1 Introduction

The development of modern combustion devices such as IC engines and gas turbines is towards higher power densities and therefore smaller surface area to volume ratios. Thus, flame-wall interaction (FWI) is of increasing importance in this context. Combustor walls are exposed to extremely high temperatures which eventually leads to fatigue failure of the material. However, combustor walls are cooled causing enthalpy loss in the combustion process. This leads to reduced efficiency and incomplete combustion which results in the formation of pollutants, e.g. CO [1, 2].

FWI strongly couples processes of flow field, flame chemistry and wall properties [3]. Flames close to a wall get quenched eventually due to enthalpy losses. FWI generally has two canonical configurations: Side-wall quenching is defined by wallparallel flame propagation; the configuration investigated in this work, head-on quenching (HOQ), is characterized by wallnormal flame propagation.

This study focuses on transient turbulent FWI of premixed flames in the HOQ configuration. The impact of a convective flow field and fuel type on FWI is investigated as it is particularly relevant in IC engines, where the flame partly impinges orthogonally on enclosing walls.

Laminar premixed flames in transient FWI of HOQ used to be investigated using simple fuels, such as CH4. Reference [4] experimentally analyzes flame structures and velocities in laminar HOQ; however the fresh gases are initially quiescent different to the flow of this study. References [5–7] use the same burner as in the present study to experimentally characterize thermokinetics, heat transfer and flame structures during HOQ. In these studies the flame is ignited in a laminar convective flow field. Numerical simulations are usually limited to 1-D calculations with simplified chemistry, e.g. [8, 9], or are a part of turbulent HOQ studies, e.g. [10, 11]. References [9, 12, 13] use detailed chemistry of different fuels in 1-D simulations of laminar HOQ, focusing on species’ concentrations, heat flux and wall temperature. Reference [13] focuses on quenching and pressure influence for various fuels.

Few studies on turbulent premixed flames in transient HOQ exist. References [14–16] use a constant volume vessel to experimentally investigate velocities and heat flux during FWI of a premixed CH4 flame. The fresh gases are initially in a tumble motion. Using simulations, ref. [10] applies 2-D DNS to analyze quenching distances, flame propagation and heat flux during transient HOQ of a premixed flame in a synthetic turbulent flow field. Reference [11] studies the effect of different Lewis numbers on quenching distances and turbulent flame lengths during transient turbulent HOQ with the aid of 2-D DNS. Reference [17] uses a 3-D DNS in a channel flow to focus on flame-flow interaction during HOQ. In general, these 2-D and 3-D DNSs are all based on simplified chemistry and synthetic turbulence.

This work has two major objectives: to analyze the impact of a turbulent convective flow field and to investigate the influence of various fuels on transient HOQ of premixed turbulent flames. Neither of these objectives has yet been subject to any experimental study. Unlike existing numerical studies this experimental work is restricted to weakly wrinkled flames but is not limited to synthetic turbulence or simplified chemistry. For the experimental investigation cinematographic stereo-PIV (to measure velocities) and planar OH-LIF (to simultaneously track flame fronts) are combined. Selected snapshots of these time-resolved measurements enable qualitative description of the transient FWI of individual quenching events. Characteristic properties of turbulent flames in the context of FWI, such as turbulent flame length, fresh gas velocity fluctuations and flame propagation are quantified.

