Unstable Density Stratification Research Articles

Linear dynamics of a thick liquid layer subjected to an oblique temperature gradient

The present study aims to demonstrate the application of the oblique temperature gradient (OTG) to manipulate the buoyancy instabilities and resulting mixing. To accomplish this, we consider a model consisting of a liquid layer supported by a perfectly conducting bottom wall from below and the upper surface is exposed to ambient inert gas in a shallow container. The stability analysis employs the pseudospectral method and perturbation energy budget approach. The imposed OTG consists of a negative vertical temperature gradient (VTG), which leads to unstable density stratification, and a negative horizontal temperature gradient (HTG) component responsible for the Hadley circulation, which can be utilised to control instabilities. Without the HTG, a Rayleigh–Bénard mode of instability exists due to the imposed VTG termed as a VTG mode. The imposed HTG induces a linear VTG term and a quintic polynomial VTG term in the bottom wall-normal coordinate $y$ . The induced linear VTG term reinforces the imposed VTG, while the induced quintic polynomial VTG term opposes the imposed VTG. For the vanishing Biot number ( $Bi$ ), the induced quintic VTG term stabilises the VTG mode for the Prandtl number, $Pr=7$ . However, for $Pr=0.01$ , the Reynolds stresses associated with the Hadley circulation destabilise the VTG mode. For non-zero $Bi$ , new streamwise and spanwise instability modes arise for $Pr=7$ due to the induced linear VTG term. Physical arguments show that the unstable density stratification near the gas–liquid interface caused by the synergistic effect of the imposed VTG and the induced linear VTG term lead to the existence of the predicted new modes. The present study thus shows that the strength of the HTG can be utilised to control the instability modes and critical parameters, thereby demonstrating the utility of the OTG in manipulating the buoyancy instabilities.

Концентрационная конвекция в замкнутой области пористой среды при заданном вертикальном перепаде концентрации и учете иммобилизации примеси

This work is devoted to the study of stability of horizontal filtration flow of a mixture through a closed porous domain taking into account impurity immobilization. The instability arises due to the vertical concentration difference of heavy impurities, which creates unstable density stratification. A general mathematical model describing the transport of an impurity through a porous medium is presented. The equations are simplified for the case of low impurity concentration. Simplification made it possible to analytically obtain the solution corresponding to homogeneous horizontal filtration and to study its stability. It is known that, in the narrow regions of porous medium and at weak intensities of the external flow, convection is excited in a monotonous manner. On the contrary, in the case of an infinite horizontal layer, oscillatory instability is observed. A study of the transition between the instability modes is presented. It is shown that the oscillatory regime is observed in long regions or at significant intensity of the external horizontal flow. At low flow intensities, convective cells do not move relative to the region and, hence, there is no reason for oscillations. It has been established that the range of flow intensity values, in which oscillations are observed, grows with increasing length of the domain. Impurity immobilization leads to the stabilization of horizontal filtration with respect to convective perturbations. Critical curves and stability maps are obtained in a wide range of problem parameters and then analyzed. For limiting cases, a comparison is made with the known results obtained for an infinite layer.

Immiscible Rayleigh–Taylor turbulence: Implications for bacterial degradation in oil spills

An unstable density stratification between two fluids mixes spontaneously under the effect of gravity, a phenomenon known as Rayleigh-Taylor (RT) turbulence. If the two fluids are immiscible, for example, oil and water, surface tension prevents intermixing at the molecular level. However, turbulence fragments one fluid into the other, generating an emulsion in which the typical droplet size decreases over time as a result of the competition between the rising kinetic energy and the surface energy density. Even though the first phenomenological theory describing this emulsification process was derived many years ago, it has remained elusive to experimental verification, hampering our ability to predict the fate of oil in applications such as deep-water spills. Here, we provide the first experimental and numerical verification of the immiscible RT turbulence theory, unveiling a unique turbulent state that originates at the oil-water interface due to the interaction of multiple capillary waves. We show that a single, non-dimensional, and time-independent parameter controls the range of validity of the theory. Our findings have wide-ranging implications for the understanding of the mixing of immiscible fluids. This includes in particular oil spills, where our work enables the prediction of the oil-water interface dynamics that ultimately determine the rate of oil biodegradation by marine bacteria.

