Lattice Boltzmann simulation of cross diffusion via Soret and Dufour effects on natural convection of experimental data based MWCNTs-H[formula omitted]O nanofluids in an L-shaped enclosure

Abstract

The cross-diffusion via Soret and Dufour impacts on natural convection in an L-shaped enclosure full of multi-walled carbon nanotubes (MWCNTs)-H2O (water) nanofluids has been studied numerically utilizing the multiple-relaxation-time (MRT) lattice Boltzmann method (LBM) through graphics processing unit (GPU) computing. The indented part of the L-shape is cold (Tc), while the left and bottom walls of the chamber maintain a heated (Th) temperature. The adiabatic boundary condition is present on the cavity’s remaining walls. The thermal conductivity and viscosity of the nanofluids are the function of fluid temperature, and the corresponding correlation was derived from the experimental data. The current methodology and the experimental data based (MWCNTs)-H2O (water) nanofluids outcomes are wholly noble. Numerical simulations have been done for numeroust relevant pertinent such as Rayleigh number, 3×103≤Ra≤3×105, buoyancy ratio, −2≤Br≤2, volume fraction, 0.0≤ϕ≤0.01, Soret number, 0.0≤Sr≤0.2, Dufour number, 0.0≤Df≤0.2, while Prandtl number, Pr=6.2, and Lewis number, Le=2 are fixed. The results are presented regarding the streamlines, isotherms, isoconcentration, velocity and temperature distribution, local Nusselt and average Nusselt number, local Sherwood and average Sherwood number, and local and total entropy production. It is discovered that with growing Rayleigh numbers, volume fractions, and Soret numbers, the Nu¯ and the Sh¯ rise. For various Rayleigh numbers Ra = (3×103, 3×104, 3×105), the Nu¯ grew by 29.95%, 19.86%, and 1.77% respectively when ϕ increased from 0.0 to 0.01. The Nu¯ graph reaches its maximum for Ra=3×105. The streamlines grow as Ra enhances. In addition, when the volume fraction rises from 0.0 to 0.02, the local Nu rises because of more potent thermal conductivity, but the local Sh falls. The highest entropy formation is found by Ra=(3×105), and the general rate of entropy formation improves with enlarged Rayleigh numbers while the Bejan number stays constant. Finally, a response surface analysis has been done, and the output closely matches the original simulation results.