2 Experiment Set-Up

2.1 Burner configuration

Transient, premixed turbulent flame-wall interaction in the head-on quenching regime is realized using a nozzle type burner (Fig. 1). A laser spark at the nozzle exit, r = 0 mm, z = 27 mm, ignites premixed air and fuel (refer to [6] for more details). A flame then propagates orthogonally against a steel wall, where it eventually gets quenched due to the enthalpy losses. The water-cooled wall is spherical-shaped (R = 300 mm) and located at z = 0 mm at a distance of 32 mm from the nozzle exit. The burner used in this study is the same as that used in ref. [5–7, 18]. Figure 1 details the burner including the turbulence grid (TG) used. The TG enhances the turbulence level to \(\frac {u^{\prime }}{u}\approx 6-7~\%\) (u is the bulk flow velocity and u′ is the axial velocity fluctuations at the nozzle exit). The integral length scale at the nozzle exit equals l i n t ≈ 4.7 mm (data taken from ref. [6]). All gas flows are controlled by electronic mass flow meters (Bronkhorst EL-Flow series). Fuel, methane (CH4) or ethylene (C2H4), and air are mixed well and fed into the burner at room temperature and atmospheric pressure. Fine meshes in the lower section of the burner homogenize the flow. The premixed jet exits the burner with a bulk flow velocity of approximately 2.62m/s. This corresponds to a Reynolds number of R e = 5000 based on the fluid properties of the unburnt mixture at 273 K, 1 atm and the nozzle exit diameter of d = 30 mm. A nitrogen coflow with a bulk flow velocity of approximately 1.31 m/s (50 % of the jet velocity) surrounds the premixed jet to shield it from external influences. Operating conditions and relevant parameters are summarized in Table 1. Based on estimations of \(\frac {u^{\prime }}{s_{l}}<1\) and \(\frac {l_{\textit {int}}}{\delta _{\textit {flame}}}= 82\)(CH4), 162(C2H4), premixed flames of both fuels are located in the wrinkled regime as in ref. [19]. The origin of a Cartesian coordinate system (detail of Fig. 1) is defined at the stagnation point of the jet in the measurement plane. The central axis z is aligned with the nozzle’s main axis; the radial axis r is perpendicular to z and the tangential coordinate t points into the plane of Fig. 1.

Fig. 1
figure 1

Sketch of the FWI impinging jet burner with turbulence grid. Dimensions are given in mm. For further details see text and refer to [5–7, 18]. The field of view (FOV) for the PIV and OH measurement in the central plane of the burner and the coordinate system are illustrated in the dashed box

Full size image
Table 1 Operating conditions for transient flame-wall interaction
Full size table

2.2 Combined OH-PLIF and PIV

Flame-wall interaction is characterized cinematographically at 10 kHz recording rate using stereoscopic PIV to measure velocities and simultaneous planar OH-LIF to track flame fronts. These techniques are applied simultaneously in the central plane of the nozzle (Fig. 1). For PIV measurements the premixed jet flow is seeded with Al2O3 particles (∅1 μ m) and illuminated by a frequency-doubled Nd:YVO4 laser (λ = 532 nm, 4.0 mJ/pulse, pulse separation 100 μs, Edgewave IS 4II-DE) at 10 kHz repetition rate. OH fluorescence is excited at the Q1(6)-transition (λ = 282.93 nm, 0.25 mJ/pulse) using a frequency doubled high-speed dye laser (Sirah Credo, Rhodamine 6G) pumped at 10 kHz by an Nd:YAG laser (λ = 532 nm, 8 mJ/pulse, Edgewave CX 16II-E). Laser beams are formed into ∼100 μm thick light sheets and are combined into one imaging plane using a dichroic mirror (high-reflective at λ = 283 nm, high-transmissive at λ = 532 nm). The dye laser is fired in between the PIV laser pulses. For stereoscopic PIV the Mie-scattering of Al2O3 particles is imaged with 180 mm f/3.5 macro lenses and Scheimpflug adapters on high-speed CMOS cameras (LaVision HSS 6). Both PIV cameras have a field of view (FOV) of 17 × 17 mm2 and are arranged symmetrically to the laser light sheets with an angle of ∼30°. OH-LIF signals are imaged with a 150 mm UV lens (Halle OUC 3.60.4) on an image intensifier (LaVision HighSpeed IRO) coupled to another CMOS camera (LaVision HSS 5). This camera system, with a FOV of 16 × 16 mm2, is arranged perpendicularly to the laser light sheet plane.

