Numerical modelling of flame flashback in premixed tube burners with turbulent flow and high hydrogen content

Numerical modelling of flame flashback in premixed tube burners with turbulent flow and high hydrogen content

van Put, Max (TU Delft Mechanical, Maritime and Materials Engineering)

Klein, S.A. (mentor)
Tummers, M.J. (graduation committee)
Roekaerts, D.J.E.M. (graduation committee)

Delft University of Technology

2021-03-05

This study presents the numerical modeling of boundary layer flame flashback in premixed tube burners with turbulent flow and high hydrogen content. The modeling approach is based on a RANS turbulence model and a flamelet combustion model using a flame surface density approach in combination with multiple turbulent flame speed correlations to close the chemical source term. The Zimont turbulent flame speed correlation is used to model flame flashback and gain insight in the relevant mechanisms driving the occurrence of the instability. However, this approach has not let to a method to predict in ad- vance that a flame is near flashback conditions. This implicates that the flashback process (when mod- eled with this specific modeling approach) seems to occur after a certain threshold has been passed. When this threshold has been passed the flashback process is self-amplifying. The ability of the mod- eling approach based on the Zimont correlation to predict flame flashback is compared to experimental data. Based on this comparison it is concluded that the numerical model performs fairly well. The most notable result is the flame regime map which shows that the numerical model is able to capture that an increase in the hydrogen concentration in the fuel leads to an increased flashback propensity, which was also seen in experimental results. Furthermore, high hydrogen concentration fuels are character- ized by high flame speeds and, if sufficiently lean, by a Lewis number lower than 1. In combination with differential diffusion this leads to the thermal diffusive instability. At an attempt to improve the nu- merical model, two turbulent flame speed correlations that account for these effects were selected and implemented in the numerical model. However, these correlations did not lead to an improved ability of the model to predict flame flashback and flame shape. This reinforces the initial findings that the numerical model with the Zimont turbulent flame speed correlation already performed fairly well.

Flame flashback
CFD modelling
RANS turbulence modelling
Combustion Instability

http://resolver.tudelft.nl/uuid:84b5e88d-72b8-4663-a597-84993aa347f7

Student theses

master thesis

© 2021 Max van Put