Natural convection effects in insulation layers of spherical cryogenic storage tanks

Natural convection effects in the insulation layers of spherical storage tanks are studied for different cryogenic fluids and tank sizes or curvatures. The non-vacuum-based insulation layers of these tanks are made of gas-filled porous materials such as perlite or glass bubbles. The large temperature difference that can exist between the inner cold wall and outer hot wall may create convective patterns with a multi-cellular flow within the insulation-filled annulus and lead to a higher boil-off rate and cryo-pumping effects. A main objective in the design of these systems is the elimination of the convective patterns by a proper choice of insulation properties. This work examines the impact of the insulation permeability or the Rayleigh number (Ra) on the type of convective flows for liquified natural gas (LNG) and liquid hydrogen (LH2) spherical storage tanks. Bifurcation diagrams of possible solutions are generated by numerical solutions of the governing equations at different Ra numbers. It is found that as Ra increases, new convective cells emerge near the south pole of the insulation layer due to unstable density stratification and lead to the distortion of isotherms, bringing the cold boundary towards the external hot boundary. When the temperature difference is large as in the case of LNG and LH2 tanks, there exist regions of multiple solutions in which multi-cell convective solutions may co-exist with a crescent-shaped weakly convective solution. In order to eliminate the possibility of multi-cell convection, inversion of the cold boundary, and reduce the boil-off rate, the Ra value of the insulation system should be below the limit point of the convective branch.

On the Potential for Cumulate Mantle Overturn in Mercury

AbstractMercury has a compositionally diverse surface exhibiting geochemical terranes that represent different periods of igneous activity, suggesting diverse mantle source compositions. Mercury's juvenile mantle likely formed after fractional solidification of a magma ocean, which produced distinct mineralogical horizons with depth. To produce the diversity of observed volcanic terranes, dynamic mixing of materials from distinct mantle horizons is required. One process that could dynamically mix the juvenile cumulate pile is cumulate mantle overturn, where dense layers in shallow planetary mantles sink into deeper, less dense layers as Rayleigh‐Taylor instabilities. Gravitationally unstable density stratification is a requisite starting condition for overturn; solidification of the Mercurian magma ocean is likely to have produced such a density inversion, with a relatively dense clinopyroxene‐bearing pyroxenite layer atop lower density dunite and harzburgite layers. Sulfides are present in abundance on Mercury's surface and would be additional mantle phase(s) if they are indigenous to the planet's interior. Sulfides have variable densities; they could potentially enhance the formation of gravitational instabilities or prevent them from developing. Exploring physically reasonable mantle density and viscosity structures, we evaluate the potential for cumulate mantle overturn in Mercury and predict the possible timing, scale, and rate of overturn for plausible physical parameter combinations. Our analysis suggests that overturn is possible in Mercury's mantle within 100 Myr of magma ocean solidification, providing a mechanism for producing the mantle sources that would melt to form surface compositions on Mercury, and overturn may control the spatial scale of volcanic provinces observed on the surface today.

Non-Oberbeck-Boussinesq effects and sub-critical primary bifurcations in porous media convection

Large temperature variations and highly unstable density stratification may exist between the cold and hot boundaries of the porous non-vacuum gas filled insulation layers of cryogenic storage tanks. For example, the density can vary by about a factor 3 in N2 filled liquefied natural gas (LNG) tank insulation layers and factor 15 in H2 filled liquid hydrogen (LH2) tank insulation layers. In such cases, the bifurcation of the convective states from the conduction state could be sub-critical leading to sudden jumps and hysteresis effects in heat transfer and boil-off rates. This work examines the impact of large physical property variation and non-Oberbeck-Boussinesq (non-OB) effects on the nature of bifurcation of convective states in gas filled porous insulation layers. Specifically, we show that the well-known super-critical (pitchfork) bifurcation of convective states that appear under the OB approximation changes to sub-critical bifurcation when the temperature difference exceeds a critical value. Further, under such large temperature variations, the emerging convective solutions could be highly asymmetric with center of circulation much closer to the cold boundary. We present linear stability results, local bifurcation analysis as well as numerical simulations of the full model equations for the two selected LNG and LH2 cases.