The PIV data are calibrated using a stereoscopic PIV target and the OH data are calibrated using a 1 mm grid target. The two systems are matched based on images of the 1 mm grid target.

LaVision Davis PIV software is used to post-process the PIV data. Adaptive interrogation windows 24 × 24 pixels in size and with 75 % overlap are used to calculate the velocity fields, resulting in a vector spacing of 144 μm. Outliers are filtered with a 3x3 median filter. Flame front positions are extracted from the instantaneous OH images. The noise in the images is reduced using a non-linear diffusion filter, as suggested by [24] for application in turbulent flames and successfully applied by [5] to laminar flames in the same head-on quenching configuration as this paper. This filter, originally developed by [25], preserves and enhances the OH gradients while removing noise. The filtered OH image is then interpolated cubically by a factor of 4 to avoid any jumps in detection of the flame front. In the final step the flame front is extracted from the filtered images by detecting the steepest OH gradient with a Canny-Edge filter [26] in Matlab.

2.3 Data-acquisition, wall position and measurement uncertainties

For each operational condition of Table 1, 20 ignition events are realized. Each event is recorded as a sequence of 200 images, which is a recording length of 20 ms and is a compromise between necessary samples for statistical data evaluation and the complexity of the experimental procedure. All events cannot be considered for some evaluated parameters, since flame front detection sometimes fails due to laser and signal intensity fluctuations. This is especially the case for CH4 flames, as they generally have a lower signal-to-noise ratio of the OH-LIF than C2H4 flames.

The exact wall position is determined from the average intensity of the 200 PIV raw images per ignition event. In the average image, the reflections from the wall due to the laser light are clearly visible. A circle with radius R = 300 mm matching the shape of the curved wall is fitted to the reflections. The wall position is then obtained. In the final step the ensemble average of the determined wall positions is calculated for each fuel. The systematic error of this procedure corresponds to the standard deviation of the determined wall position, which is 160 μm in the radial direction r and 5 μm in the wall normal direction z.

As mentioned previously, the PIV and the OH systems are matched using a grid target with 1 mm line spacing. The OH image resolution is the limiting factor. Based on line thickness of the target the resolution is estimated as being 65 % inferior to that of the PIV image. The systematic matching error is estimated to be 82 μm based on the line positions. This is assumed as the dominant error source of this measurement compared to other possible uncertainties (e.g. time and spatial resolution).

2.4 Determination of quenching and spatial representation

Flame quenching is determined from the OH images based on the OH gradient. During wall approach, the OH gradient in vertical direction z is tracked for each flame element. When the flame senses the wall, the OH gradient decreases until it vanishes entirely. The instant of quenching is defined as being where the OH gradient, normalized to its maximum value far-off the wall, falls below 0.5. In ref. [27], this approach is successfully applied to side-wall quenching flames and validated by comparison to chemoluminescence imaging. Thus the quenching distance δ q u e n c h corresponds to the location z of this event corrected for the wall position at the specific radial location r. This correction is necessary due to the curvature of the wall (Fig. 1).

When analyzing the characteristic processes of transient FWI, spatial representation depending on the distance between the flame and the wall can suppress intermittence effects observed in a laboratory-fixed representation of turbulent flames. To achieve this, the instantaneous axial flame front position is sorted into radial Δr and wall normal Δz bins (Fig. 2). First, Fig. 2a, each instantaneous flame front is divided radially into bins of Δr = 500 μm. Next, in each bin, the wall normal flame position is corrected for the wall position and a mean flame position z F F is calculated. The spatial representation is obtained by sorting z F F and flame properties, such as the turbulent flame length L t at this position into wall normal bins of Δz = 500 μm (Fig. 2b). This method has the drawback that in the case of highly wrinkled flames z F F is the average position of several flame elements in one radial bin Δr (compare Fig. 2a 6.5m m < r < 8 mm). However, this is only observed occasionally for CH4 flames. Far off the wall, z > 4 mm, 5 out of 20 CH4 ignition events exhibit this phenomena in certain radial bins Δr. At maximum only 4 % of all Δr per instantaneous flame front are concerned. Close to the wall, z < 4 mm, this phenomena does not occur any more. Further consequences are discussed for L t in Section 3.3.

