FLAME PROPAGATION IN SWIRLING FLOWS—EFFECT OF LOCAL EXTINCTION ON THE COMBUSTION INDUCED VORTEX BREAKDOWN

Abstract

The interaction of chemical reaction and swirling flows can lead to accelerated flame propagation. In this context, the combustion induced vortex breakdown (CIVB) has been identified in previous studies as prevailing flashback mechanism in swirl burners. However, quenching of the chemical reaction can prevent the upstream flame propagation due to a CIVB. The present study shows that this limiting criterion can be described by a quenching constant that is balancing the chemical time scale with a time scale for the flow. The most appropriate chemical time scales are based on perfectly stirred reactor (PSR) simulations. To consider the influence of the preferential diffusion typical for hydrogen, the PSR time scales are corrected by the Lewis number. Once the quenching constant is determined experimentally for a specific burner configuration, the flashback limits for various operating conditions and fuels can be predicted. Finally, the pressure scaling of the flashback limits is discussed.

Keywords:

• Combustion induced vortex breakdown
• Flame propagation
• Flashback
• Flashback limits
• Gas turbine combustion

Notes

¹There was no optical access at z ≈ − 0.5 [sdot] d. Further on it has to be mentioned that the stagnation point may be shifted if a reaction is stabilized in the recirculation zone.

² S _L is calculated from a mixture equation (Kröner, 2003; Kröner et al., 2003) and equations for and of Peters (1994).

³The flashback limits for BC1 and BC3 are discussed later.

⁴Eq. (6) can be transfered to a Peclet Number model proposed by Putnam and Jensen (Putnam and Jensen, 1948) by multiplying with (d ²/aτ _c ) (Kröner et al., 2003), (Kröner, 2003).

⁵Here, the convection induced vortex structures define the local flame front geometry. In particular, the thickness of the reacting layer is no longer depending on the thermal or diffusive transport properties of the fuel air mixture as in a undisturbed laminar flame front, which is a precondition for the derivation of the chemical time scale . Therefore, the thickness of the flame front is initially independent of the gas temperature or the fuel properties in this case.

⁶Abdel-Gayed et al. (1984), (Abdel-Gayed and Bradley, 1985 1989) propose a similar approach correcting the Karlovitz stretch factor for a quenching criterion in turbulent flames: K [sdot] L e ⁿ  = const. They determined the exponent empirically to n = 1, which leads to the quenching criterion K [sdot] Le ≥ 1.5. Schmid (1995) also used a empirical H ₂-correction factor to adapt a model for the turbulent flame speed for hydrogen air flames. It can be shown that can be deduced by the Lewis number time scale modification, which is presented here (Kröner, 2003). Kurdymov et al. (Kurdymov et al., 2000) calculated numerically the flashback limits in the wall boundary layer for different fuel Lewis numbers. The criterion for the Karlovitz stretch factor 0.057 < K < 0.085 (based on a chemical time scale and the wall boundary velocity gradient) was proposed. Interestingly, the flashback limits of Kurdymov et al. can be deduced to a fuel independent limit of K = 0.068 with the time scale modification presented here.

⁷The fuel is treated as a homogeneous gas.

⁸For this study, small variations due to different air excess ratios are negligible.

⁹Transforming Eq. (11) to τ _PSR  = C _quench ∗d/ū [sdot] Le, for a given flow rate (ū) the critical time scale can be calculated. With help of Figures 15–18 the critical air excess ratio can be determined.

¹⁰As Reynolds numbers Re < 55.000 are not relevant for technical combustion systems, the isothermal velocity profiles at have not been determined at those flow rates.