No abstract

In this study we investigate the onset of three-dimensionality in an otherwise two-dimensional periodic wake of a square cylinder. Floquet stability analysis is employed to extract the different modes of three-dimensional instabilities. It is observed that the three-dimensional transition process for a square cylinder is similar to that of a circular cylinder. Most notably, there is a long-wavelength (mode A) three-dimensional disturbance that becomes unstable first at a Reynolds number of about 161, followed by a short-wavelength (mode B) three-dimensional disturbance that becomes unstable at a Reynolds number of about 190. In addition, a third intermediate-wavelength mode is also observed to become unstable at around Re=200. Unlike modes A and B, the intermediate-wavelength mode is subharmonic with a period of twice the shedding period of the two-dimensional base state. This mode also breaks the reflection translation symmetry observed in the other two modes and as a result appears with multiplicity two. The space–time symmetries of the three modes are explored in detail.

The heat generation and fluid convection induced by plasmonic nanostructures is attractive for optofluidic applications. However, previously published theoretical studies predict only nanometre per second fluid velocities that are inadequate for microscale mass transport. Here we show both theoretically and experimentally that an array of plasmonic nanoantennas coupled to an optically absorptive indium-tin-oxide (ITO) substrate can generate 4micro-metre per second fluid convection. Crucially, the ITO distributes thermal energy created by the nanoantennas generating an order of magnitude increase in convection velocities compared with nanoantennas on a SiO 2 base layer. In addition, the plasmonic array alters absorption in the ITO, causing a deviation from Beer-Lambert absorption that results in an optimum ITO thickness for a given system. This work elucidates the role of convection in plasmonic optical trapping and particle assembly, and opens up new avenues for controlling fluid and mass transport on the micro-and nanoscale.

Unsteady three-dimensional flow in the mold region of the liquid pool during continuous casting of steel slabs has been computed using realistic geometries starting from the submerged inlet nozzle. Three largeeddy simulations (LES) have been validated with measurements and used to compare results between full-pool and symmetric half-pool domains and between a full-scale water model and actual behavior in a thin-slab steel caster. First, time-dependent turbulent flow in the submerged nozzle is computed. The time-dependent velocities exiting the nozzle ports are then used as inlet conditions for the flow in the liquid pool. Complex time-varying flow structures are observed in the simulation results, in spite of the nominally steady casting conditions. Flow in the mold region is seen to switch between a "doubleroll" recirculation zone and a complex flow pattern with multiple vortices. The computed time-averaged flow pattern agrees well with measurements obtained by hot-wire anemometry and dye injection in fullscale water models. Full-pool simulations show asymmetries between the left and right sides of the flow, especially in the lower recirculation zone. These asymmetries, caused by interactions between two halves of the liquid pool, are not present in the half-pool simulation. This work also quantifies differences between flow in the water model and the corresponding steel caster. The top-surface liquid profile and fluctuations are predicted in both systems and agree favorably with measurements. The flow field in the water model is predicted to differ from that in the steel caster in having higher upward velocities in the lower-mold region and a more uniform top-surface liquid profile. A spectral analysis of the computed velocities shows characteristics similar to previous measurements. The flow results presented here are later used (in Part II of this article) to investigate the transport of inclusion particles.

No abstract

No abstract

The fully developed turbulent flow in a straight duct of square cross section has been simulated using the large eddy simulation (LES) technique. A mixed spectral-finite difference method has been used in conjunction with the Smagorinsky eddy-viscosity model for the subgrid scales. The simulation was performed for a Reynolds number of 360 based on friction velocity (5810 based on bulk velocity) and duct width. The simulation correctly predicted the existence of secondary flows and their effects on the mean flow and turbulence statistics. The results are in good qualitative agreement with the experimental data available at much higher Reynolds numbers. It is observed that both the Reynolds normal and shear stresses equally contribute to the production of mean streamwise vorticity.

Particle motion and capture in continuous steel casters were simulated using a Lagrangian trajectorytracking approach, based on time-dependent flow fields obtained from large-eddy simulations (Part I of this article). A computation was first conducted on a water model of a full-scale standard slab caster, where measurements were available. It simulated the transport of 15,000 plastic particles and their removal by a screen positioned near the mold top surface. The computation shows the screenremoval fractions to be 27 Ϯ 5 pct for 0 to 10 seconds and 26 Ϯ 2 pct for 10 to 100 seconds, which agrees with previous measurements. The flow exiting the nozzle was relatively uniform, and turbulent motion in the domain was very chaotic, so particle removal did not depend on the initial location of particles introduced in the nozzle port. A computation of motion and capture of 40,000 small inclusions (10 and 40 m) was then performed in an actual thin-slab steel caster. The particles moved through the mold region with an asymmetrical distribution, which was caused by transients in fluid turbulence in the lower recirculation region, rather than by inlet variations at the nozzle port. Only about 8 pct of these small particles were removed to the top surface. This removal fraction was independent of both particle size and density, likely because all the simulated particles were too small to deviate significantly from the surrounding fluid flow. Finally, the computational results were further processed to predict the ultimate distribution of impurity particles in the solid thin slab after a short burst of inclusions entered the mold. They were reprocessed to reveal the distribution of total oxygen content for a steady inclusion supply from the nozzle. The results of this work confirm the important role of flow transients in the transport and capture of particles during continuous casting and can serve as a benchmark for future simplified models.