Fig. 2
figure 2

Post-processing steps for spatial representation of FWI parameters, depending on the flame-wall distance

Full size image

The influence of the bin widths (Δrand Δz) on the behaviour of the quantities evaluated in this study has been analyzed and results (not shown) show similar behaviour independent of the size of Δ. Quenched flame elements are not considered in the statistical evaluation.

3 Results

3.1 Flame evolution

Transient FWI, leading to head-on quenching is discussed qualitatively in this section, based on Fig. 3. Instantaneous snapshots including the evolution of the flame front and the absolute velocity field \(\left | u \right |=\sqrt {u_{axial}^{2}+u_{radial}^{2}+u_{tangential}^{2}} \) of a single ignition event of CH4 a) and C2H4 b), are shown. Overall, the FWI process of the C2H4 flame is considerably faster than the CH4 flame because of the higher flame speed (Table 1).

Fig. 3
figure 3

Flame front (white line) and velocity field for CH4 (a) and C2H4 (b) at different time steps for a single quenching event. τ = 0 ms at ignition

Full size image

Regarding the velocity in the r-z-plane, the acceleration of the burnt gases is up to twice the absolute velocity of the fresh gases and the turnover of the vectors in the wall normal direction is due to the passage of the flame. The flame passage disturbs the typical stagnation point flow, with high radial velocities for |r| ≫ 0 mm.

In flame front evolution, the impact of the wall is obvious since the turbulent flame fronts are laminarized during wall approach, which confirms the DNS observations of [10] and [11]. Additionally, Fig. 3 reveals that at times τ = 9.8 ms for CH4 a) and τ = 5.9 ms for C2H4 b) quenching does not occur homogenously along the wall surface. It starts at particular points, corresponding usually to convex flame elements (not shown), and then propagates parallel to the wall until the whole flame is quenched entirely. This general experimental observation is in full agreement with the DNS-based explanations of [10]. Concave flame elements, e.g. Fig. 3b at τ = 5.9 ms remain relatively long at the wall, before they finally get quenched. Low flow field velocities and disturbed flame chemistry slow the consumption of the remaining gas thus causing this phenomenon. The CH4 flame is clearly more sensitive to flame-flow interactions inducing strong flame front wrinkling (Fig. 3a at τ = 3.8 and 5.8 ms). This can be related to the different Lewis numbers of the investigated fuels. For L e > 1 (C2H4) flame wrinkling due to external disturbances, such as velocity fluctuations u′/s l , vanishes quickly, whereas it remains or is amplified for L e ≤ 1 (CH4) [11, 23, 28, 29].

3.2 Quenching distances

An initial characteristic measure of flame quenching at a wall is the quenching distance δ q u e n c h . Non dimensionalizing δ q u e n c h by the flame thickness δ f l a m e (Table 1) leads to the quenching Peclet number P e. Table 2 summarizes the results obtained. The standard deviation of δ q u e n c h and P e is related to the turbulent nature of the FWI. Systematic errors are estimated and discussed in Section 2.3

Table 2 Mean quenching distances evaluated between r = [0–12] mm
Full size table

Reference [6] measured δ q u e n c h =170 μm for CH4 based on a temperature criterion on the same burner at otherwise identical operational conditions except for the laminar flow conditions. According to [10, 11] δ q u e n c h of turbulent and laminar flames is similar. This is the case for the turbulent CH4 flame of this study with δ q u e n c h =179 μm

Peclet numbers P e are in an expected range for head-on quenching [3]. However values for the two fuels are different, which contrasts to ref. [3, 13], who state that P e is independent of the fuel type for hydrocarbons. Based on DNS [11] states that smaller Lewis numbers lead to higher Peclet numbers. This was not observed in the experimental data of this work, which might be due to the different flame regime investigated here.

