My Journey

Phase 1: This study aims at simulating the coherent flow structures in the wake of axisymmetric bodies at high angles of attack and quantifying the unsteady body forces due to vorticity, shear layer, and other aerodynamic structures associated with the cross-stream flow. For the direct numerical simulation (DNS), the unsteady incompressible Navier-Stokes equations were solved on curvilinear staggered grids by using a second-order fractional step method in time and a central difference method in space. The momentum equations are solved by a matrix-free Newton–Krylov solver. The Poisson equations are solved using an iterative solver with a multigrid preconditioner. The surface of an axisymmetric body is represented with triangular meshes and a sharp interface immersed boundary method is used to impose wall boundary conditions. Simulations were conducted to investigate the flow over the axisymmetric cone and cone-cylinder model at different angles of attack and Reynolds numbers. For the identification and isolation of the most important flow features, first, an anisotropic Gaussian filter is applied to remove unnecessary small-scale structures. The size and strength of the filter are determined based on the spatial characteristics of the smallest length scale of the vortical elements in the flow. Secondly, on this filtered field, an expectation-maximization (EM) clustering algorithm is applied for systematic identification of the core of the primary vorticity. 

Shockwave and boundary-layer interaction over a flexible surface with different thermal boundary conditions is studied in this project. In our simulation, a sixth-order compact central finite difference scheme and a third-order total variation diminishing Runge-Kutta (RK3-TVD) method are used for spatial and temporal discretization of the compressible governing NS equations, respectively. To ensure the numerical stability near the shock, artificial shear viscosity, bulk viscosity, and thermal diffusivity are added. This in-house flow solver is coupled with a finite-element structural solver via a partitioned approach. At each time step, the flow is solved first with the present status of the interface. Then, the fluid forces are exerted on the boundary of the structure and the structural solver finds the deformation parameters, strain, and principal stress in the solid. Then, again, velocity boundary conditions at the interface are calculated from the velocity and displacement of the fluid-solid interface. The process is iterated until both dynamic and kinematic boundary conditions at the interface are satisfied. A synthetic digital filtering approach is used to generate the turbulent boundary layer at the inlet by adding the velocity fluctuations to the mean fully developed turbulent compressible boundary. The flow is characterized by an oblique impinging shock on a compliant panel and the corresponding separation bubble, expansion fan, reflected, and re-attachment shock. An unsteady self-excited decaying oscillation is observed, which results in significant changes in the interaction region and downstream wake over time. Modal analysis of the panel helps to find the mode shape that is mostly responsible for energy transfer in this FSI system and separate the unsteady forces coming from the flow and forces due to the motion of the panel. The recent improvement of the MPI communication framework between the structural solver and the flow solvers makes the code 21.77% faster.