In many fluid flows, transition of boundary layers from laminar to turbulence is forced by free-stream perturbations. This phenomenon is called Bypass Transition, and affects various engineering applications including, for example, turbo-machinery flows. As a result, transition prediction is recognized as a key factor in improving the design of these machines.

The clip below shows iso-surfaces of streamwise velocity perturbations, and vertical plane shows the free-stream turbulence. The flow is initially laminar and characterized by long streamwise perturbations — the boundary layer streaks also known as Klebanoff distortions. A turbulent spot is observed downstream, followed by the fully turbulent boundary layer. The inception of turbulent spots or patches is sporadic, both in space and time. Its frequency and propensity dictate the length of the transitional region.

We have also developed novel structure identification and tracking algorithms that enable us to examine the entire temporal and spatial evolution of the pre-transitional streaks, and to extract the particular streaks that break down into spots.  These streaks are generally much high amplitude than the rest of the population, and are often two- to three-times higher in magnitude than the average streak.

Beyond the secondary instability of the steaks, the emergent turbulent spots can be tracked as well.  This enable us to examine the development of turbulence and the dynamics of the laminar-turbulence interface.  The video below shows our spot tracking as the turbulent patch expands and occupies a larger volume within the boundary layer, and ultimately merge with the downstream fully-turbulent flow.

To aid in transition modeling, we have performed detailed direct numerical simulations (DNS) of bypass transition over a wide range of conditions.  These computational experiments were only possible through the development and implementation of efficient, scalable algorithms.  The example below compares transition onset at different levels of streamwise pressure gradient, starting with a favorable, then zero and finally two progressively more adverse pressure gradients.