3.3 Flame length

Turbulent flame front length L t is an appropriate way to quantify laminarization of the flame front during wall approach (Fig. 3). L t is defined here as the length of each flame element in a radial bin Δr (compare Fig. 2a). The obtained L t are sorted according to the flame position z F F ; and the mean and standard deviation of L t in each bin Δz is calculated (compare Fig. 2b). The results are nondimensionalized by the bin width Δr, which is assumed to correspond to the unwrinkled laminar flame length L l . As mentioned in Section 2.4 this method has some drawbacks: For L t this means that several flame elements in one radial bin Δr contribute to L t in the case of intense flame wrinkling (compare Fig. 2a 6.5 mm < r < 8 mm). Nevertheless, this approach clearly differentiates between laminar and turbulent flames, which is the aim of this section. The problem could be circumvented using the approach of ref. [27]: The instantaneous L t is determined spatially in each equal-sized square of a grid. However, this approach is not feasible here, since the sample size is relatively low.

Figure 4 shows the evolution of L t as a function of the flame positionz. Depending on the fuel, different zones can be defined. For z > 3 mm FWI of CH4 is dominated by flame-turbulence interaction, and flame wrinkling is strongly damped during wall approach (compare Fig. 4a and b). Referencing [10], this zone is called “inhibition zone”. C2H4 does not exhibit such a zone, mainly due to the higher Lewis number, which causes the flame to be less wrinkled [11, 23, 29]. Between 1 < z < 3 mm L t shows similar behaviour for both fuels with almost constant L t . However, C2H4 features slightly higher fluctuation levels, although they remain at the same level as for z > 3 mm (compare Fig. 4b). Finally, for z < 1 mm L t decreases and approaches the laminar flame length L l . Referring to [10] this zone could be considered as the transition to the “influence zone”, where the flame starts sensing the wall and the quenching process is induced. Generally, the experimental observations confirm the DNS studies of [11], which state that during FWI higher Lewis numbers lead to less flame wrinkling and in certain cases to almost constant L t .

Fig. 4
figure 4

Normalized flame lengths for CH4 (-, 19/20 ignition events) and C2H4 (- -, 19/20 ignition events) as a function of the flame position z. a mean and b fluctuations evaluated for r ∈ [0–12] mm, with Δr =Δz = 500 μm

Full size image

3.4 Fluctuations

The velocity fluctuations of the fresh gases in the direction of flame propagation are analyzed to judge the influence of flame-flow interaction on transient head-on quenching processes. First, the instantaneous flame normal vector \(\vec {n}\) pointing towards the fresh gases is calculated in the r-z-plane for each flame front element in a radial bin Δr (Section 2.4 and Fig. 2). Second, the axial \(\vec {u}_{axial}\) and radial \(\vec {u}_{radial}\) velocity components are extracted at a distance of Δn = 1 mm ahead of the flame front to circumvent unknown density effects caused by the expanding flame [27]. The influence of Δn on the results of this section has been analyzed in a sensitivity study within a range of Δn∈[0.5,1.5]m m No major differences have been observed (not shown here). After projecting these components onto the flame normal direction \(\vec {n}\), the mean velocity is calculated in each bin Δr. The sum of the obtained components \(\vec {u}_{radial,n}\) and \(\vec {u}_{axial,n}\) is \(\vec {u}_{absolute,n}\) (Fig. 5). Next, the obtained values are sorted according to the vertical flame position z F F . The standard deviation in each axial bin Δz then results in the velocity fluctuations u′ r a d i a l,n , u′ a x i a l,n and u′ a b s o l u t e,n . Results are normalized by the laminar flame speed s l to compare the two fuels.