Пассивная конвективная вентиляция в двойном воздушно-пористом слое с внутренним тепловыделением, зависящим от доли твердой фазы

The convective stability of a two-layered system consisting of a heat-generating porous region underlying an air region has been numerically studied. The linear dependence of the heat release on the solid volume fraction is taken into account in the porous region. The equal constant temperature values are fixed on the external impermeable boundaries of the system. The critical internal Rayleigh-Darcy number at which the convection is induced in the system in the form of two-dimensional roll patterns with a given wave number has been determined. The convective flow is possible due to the formation of unstable density stratification in the presence of internal heat release. Two types of stationary convection, namely, the local and the large-scale convection, have been studied. The local flow arises in the air sublayer and scarcely penetrates into the porous sublayer. The large-scale convection covers both sublayers. The change in the convective regime occurs with the growth of one or another parameter of the system and indicates the variation of the instability type. It is accompanied by an abrupt (by times and tens of times) change in the critical wave number of roll patterns. Numerical calculations show a decrease in the onset value for both types of convection with increasing solid volume fraction ϕ in the porous sublayer and increasing relative thickness d of the air sublayer. The growth of the Darcy number (the dimensionless per-meability of the porous sublayer) also causes destabilization of the air motionless state at the given ϕ and d. The variation of the convection regime from a large-scale flow to a local one occurs with increasing relative thickness of the air sublayer, whereas an opposite transition from the local to the large-scale convection regime is observed with increasing Darcy number.

Mechanism for the subglacial formation of cryogenic brines

AbstractCryogenic brines are under-studied, despite the fact that they may contain information about past ice-sheet behavior. Cryogenic brines form through cryoconcentration of seawater, although the specific setting and mechanism of formation have been debated. Previous conceptual models of brine formation require seawater isolation from the ocean in a closed basin experiencing freezing. We propose instead that they may form in pore spaces of marine sediments subjected to repeat cycles of ice-sheet advance and retreat. During periods of basal freezing, cryoconcentration produces hypersaline brines which experience downward flow driven by unstable density stratification. Our advection-diffusion model of porewater chemistry evolution successfully recreates the porewater chemistry of two deep Antarctic cores containing cryogenic brines (AND-1B and AND-2A), suggesting that cryogenic brines can be formed through the repeated isolation and cryoconcentration of marine waters within subglacial sediment pore spaces of modern and past ice sheets.

Genesis of Convolutions in Lacustrine Complexes in Regions with Comparatively Low (Baltic Shield) and High (Tien Shan) Paleoseismic Activity

Abstract —Structural and lithological comparison of intraformational plicative liquefaction deformations (convolutions) in Late Quaternary lacustrine complexes of the Baltic Shield (sediments of small glacial lakes) and the Tien Shan (sediments of a large stationary basin in the Issyk-Kul depression) is carried out in order to clarify their genesis. These regions have sharply different levels of seismotectonic activity, which makes it possible to most fully determine the role of a seismic factor in the formation of bedding disturbances. Convolutions in the Baltic Shield are represented by load casts, flame structures, pseudonodules, and regular folds with more pronounced anticlinal bends. It is shown that convolutions occur here only in the most liquefied soils (siltstones and fine-grained sands) with low cohesion and fluid permeability in the case of unstable (increased density and coarsening of the sediment from bottom to top) or neutral (homogeneous composition) stratification of weakly consolidated sediments, and the ratio of the widths of the synclinal and anticlinal parts of the folds (Ksyn) for these structures is 1.0–7.5. These signs point to the formation of disturbances because of a spontaneous instability or an instability initiated by weak mechanical influences in the sedimentary stratum under conditions of a normal vertical gradient of the sediment viscosity. Deformations in the second region are structures typical of the Baltic Shield, columnar structures of deep interpenetration of layers, diapirs, and clastic dikes. There are also flexural folds with Ksyn ≈ 1, which formed at a higher degree of sediment consolidation than the actual liquefaction structures. The convolutions of the Issyk-Kul depression are developed not only in sand–silty sediments but also in relatively lowly liquefied soils (clay, gravel, and pebbles). These structures formed under both unstable and stable density stratification of the sediment (for example, clay-on-sand), and Ksyn = 0.3–2.5. Thus, the formation of convolute structures in the Issyk-Kul depression is impossible without strong dynamic impacts on stratified sediments during their accumulation. The results obtained make it possible to substantiate the predominantly diagenetic (convective and landslide) genesis of convolutions in the lacustrine complexes of the Baltic Shield and the seismogenic genesis in the limnogenic deposits of the Tien Shan. A number of new criteria for identifying seismites are proposed, which determines the practical significance of the study.

