A robust methodology for the detection of turbulent/non-turbulent interfaces in temporally evolving turbulent flows

Abstract A new interface detection technique is proposed to identify turbulent/non-turbulent interfaces (TNTIs) in temporally-evolving flows, and its efficiency in comparison to existing techniques is assessed for two fundamentally different test cases. The first consisted of a temporally-evolving cylinder wake, and the second a decaying stream of grid-turbulence. The turbulent flows were marked with a high Schmidt number dye and combined PIV/PLIF experiments were performed. The TNTIs were identified using PLIF image intensity and intensity-gradient based methods. Since the bulk concentration decayed with time, a time dependent normalisation of the images had to be performed in all methods. A novel intensity-based method was proposed where the intensity field was transformed using a combination of pixel-based intensity and the corresponding cross-stream, y, location. The efficacy of the different methods were compared using metrics such as total enstrophy contained within the interface and enstrophy gradient along the interface. The new intensity-based method showed a superior performance as it captured both high total enstrophy and enstrophy gradient.