter also consider the important duality principle based on which the constrained optimization problem can be addressed as either the primal (original) problem, or the dual problem, which is equivalent to the primal if a set of KKT conditions are satisfied, in the sense that the solution of the dual is the same as that for the primal. The chapter further considers two methods, linear and quadratic programming, of which the latter is the foundation for support vector machine (SVM), an important classification algorithm to be considered in a later chapter.