Effects of buoyancy on turbulent scalar flux closure for turbulent premixed flames in the context of Reynolds Averaged Navier–Stokes simulations

The effects of body force on the statistical behaviour of turbulent scalar flux and its closure in the context of Reynolds Averaged Navier–Stokes simulations have been studied using Direct Numerical Simulations (DNS) of statistically planar turbulent premixed flames under different turbulence intensities and Froude numbers. An increase in body force magnitude in the case of unstable density stratification has been found to augment flame wrinkling, burning rate and gradient transport in comparison to a case without body force but with statistically similar unburned gas turbulence. By contrast, an increase in body force magnitude in the case of stable stratification reduces the flame wrinkling, burning rate and gradient transport in comparison to the flame without body force subjected to statistically similar unburned gas turbulence. Based on a-priori DNS analysis, an algebraic closure for turbulent scalar flux has been identified where the Froude number effects are explicitly accounted for. The body force has been found to have significant influence on the statistical behaviours and magnitudes of the terms of the scalar flux transport equation and this effect is particularly strong for the mean pressure gradient term in the scalar flux transport equation. Based on a detailed a priori DNS analysis, suitable model expressions have been identified for the turbulent transport, pressure gradient, dissipation and reaction rate-velocity correlation terms of the scalar flux transport equation by incorporating the effects of body force (e.g. Froude number effects) for improved model performance.

Onset of convective instability in an inclined porous medium

The diffusion of a solute from a concentrated source into a horizontal, stationary, fluid-saturated porous medium can lead to a convective motion when a gravitationally unstable density stratification evolves. In an inclined porous medium, the convective flow becomes intricate as it originates from a combination of diffusion and lateral flow, which is dominant near the source of the solute. Here, we investigate the role of inclination on the onset of convective instability by linear stability analyses of Darcy's law and mass conservation for the flow and the concentration field. We find that the onset time increases with the angle of inclination (θ) until it reaches a cutoff angle beyond which the system remains stable. The cutoff angle increases with the Rayleigh number, Ra. The evolving wavenumber at the onset exhibits a lateral velocity that depends non-monotonically on θ and linearly on Ra. Instabilities are observed in gravitationally stable configurations (θ≥90°) solely due to the nonuniform base flow generating a velocity shear commonly associated with Kelvin–Helmholtz instability. These results quantify the role of medium tilt on convective instabilities, which is of great importance to geological CO2 sequestration.

Effects of Body Forces on the Statistics of Flame Surface Density and Its Evolution in Statistically Planar Turbulent Premixed Flames

Body forces such as buoyancy and externally imposed pressure gradients are expected to have a strong influence on turbulent premixed combustion due to the considerable changes in density between the unburned and fully burned gases. The present work utilises Direct Numerical Simulation data of three-dimensional statistically planar turbulent premixed flames to study the influence of body forces on the statistical behaviour of the flame surface density (FSD) and its evolution within the flame brush. The analysis has been carried out for different turbulence intensities and normalised body force values (i.e., Froude numbers). A positive value of the body force signifies an unstable density stratification (i.e., body force is directed from the heavier unburned gas to the lighter burned gas) and vice versa. It is found that for a given set of turbulence parameters, flame wrinkling increases with an increase in body force magnitude in the unstable configuration. Furthermore, higher magnitudes of body force in the unstable density stratification configuration promote a gradient type transport of turbulent scalar and FSD fluxes, and this tendency weakens in the stable density stratification configuration where a counter-gradient type transport is promoted. The statistical behaviours of the different terms in the FSD transport equation and their closures in the context of Reynolds Averaged Navier–Stokes simulations have been analysed in detail. It has been demonstrated that the effects of body force on the FSD and the terms of its transport equation weakens with increasing turbulence intensity as a result of the diminishing relative strength of body force in comparison to the inertial force. The predictions of the existing models have been assessed with respect to the corresponding terms extracted from the explicitly averaged DNS data, and based on this evaluation, suitable modifications have been made to the existing models to incorporate the effects of body force (or Froude number).

Asymptotic modelling and direct numerical simulations of multilayer pressure-driven flows

Abstract The nonlinear dynamics of two immiscible superposed viscous fluid layers in a channel is examined using asymptotic modelling and direct numerical simulations (DNS). The flow is driven by an imposed axial pressure gradient. Working on the assumption that one of the layers is thin, a weakly-nonlinear evolution equation for the interfacial shape is derived that couples the dynamics in the two layers via a nonlocal integral term whose kernel is determined by solving the linearised Navier–Stokes equations in the thicker fluid. The model equation incorporates salient physical effects including inertia, gravity, and surface tension, and allows for comparison with DNS at finite Reynolds numbers. Direct comparison of travelling-wave solutions obtained from the model equation and from DNS show good agreement for both stably and unstably stratified flows. Both the model and the DNS indicate regions in parameter space where unimodal, bimodal and trimodal waves co-exist. Nevertheless, the asymptotic model cannot capture the dynamics for a sufficiently strong unstable density stratification when interfacial break-up and eventual dripping occurs. In this case, complicated interfacial dynamics arise from the dominance of the gravitational force over the shear force due to the underlying flow, and this is investigated in detail using DNS.