Fig. 5
figure 5

Sketch of the flame normal vector \(\vec {n}\) of a single flame front element and the related velocity vector components \(\vec {u}_{radial,n}~\vec {u}_{axial,n}\) and \(\vec {u}_{absolute,n}\)

Full size image

Figure 6 shows components of the fresh gas velocity fluctuations in the flame normal direction. The results clearly show that the absolute turbulence level of the fresh gas decreases during wall approach (Fig. 6a). Consequently, the wall damps turbulence in the fresh gas flow. The rate of turbulence decrease in axial direction (Fig. 6b) can be considered as constant for positions of z < 4m m. For the radial component turbulence decrease is only observed very close to the wall (Fig. 6c z < 1.75 mm for CH4 and z < 3 mm for C2H4).

Fig. 6
figure 6

Velocity fluctuations in the fresh gas in flame normal direction for CH 4 (–, 19/20 ignition events), C 2 H 4 (- -, 18/20 ignition events). Values are extracted at Δn = 1 mm ahead of the flame front and evaluated for Δr =Δz = 500 μm, r ∈ [0–12] mm. a absolute fluctuations, b axial component and c radial component

Full size image

While the axial component, Fig. 6b, shows similar behaviour for both fuels, the radial component Fig. 6c shows clear differences. The radial component is quite revealing since its amplitude for CH4 is inferior to that of C2H4. It is well known from turbulent flame theory and experimental studies, e.g. [22], that higher turbulence causes more flame wrinkling. However, we observe the opposite for the near wall flame. When neglecting the possible influence of the processing method for L t , one can therefore conclude that the higher wrinkling of the CH4 flame (compare Fig. 3a and 4a) is related to Lewis number effects as discussed in Section 3.1. Hence diffusion processes are suspected to be the reasons for stronger flame wrinkling [11, 23, 28, 29].

In the case of FWI, local values of u′/s l close to the instantaneous position of the flame rather than global values at the nozzle exit should be considered. The ratio u′/s l is influenced by the presence of the wall and can be significantly different compared to the unbounded flow. This might be the reason for discrepancies to observations from the previous work by Renou and Boukhalfa [22]. In contrast to the present findings, a higher Lewis number caused increased flame wrinkling at a similar turbulence level to this work (u′/s l ≈ 0.9). At a slightly higher turbulence level (u′/s l ≈ 1.3), however, an increasing Lewis number decreased the flame wrinkling which appears similar to the observations of this work.

3.5 Flame propagation

Flame propagation during FWI is important, since enthalpy losses to the wall strongly disturb the combustion process [3, 6]. Calculation of the turbulent flame displacement speed s t , e.g. in [30], from the experimental data is not feasible because of the low sample number and the 3-dimensional flame propagation. Since only a few studies have characterized transient head-on quenching in a convective flow, the evolution of the flame front position z F F as a function of time τ is a parameter of interest. Figure 7 shows the mean evolution of this parameter along the centreline of the burner at Δr = [0−0.5]m m. For each ignition event the instant of quenching at this radial position is determined and is set to τ = 0m s. This allows the calculation of an ensemble average of the flame position z F F as a function of time τ prior to quenching.

Fig. 7
figure 7

Flame position z F F over time τ (referenced to the instant of quenching) for CH 4 (–, 17/20 ignition events) and C 2 H 4 (- -, 19/20 ignition events) evaluated for r ∈ [0–0.5] mm. Error bars indicate the standard deviation of the flame position (for better readability only every second error bar is shown)

Full size image

The results of Fig. 7 quantify and confirm the qualitative findings of Fig. 3, stating that the C2H4 flame reaches the wall faster than the CH4 flame. The decreasing standard deviation of z F F in Fig. 7 clearly proves the laminarization of the flame front, especially for times τ > −2m s prior to quenching. For both fuels z F F (τ) has a parabolic shape with asymptotic behaviour close to the wall. The experimental results of z F F (τ) for turbulent flames from this study demonstrate a similar behaviour to the findings in [7] resulting from 1-D CARS thermometry measurements of a laminar CH4 flame at the same ϕ and conditions on the same burner. For turbulent flames the trend of the measured z F F (τ)corresponds well to the DNS observations in [10] and [11].

