Experimental observations of high-energy surface melting processes, such as laser welding, have revealed unsteady, often violent, motion of the free surface of the melt pool. Surprisingly, no similar observations have been reported in numerical simulation studies of such flows. Moreover, the published simulation results fail to predict the post-solidification pool shape without adapting non-physical values for input parameters, suggesting the neglect of significant physics in the models employed. The experimentally observed violent flow surface instabilities, scaling analyses for the occurrence of turbulence in Marangoni driven flows, and the fact that in simulations transport coefficients generally have to be increased by an order of magnitude to match experimentally observed pool shapes, suggest the common assumption of laminar flow in the pool may not hold, and that the flow is actually turbulent. Here, we use direct numerical simulations (DNS) to investigate the role of turbulence in laser melting of a steel alloy with surface active elements. Our results reveal the presence of two competing vortices driven by thermocapillary forces towards a local surface tension maximum. The jet away from this location at the free surface, separating the two vortices, is found to be unstable and highly oscillatory, indeed leading to turbulence-like flow in the pool. The resulting additional heat transport, however, is insufficient to account for the observed differences in pool shapes between experiment and simulations.

Control of self-sustained jet oscillations in confined cavities is of importance for many industrial applications. It has been shown that the mechanism underlying these oscillations consists of three stages: (i) growth of the oscillation, (ii) amplitude limitation and (iii) delayed destruction of the recirculation zone bounding the jet. It has also been shown that oscillations may be enhanced or suppressed by applying (e.g. electromagnetic) body forces. In the current paper we study the influence of electromagnetic forces oriented aligned with or opposite to the direction of the jet on the oscillation mechanism. The influence of the forcing is found to depend on the Stuart number N in relation to a critical Stuart number Ncrit. We demonstrate that for |N| < Ncrit , the oscillation mechanism is essentially unaltered, with moderate modifications in the jet oscillation amplitude and frequency compared to N=0N=0. For N > Ncrit, electromagnetic forcing leads to total suppression of the self-sustained oscillations. For N < Ncrit , electromagnetic forces dominate over inertia and lead to strongly enhanced oscillations, which for N≪−NcritN≪−Ncrit become irregular. As was earlier demonstrated for N=0,N=0, the present paper shows that for −6Ncrit<N<Ncrit−6Ncrit<N<Ncrit the oscillatory behaviour, i.e. frequencies, oscillation amplitudes and wave shapes, can be described quantitatively with a zero-dimensional model of the delay differential equation (DDE) type, with model constants that can be a priori determined from the Reynolds and Stuart number and geometric ratios.

We study the flow in a layer of conductive liquid under the influence of surface tension, gravity, and Lorentz forces due to imposed potential differences and transverse magnetic fields, as a function of the Hartmann number, the Bond number, the Reynolds number, the capillary number and the height-to-width ratio A. For aspect ratios A 1 and Reynolds numbers Re≤A, lubrication theory is applied to determine the steady state shape of the liquid surface to lowest order. Assuming low Hartmann (Ha ≤ O(1)), capillary (Ca ≤ O(A⁴)), Bond (Bo ≤ O(A²)) numbers and contact angles close to 90°, the flow details below the surface and the free surface elevation for the complete domain are determined analytically using the method of matched asymptotic expansions. The amplitude of the free surface deformation scales linearly with the capillary number and decreases with increasing Bond number, while the shape of the free surface depends on the Bond number and the contact angle condition. The strength of the flow scales linearly with the magnetic field gradient and applied potential difference and vanishes for high aspect ratio layers (A → 0). The analytical model results are confirmed by numerical simulations using a finite volume moving mesh interface tracking method, where the Lorentz force is calculated from the equation for the electric potential. It is shown that the analytical result for the free surface elevation is accurate within 0.4% from the numerical results for Ha² ≤ 1, Ca ≤ A⁴, Bo ≤ A², Re ≤ A and A ≤ 0.1 and within 2% for A=0.5. For A=0.1, the solution remains accurate within 1% of the numerical solution when either Ha² is increased to 400, Ca to 200A⁴ or Bo to 100A².