Investigation of the influence of wind stress in autumn leading to unstable stratification in a meromictic lake using three-dimensional numerical modeling

Currents in lakes affect the distribution of hydrophysical and hydrobiological characteristics. Three-dimensional model study is an important step toward an improved understanding and reliable prediction of the distribution of hydrophysical fields. Salt stratified lake Shira refers to meromictic reservoirs. The depth of the upper mixed layer depends on many factors (even a few isolated, fairly rare facts of its complete mixing are known). The study of the influence of each of them separately is the task of mathematical modeling. The model results indicate that strong wind stress in the autumn can result in unstable density stratification, leading to intense vertical circulation.

Mixing driven by transient buoyancy flows. I. Kinematics

Mixing of two miscible liquids juxtaposed inside a cavity initially separated by a divider, whose buoyancy-driven motion is initiated via impulsive perturbation of divider motion that can generate the Richtmyer-Meshkov instability, is investigated experimentally. The measured Lagrangian history of interface motion that contains the continuum mechanics of mixing shows self-similar nearly Gaussian length stretch distribution for a wide range of control parameters encompassing an approximate Hele-Shaw cell to a three-dimensional cavity. Because of the initial configuration of the interface which is parallel to the gravitational field, we show that at critical initial potential energy mixing occurs through the stretching of the interface, which shows frontogenesis, and folding, owing to an overturning motion that results in unstable density stratification and produces an ideal condition for the growth of the single wavelength Rayleigh-Taylor instability. The initial perturbation of the interface and flow field generates the Kelvin-Helmholtz instability and causes kinks at the interface, which grow into deep fingers during overturning motion and unfold into local whorl structures that merge and self-organize into the Rayleigh-Taylor morphology (RTM) structure. For a range of parametric space that yields two-dimensional flows, the unfolding of the instability through a supercritical bifurcation yields an asymmetric pairwise structure exhibiting smooth RTM that transitions to RTM fronts with fractal structures that contain small length scales for increasing Peclet numbers. The late stage of the RTM structure unfolds into an internal breakwave that breaks down through wall and internal collision and sets up the condition for self-induced sloshing that decays exponentially as the two fluids become stably stratified with a diffusive region indicating local molecular diffusion.

Solutal natural convection flows in ternary mixtures

It is known that the cross diffusion terms generate four different types of solutions to the one-dimensional unsteady diffusion equations for ternary mixtures. The stability of the fluid column corresponding to these solutions can be classified depending on the sign of the first derivative of density with respect to the direction of the gravity vector (i.e. ∂ρ/∂y) and the sign of (∂2ρ/∂y2)/y, where y=0 is located at the center of the diffusion layer. The type of solution depends on the initial conditions and on the set of diffusion coefficients considered. One type of solution corresponds to a stable fluid column with (∂ρ/∂y<0) and (∂2ρ/∂y2)/y>0. Two types of solutions generate fluid layers with unstable density stratification (∂ρ/∂y>0) and the fourth type shows a fluid layer with (∂2ρ/∂y2)/y<0. We analyzed the unsteady diffusion processes in a ternary mixture under the conditions of experiments carried out to determine the four diffusion coefficients of the ternary system 1,2,3,4-tetrahydronaphthaline-isobutylbenzene-dodecane (THN-IBB-nC12). These measurements; performed in diffusion cells with an initially stable stratification within the cell, with the denser mixture at the bottom and the lighter at the top; are usually based on the validity of the one-dimensional unsteady diffusion mass transfer equations. The linear stability analysis for the onset of convection in the unstable layers with unstable density stratification (∂ρ/∂y>0) indicates that the critical thickness of these layers depends on the Rayleigh numbers, on the diffusion coefficients and on the initial conditions. To illustrate the flow structures that can be generated in these unstable conditions, we performed numerical simulations for selected sets of diffusion coefficients at different Rayleigh numbers. The results of these simulations are in general agreement with the predictions of the linear stability analysis and indicate that, under specific conditions, the convective motions developed in the cell produce significant departures of the concentration distributions from the pure diffusion situation.

