School of Mechanical & Systems Engineering
Seminar Series: 2011/2012

"Obtaining 3D turbulent premixed combustion statistics from 2D experimental measurements:  Development of experimental diagnostics using Direct Numerical Simulations"

Professor Nilanjan Chakraborty
(School of Mechanical and Systems Engineering)

Thursday 13 October 2011
4pm, Stephenson Building, room F16 (first floor)

To date, in most existing experimental measurements the flame location and structure are often identified by Rayleigh scattering, Planar Laser Induced Fluorescence (PLIF) and tomography of small vaporising droplets, while the velocity and velocity gradient measurements are carried out using Particle Image Velocimetry (PIV). Very often these measurements have been carried out in two-dimensions for the measurements of the local flame propagation speed, reacting scalar gradient, curvature, strain rate and flame stretch rate. However, turbulent combustion is inherently three-dimensional in nature and thus the two-dimensional experimental measurements cannot capture the three-dimensional flame propagation and the species and velocity fields associated with it. Although a few three-dimensional measurements have been carried out where the experimental data is obtained from the intersection line of two intersecting planes. Such methods are still very expensive and thus two-dimensional measurements will have relevance in the foreseeable future. This fact essentially necessitates the estimation of the correction factors for extracting three-dimensional quantities from two-dimensional measurements so that two-dimensional experimental data can be used for developing new models and assessing the performance of new combustion models.

Direct Numerical Simulations (DNS) allows for simulation of three-dimensional turbulent flows without any kind of physical approximation. In DNS the smallest relevant length and times of turbulent flows are adequately resolved and thus DNS data can be treated as experimental data with resolution up to the Kolmogrorov length scale, which is either difficult or impossible to obtain in routine combustion measurements. In the present study three-dimensional compressible DNS data of freely propagating statistically planar turbulent premixed flames has been used to assess the accuracy of the isotropy derived correction factors, which relate the two-dimensional projections of the various key quantities for the turbulent combustion modelling with their corresponding actual three-dimensional counterparts for different values of Karlovitz number Ka, Lewis number Le, heat release parameter tau and turbulent Reynolds number Rei. The isotropic distribution of the surface area weighted probability density function (pdf) of the angle phi between the normal vectors on the measurement plane and on the flame surface provides a number of simple algebraic relations between the two-dimensional projections of a number of key quantities related to the scalar gradient transport with their three-dimensional counterparts. It is found that the threshold value of Rei, above which the assumption of isotropy yields an accurate result, depends on the values of heat release parameter, Lewis number and the regime of the prevailing combustion process. Detailed physical explanations are provided for the aforementioned dependence of the isotropy-derived correction factors on Lewis number, heat release parameter, Karlovitz number and the turbulent Reynolds number of the turbulent premixed combustion in question. 

Nilanjan Chakraborty is Professor of Fluid Dynamics and heads the MSE Multiphase Flow and Thermal Systems research group.  You can view his profile at