Abstract

In this work, the fluid dynamic behaviour around a proposed micro-sensor geometry is assessed. The sensor is intended for use in micro devices and is represented by two cubes (or elements) set in tandem. The flow is described by the Navier-Stokes equations and is solved by Direct Numerical Simulation (DNS) using ANSYS-CFX. Flow visualization using an existing experimental setup is carried out to visualize the flow pattern around the sensor using the soap film technique, where the flow is considered to be two-dimensional incompressible. This visualization is intended to verify the DNS that is carried out for the same cases. Results for the flow pattern and the vortex shedding frequency that are obtained from both numerical simulations and experimental investigations compare favorably, for three different values of Reynolds number, which verifies the numerical approach.
A parametric study on the effect of geometry in the limit of 2D incompressible flow of water is carried out. This study shows that the inter-element spacing strongly affects the flow in the inter-element cavity; it also shows that the thickness of the downstream element affects the downstream shear layer. Both of these geometric parameters control the vortex shedding in the wake and the drag coefficient particularly on the downstream element. This parametric study also suggests that a ‘general’ linear correlation between Strouhal and Reynolds numbers (modified to include geometric parameters) is valid for all variables investigated in this work.
The DNS of air in the limit of 2D and 3D flow is considered at three subsonic inflow Mach numbers. The flow simulation results are verified against basic flow physics, available experimental data and interpretations of vortex shedding behaviour particularly in 3D flow. For air flow in the 2D limit, the vortex shedding frequency expressed in terms of Roshko number (rather than Strouhal number) correlates well with Reynolds number at all Mach numbers. For both 2D and 3D flows, the vortex shedding frequency, flow behavior and drag coefficient compare reasonably well with available experimental data.
The drag coefficient and Strouhal number computed from DNS will serve as a first step towards inferring the flow pressure and velocity, when the proposed sensor is built. Hence achieving one of the main goals of this work.