A cost-effective large-eddy simulation framework for boundary layer transition at low turbulence intensities

  • Abhi Beechook

Student thesis: Doctoral ThesisDoctor of Philosophy

Abstract

When using large-eddy simulation (LES) to study natural transition, two-dimensional disturbancesin the form of Tollmien-Schlichting (T-S) waves are introduced in the laminar boundary layereither via a wall \blowing and suction" strip or an inow boundary condition. In order to performreliable natural transition simulations it is crucial, and often challenging, to prescribe appropriateinitial conditions, for instance, the disturbance amplitude. It is indisputable that the initial T-Swave amplitude is governed by the transformation mechanism by which external environmentalperturbations enter the laminar boundary layer, a process referred to as receptivity.One of the major challenges in transition prediction lies in determining appropriate initial T-Swave amplitudes. To this date, this challenge remains unsolved due to the complexity and nonuniversalityof the receptivity mechanism. In this work, a new framework for estimating the initialamplitude within the linear region of the laminar boundary layer for a spatially developing T-S waveis presented. This framework is developed, for the rst time, exploiting the empirical observationthat the growth of T-S waves becomes non-linear once the r.m.s of the amplitude reaches a valueof approximate 1% of the freestream velocity. The new framework also oers the exibility toprovide approximations for initial T-S wave amplitudes at any streamwise location between theregion where a T-S wave starts to amplify and the location where its amplitude reaches 1% of thefreestream velocity.It is known that linear stability theory (LST) predictions (as part of the eN transition predictionframework) can be used to relate the transitional Reynolds number (transition location onset) andthe turbulence intensity of the freestream. Based on this knowledge, the amplitude estimationframework is extended to provide approximations of suitable initial amplitudes as a function of thefreestream turbulence intensity level (ranging between 0 and 0:5%). To the author's knowledge, thisallows, for the rst time, the application of LES to congurations where it is important to investigatethe eects of the turbulence level in the environment. The transition predicting capability of the newamplitude framework is assessed on natural transition experimental and 3D LES and the agreement is excellent for the congurations tested. The transition location ispredicted (in terms of the skin friction coecient) within a maximum deviation of 1:2% from theexperimental data of Schubauer and Klebano [133]. In addition, the new framework showed theability to reproduce transitional ow features at low turbulence intensities that are challenging toexplore under experimental conditions. As part of the methodology towards realising this, a newLES inow boundary condition is developed that embeds LST information as analytical functions,providing a means to facilitate boundary layer transition computations at low turbulence intensitiesat a reduced computational cost (ranging between 25 and 60% savings).Whilst initially developed for, and evaluated on, a Blasius boundary layer, the proposed amplitudeframework and the initial amplitude to turbulence intensity relationship can help the modellingcommunity to perform ecient LES computations and generate high-delity data. This opens avenuesfor in-depth exploration of the physics of natural transition, at a reduced computational cost,and can inform the development of new transition predicting tools. Finally, a key advantage of thenew framework is that it oers a simple alternative method for estimating initial T-S wave amplitudesi.e. without the need for receptivity calculations. The new amplitude method is reasonablystraightforward to use and results in a reduction of computational cost.at plate test cases using 2.5D
Date of Award2021
Original languageEnglish
Awarding Institution
  • Coventry University
SupervisorHumberto Medina (Supervisor)

Cite this

'