Simulations of turbulent interfaces produced by positive and negative buoyancy are conducted by moving blocks of fluid in the direction of the flow. The second moment of the blocks increases at a rate proportional to the diffusivity. The block simulation is free of numerical oscillations. Unlike most classical methods, the error associated with Lagrangian block simulation is not cumulative. Artificial diffusion error is negligible. The non-diffusive Lagrangian block simulations have provided reliable data to evaluate the performance of (i) sub-grid scale modelling and (ii) K-ε modelling of turbulent flow under the opposing influence of buoyancy.
INTRODUCTIONMost computational fluid-dynamics codes are developed using the Eulerian description. To find the numerical solution, fluxes are estimated on the surface of the finite volume using a truncation series. Spurious numerical oscillations and artificial numerical diffusion are consequences, particularly in regions across flow discontinuities. Diffusion often is introduced synthetically in many schemes to gain computational stability. Occasional switching to a diffusive upwind scheme is one classic strategy to manage the numerical oscillations [1,2,3,4,5]. Lagrangian block simulation (LBS) offers an alternative that could eliminate the spurious numerical oscillations and false diffusive error [6,7]. The blocks move in the direction of the flow. The squares of the block widths expand in proportion to the diffusivities. The block simulation procedure consists of three steps: (i) Lagrangian advection and diffusion, (ii) division into portions, and (iii) reassembly of the portions into new blocks. The blocks are renewed in each time increment to prevent excessive distortion. In this paper simulation of buoyancy at turbulence interfaces has been carried out using the LBS method. In one series of simulations, the Kelvin-Helmholtz instabilities initiate turbulence across