The absolute flame speed u F F in the r-z plane provides further insight into transient head-on quenching of premixed turbulent flames. The absolute flame speed u F F results from the superposition of the convective velocity of the unburnt gases and the local flame displacement speed s t [30]. Due to the relatively small sample number in this study, a time derivation of the curves in Fig. 7 would cause noisy u F F . Thus, a different procedure for determining u F F is used: First, the mean axial flame position z F F is calculated for each instantaneous flame front in Δr = [0−0.5]m m. This leads to a function z F F (τ), which is then derived in time for each ignition event. In the last step, d z F F /d τ = u F F is sorted according to z F F and the mean is calculated in each bin Δz (Section 2.4). Normalization to the undisturbed laminar flame speed s l allows better comparability of the two fuels. It is expected that u F F → 0 m/s at the wall. Due to collection into bins Δz = 500 μm, this is not observable in this study.

The results for u F F (z) including the standard deviation are shown in Fig. 8. Previous observations on laminarization and the faster FWI process of C2H4 (hidden here by the normalization to s l ) are observed again. Generally, far off the wall, for z > 1 mm, flame propagation is driven primarily by the convective flow field. Additionally, u F F strongly decreases at quasi-constant rates during wall approach. Since both fuels show the same normalized u F F differences in the overall propagation process are related to the reactive/diffusive properties of the fuels which is expressed through the laminar flame speed s l (Table 1). Consequently, the influence of convection on the flame propagation of both fuels is identical.

Fig. 8
figure 8

Absolute flame speed u F F normalized by laminar flame speed s l , including standard deviation for CH 4 (–, thin error bars, 17/20 ignition events), C 2 H 4 (- -, thick error bars, 19/20 ignition events) as a function of the distance to the wall. Evaluated for r ∈ [0–0.5] mm with Δr =Δz = 500 μm. For better readability only every second error bar is shown

Full size image

Close to the wall, for z < 1 mm, both fuels exhibit |u F F /s l |≤1. This again proves the transition to the influence zone (compare to Section 3.3). This observation strongly indicates a decrease in the flame displacement speed s t and changes in the laminar flame speed [10, 27].

Experimental quantification of u F F during HOQ has not yet been published in literature. Qualitative comparison of the results with various numerical simulations shows that they are in good agreement: [8] uses 1-D simulations to calculate the absolute flame speed for a laminar premixed flame during HOQ. However, the fresh gas is initially quiescent. In turbulent premixed HOQ, the absolute flame speed could be deducted from the flame position over time as in [10] and [11].

4 Conclusion

Flame-flow interactions (FWI) during transient head-on quenching of turbulent premixed CH4 and C2H4 flames were investigated. Qualitative observations based on selected instantaneous snapshots from the time series data of the velocity field and the flame position clearly show the different behaviour of the flame shape and the overall FWI process speed. However, both fuels show laminarization of flow and flame during wall approach. The quantitative findings of this work, such as the quenching distance δ q u e n c h and the turbulent flame length L t , are in good agreement with DNS calculations, which are restricted to simplified chemistry and synthetic turbulence and handle different flame regimes. This work is the first known quantification of fresh gas velocity fluctuations u′/s l in the context of FWI. u′/s l induces flame wrinkling which is expressed by L t . The different fuels reveal that it is primarily Lewis number effects not the amplitude of the flow field fluctuations, that mainly influence L t . Finally, flame propagation shows that convection dominates flame propagation far off the wall and differences are only related to different reactive/diffusive properties, expressed by the laminar flame speed of the fuel. Close to the wall the absolute flame speeds are inferior to the undisturbed laminar flame speeds and thus strongly indicate disturbances in the flame chemistry.