LES of Delft Jet-in-Hot-Coflow (DJHC) with Presumed PDF Model and Tabulated Chemistry


Summary of the MSc Thesis Applied Physics by Alexander Klaessen, presented on August, 21, 2012

Supervisors: Prof. dr. D.J.E.M. Roekaerts, Dr. M.K. Stoellinger, G. Sarras MSc

A Large Eddy Simulation (LES) of the Delft Jet-in-Hot-Coflow (DJHC) burner has been performed with the software package OpenFOAM. At the start of this thesis, only Reynolds Averaged Navier-Stokes (RANS) codes had been applied in the numerical simulation of the DJHC burner. Therefore the goal of this thesis is to make a first attempt on modeling the DJHC burner with a LES code. The expectation from LES is that it can handle accurately the effects of large-scale turbulent structures on the flame stabilization process.

In this initial LES study a combination of simple models for chemistry and turbulence-chemistry interaction is used. To describe mixing between fuel and coflow and effects of combustion reactions, the flamelet/progress variable approach is used with a presumed probability density function (PDF) method to close the source term. For simplicity the experimentally observed heat loss effects at the inflow boundary (enthalpy deficit) and effects of entrainment of surrounding air are neglected. Turbulent fluctuations at the inflow boundary are taken into account using a synthetic turbulence generator.

A simulation of one of the experimentally studied cases was successfully completed. The consequences of the simplifying assumptions concerning enthalpy deficit and mixing with surrounding air can be clearly seen in the predicted profiles of mean temperature. The influence of the limitations of the synthetic turbulence generator can be seen in the velocity profiles, but this effect is less pronounced compared to the other mentioned aspects. The prediction of the mean velocity profile is in agreement with experiments, but the predicted Reynolds stress shows deviations.

The recommendations for future work are as follows: it is advised to include enthalpy deficits and mixing with the surrounding air in the model. The latter can be accomplished by introducing a three-stream problem. To improve the velocity predictions, also the inflow turbulence generation algorithm must be changed or improved.


Involved People:

Facilities used: