Direct numerical simulation of compressible two-species mixing layer

Abstract

Over the past several decades, the plane mixing layer of the two parallel streams of fluid with different velocities has been studied. An understanding of the physics knowledge of turbulence in the mixing layer is very important for engineering design, especially with the design of high-speed jet engine combustion chambers. To predict the efficiency of combustion, the turbulence in the mixing layer of two substances has to be estimated accurately. In this research, the mixing of two fluids with different densities and viscosities in a subsonic region is investigated using Direct Numerical Simulation (DNS) solver. A DNS solver of the two-species flow is first developed. A mass fraction model together with a mixture rule is employed to capture material interfaces. A novel sixth-order Characteristic Weighted Essentially Non-Oscillatory with Central-Upwind Localized Dissipative nonlinear interpolation (CWENO6-CULD) is implemented to discretize the convective term and deal with the discontinuous interface of the different fluid properties, while the viscous term is discretized with the sixth-order Central Differencing scheme (CD6). The third-order Total Variation Diminishing Runge-Kutta (RKTVD) is applied for the temporal discretization to increase the stability of the numerical schemes. A convergence test of the CWENO6-CULD is conducted to confirm its sixth-order convergence rate. Additionally, various test cases are performed to ensure the capability of the CWENO6-CULD in capturing sharp discontinuities, while correctly solving small-scale structures. The flow physic of the compressible two-species mixing layer is then investigated. The simulation results reveal that the difference in densities and vicosities does not much affect the evolution of the flow in laminar and fully turbulent regions. On the other hand, it affect so much in the transition region. The flow mechanism is altered when the lighter fluid was on top of the heavier fluid. This resulted in a decrease in the transition length.