increased quenching distances δq. Both measures reduce the critical velocity gradient (Eq. 1.8) and, thus, increase the resistance against wall boundary layer FB. Baumgartner and Sattelmayer [69] conducted an experimental investigation regarding the effect of varied dilution air mass flows and injection angles and reported significantly extended FB limits for premixed hydrogen–air mixtures. Accordingly, air dilution in the premixing section is also applied in the burner setup of the current thesis.
Swirl number reduction
As discussed in the previous section, swirl is imposed on the flow in order to achieve fuel–air mixing and create a central recirculation zone that provides low flow velocities for flame anchoring. Between a swirl number too low for the occurrence of VB and a swirl number so high, that it requires an unaffordable pressure loss, a certain margin for swirl number variation exists. In case of incomplete mixing, the high flame temperatures associated to rich pockets of hydrogen would lead to increased NOx emissions. This poses an argument in favor of a high swirl number for hydrogen combustion. An argument in favor of a lower swirl number is the reported increased FB resistance. Sayad et al. [53]
reported a significantly extended operational range when decreasing the swirl number from S = 0.66 to S = 0.53 for a generic swirl burner operating on syn gases containing up to 80 vol.-% hydrogen. With decreasing swirl number, the swirling jet opening angle downstream of the mixing tube outlet was also reduced. This reduction in jet opening angle was previously reported by Terhaar et al. [70] and Reichel et al. [2] to also occur when the swirl number was reduced due to increasing injection rates of a non-swirling central air jet. Similar to the swirl number reduction of Sayad et al., the non-swirling air jet also increased FB resistance. This is reasonable, since a decreasing jet discharge angle reduces the area consumed by the mixture downstream of the mixing tube and leads to higher axial velocities.
The same tendency of increased FB resistance with reduced swirl number was also reported by Konle and Sattelmayer [66], when they conducted experimental investigations on a swirl-stabilized premix burner at atmospheric conditions firing natural gas. Moreover, they report velocity measurements in the premixing section at isothermal and reacting conditions. At isothermal conditions, the axial velocity profile 0.5D upstream of the mixing tube outlet exhibits the typical axial velocity deficit found in strongly swirling flows. At reacting conditions, although the flame is stabilized downstream of the mixing tube outlet, the axial velocity on the central axis is further reduced. Simultaneously, on a higher radius, r/D= 0.2, a velocity increase is observed. This observed difference in isothermal and reacting velocity field in the mixing tube is an important mechanism that needs to be taken into account, when judging FB resistance of combustor geometries based on isothermal flow fields.
1.6 Controlling Parameters for Flashback Prevention 29
An opposite trend in the correlation of swirl number with respect to FB is observed by Syred et al. [52]. They report increased FB resistance with increasing swirl number in the range ofS = 0.8–1.47 from their investigation of blow off and FB limits of a generic swirl burner firing high-hydrogen content coke oven gas (65% H2 , 25% CH4, 6% CO, 4% N2).
In this case the complex interaction of partially premixed fuel injection with the flow field seems to contribute to this somewhat unexpected correlation.
Obviously, the definition of a swirl number becomes to some extent ambiguous, when the high volumetric fuel flow rates associated to hydrogen-rich fuels alter flow field characteristics like the swirl number. Since this important effect is under investigated and not fully understood for partially premixed, swirling combustor flows further light will be shed on the correlation of swirl number and FB with respect to varying AI rates and fuel flow rates throughout this thesis.
Non-swirling central air jet (axial air injection)
In order to avoid FB of the types three and four, turbulent flame propagation in the core flow and combustion induced vortex breakdown, “a major design criterion for nozzle aerodynamics is that the axial velocity must be as high and as uniform as possible and free of strong wakes” [34]. The applications of non-swirling central air jets in swirling combustor flows had been reported previously, e.g., by McVey et al. [71]. However, Burmberger and Sattelmayer [72] first suggested its application to influence the position of vortex breakdown aiming to increase FB resistance of high reactivity fuels. They suggest that, the mechanism by which a non-swirling central air jet delays VB and, thus increases FB resistance, is is linked to the axial gradient of the azimuthal vorticity∂ωφ/∂z, here expressed in cylindrical
a) b)
Figure 1.10: Effect of azimuthal vorticity on streamline divergence: a) decreasing az-imuthal vorticity causes streamline divergence (ur >0); b) increasing azimuthal vorticity prevents streamline divergence (ur<0)
coordinates. In a swirling flow without AI, the absolute value of the azimuthal vorticity ωφdecreases in flow direction, i.e. ∂ωφ/∂z <0. This causes a declining axial velocity on the axis of rotation which results in streamline divergence (ur >0), as is illustrated in Fig. 1.10a. The streamline divergence may trigger the isothermal vortex breakdown to occur upstream of the combustion chamber leading to flame FB. A non-swirling axial air jet delays the streamline divergence and shifts the location of VB downstream (Fig. 1.10b).
Burmberger et al. [73] proved that a burner, applying this concept, allows for FB-free operation of stoichiometric methane-diluted hydrogen mixtures under unconfined conditions. Besides the use of diluted hydrogen and unconfined flames, the study is also limited to perfectly premixed combustion and omits the challenges related to achieving FB safety as well as low NOx emissions for pure hydrogen in a technically premixed case.
Previous investigations revealed that, in the presence of a high momentum [74,75,76] or low momentum [70] non-swirling jet the flow field is less prone to exhibit self-excited flow oscillations. Such self-excited flow instabilities were previously observed by Paschereit et al. [77] in swirling combustor flows for both, isothermal and reacting conditions. Galley et al. [78] report a hydrodynamic instability to trap the fuel in a precessing vortex core and lead to strong temporal fuel concentration fluctuations. Consequently, the suppression of a hydrodynamic instability, could increase the temporal mixing quality. This is identified as one possible interaction mechanism between axial injection of air and fuel–air mixing quality.
Terhaar et al. [70] recorded isothermal and reacting flow field data for methane in the presence of varied amounts of axial air injection, which were controlled by a mass flow meter. They present a correlation where, depending on the combination of primary swirl number and amount of axial air injection, the hydrodynamic instability is suppressed.
This suppression is observed to coincide with a change in VB type from bubble type to cone type (compare Billant et al. [79]). Whereas the bubble type VB exhibits a local minimum in axial velocity in the plane upstream of VB, the cone type VB does not and is therefore preferred for FB safety.
Depending on the initial swirl number, i.e., swirl in the absence of axial injection, above a certain AI rate, VB is suppressed as reported by Terhaar et al. [70]. Therefore, the use of a central air jet is limited as such, that for the sake of flame stability, a sufficiently large inner recirculation zone has to be preserved at all times. Another potential limiting factor of AI is the fuel–air mixing quality. Application of a central non-swirling jet alters the flow field, reduces the resulting swirl number and, thus, potentially reduces fuel–air mixing quality. However, other mechanisms like the suppression of a hydrodynamic instability could positively affect fuel–air mixing. The net impact of varied AI rates on fuel–air mixing is currently not understood and therefore subject of the current thesis.