Flow past a moderately heated horizontal cylinder at low Reynolds number

Flow past a heated two-dimensional cylinder in the laminar regime is investigated. A hybrid finite element–finite volume method is used to solve Navier–Stokes and energy equations. The vortex shedding frequency defined nondimensionally as Strouhal number is observed to be increasing with heating. Vortex shedding behind heated cylinder exhibits asymmetry in wake region due to stable and unstable density stratification below and above the cylinder, respectively. A negative lift is generated due to heating, in an otherwise symmetric zero time-averaged lift configuration, due to skewness in vortex shedding towards the direction of buoyancy. We also note that the heating brings down the drag for the cases considered here. The thermal energy transport due to convection, defined by nondimensional Nusselt number is found to increase only marginally with cylinder heating.

Multiple steady states in exchange flows across faults and the dissolution of

We develop a model of the steady exchange flows which may develop between two aquifers at different levels in the geological strata and across which there is an unstable density stratification, as a result of their connection through a series of fractures. We show that in general there are multiple steady exchange flows which can develop, depending on the initial conditions, and which may involve a net upwards or downwards volume flux. We also show that there is a family of equilibrium exchange flows with zero net volume flux, each characterised by a different interlayer flux of buoyancy. We present experiments which confirm our simplified model of the exchange flow. Such exchange flows may supply unsaturated water from a deep aquifer to drive dissolution of a structurally trapped pool of geologically stored $\text{CO}_{2}$, once the water in the aquifer containing the trapped pool of $\text{CO}_{2}$ has become saturated in $\text{CO}_{2}$, and hence relatively dense. Such exchange flows may also lead to cross-contamination of aquifer fluids, which may be of relevance in assessing risks of geological storage systems.

The feasibility of thermal and compositional convection in Earth's inner core

Inner core convection, and the corresponding variations in grain size and alignment, has been proposed to explain the complex seismic structure of the inner core, including its anisotropy, lateral variations and the F-layer at the base of the outer core. We develop a parametrized convection model to investigate the possibility of convection in the inner core, focusing on the dominance of the plume mode of convection versus the translation mode. We investigate thermal and compositional convection separately so as to study the end-members of the system. In the thermal case the dominant mode of convection is strongly dependent on the viscosity of the inner core, the magnitude of which is poorly constrained. Furthermore recent estimates of a large core thermal conductivity result in stable thermal stratification, hindering convection. However, an unstable density stratification may arise due to the pressure dependant partition coefficient of certain light elements. We show that this unstable stratification leads to compositionally driven convection, and that inner core translation is likely to be the dominant convective mode due to the low compositional diffusivity. The style of convection resulting from a combination of both thermal and compositional effects is not easy to understand. For reasonable parameter estimates, the stabilizing thermal buoyancy is greater than the destabilizing compositional buoyancy. However we anticipate complex double diffusive processes to occur given the very different thermal and compositional diffusivities.

The effect of natural and forced melt convection on dendritic solidification in Ga–In alloys

The directional solidification of Ga–25wt%In alloys within a Hele-Shaw cell was visualized by means of X-ray radioscopy. The experimental investigations are especially focused on the impact of melt convection on the dendritic growth. Natural convection occurs during a bottom up solidification because lighter solute is rejected at the solid–liquid interface leading to an unstable density stratification. Forced convection was produced by a rotating wheel with two parallel disks containing at their inner sides a set of permanent NdFeB magnets with alternating polarization. The direction of forced melt flow is almost horizontal at the solidification front whereas local flow velocities in the range between 0.1 and 1.0mm/s were achieved by controlling the rotation speed of the magnetic wheel. Melt flow induces various effects on the grain morphology primarily caused by the convective transport of solute. Our observations show a facilitation of the growth of primary trunks or lateral branches, suppression of side branching, dendrite remelting and fragmentation. The manifestation of all phenomena depends on the dendrite orientation, local direction and intensity of the flow. The forced flow eliminates the solutal plumes and damps the local fluctuations of solute concentration. It provokes a preferential growth of the secondary arms at the upstream side of the primary dendrite arms, whereas the high solute concentration at the downstream side of the dendrites can inhibit the formation of secondary branches completely. Moreover, the flow changes the inclination angle of the dendrites and the angle between primary trunks and secondary arms.