Body Text
Research
>> back
A schematic of the optical arrangement used in these experiments is given above. While the frequency-doubled 532 nm output of two Nd:YAG lasers can be used for the particle imaging for PIV, the OH fluorescence is excited by the 285 nm, frequency-doubled output of an Nd:YAG-pumped dye laser, as shown. The OH fluorescence is imaged on an ICCD array, while the double-exposed PIV images are imaged from a spinning mirror onto a 2K x 2K CCD array.
Simultaneously acquired vector and OH PLIF fields are shown above (a and b) for a low Reynolds number flame. These images show that this flow is characterized by a small range of velocity fluctuation scales and laminar-like reaction zones and that the high velocity jet fluid is essentially bounded by each side of the flame. It is also observed that there exists a high level of shear across the reaction zones that is more clearly seen by considering the normalized vorticity contour plot above (c), where the vorticity is reasonably well correlated with the reaction zone location. The high vorticity does not imply that vortical eddies exist within the reaction zone; rather, the high vorticity is related to the shear across the reaction zone. The magnitude of the vorticity near the reaction zones is an order of magnitude larger than the outer-scale frequency. This results from the fact that nearly the entire velocity difference of the jet is taken across the relatively thin reaction zones.
The magnitude and direction of the minimum principal strain rates (S') are also shown above (d and e). Note that the majority of the strain field is compressive (negative). It is seen that along the reaction zones, the principal compressive strain rates are primarily aligned at 45� to the flow direction. The preferred direction of principal strain is expected given the high shear across the reaction zone (as inferred from the vorticity contours of c). The reaction zone, which is primarily vertical, is not typically aligned normal to the principal compressive strain. This lack of alignment is a characteristic of laminar flows where the stabilizing effect of viscosity does not allow material elements to freely adjust to the applied strain.
Similar velocity and OH fields for a more turbulent jet flame are shown above, where the downstream location and field-of-view are identical to the previous case. Here we see that the high velocity flow also appears to be largely bounded by the reaction zones, although this is less clear than with the lower Reynolds number case. As was also noted in previous work in round turbulent jet flames, the OH zones reveal larger diffuse regions typically connected by thin OH regions.
The vorticity field (c) shows that vorticity is again fairly well correlated with the general reaction zone, although the correlation is not as strong as at the lower Reynolds number. Similar to the lower Reynolds number case, the maximum values of the vorticity are an order of magnitude larger than the outer-scale frequency. These large vorticity magnitude regions are more commonly associated with thick, rather than thin, OH zones.
Examination of the principal compressive strain rate field (d and e) reveals that, similar to the low Reynolds number case, the large shear across the reaction zone results in principal compressive strain axes that are primarily aligned at 45� to the flow direction. However, now the thinnest reaction zones (e.g. center-left) are now aligned normal to the direction of the principal compressive strain, in addition to being associated with high values of compressive strain. In general, the shapes of the reaction zones and high strain regions differ since strain rate variations occur throughout the turbulent flow field (regardless of stoichiometry) whereas the reaction zone location is restricted to lie along the stoichiometric surface. Although not shown, our results also indicate that the thin OH regions are under extensive strain in a direction parallel to the reaction sheet. Another interesting feature of the above figure is that the diffuse OH regions are often oriented at about 90� with respect to the thin regions, thus they are aligned parallel to the direction of the principal compressive strain or normal to the extensive strain. This extensive strain, diffusion and possibly small-scale turbulence, broaden the reaction zone.
Superimposing reaction field data and vorticity/strain fields clarifies the interactions described above. A highly strained reaction zone with local extinction is shown in the above figure, where the reaction zone lies between the ambient fluid at left and jet fluid at right. Flow conditions and location are the same as the previous case. The thin OH regions are a result of alignment normal to the principal compressive strain (approximately 45� to the flow direction, while the thicker regions are not in alignment. The above seems to explain a common observation of OH fields in turbulent jet flames: that they tend to exhibit an alternating pattern of thick and thin regions often aligned at 90� with respect to each other. This is apparently due to the interaction of the reaction zones with the preferred direction of the principal compressive strain that acts at 45�. If a certain part of the reaction zone is acted on by a particularly large compressive strain, then the reaction zone will become oriented normal to it. Since the entire reaction zone cannot become aligned at 45� (since the flame must always follow the stoichiometric surface), the OH regions upstream and downstream must become oriented at approximately 90� to the thin regions, or parallel to the compressive strain and normal to the extensive strain. This alignment of the reaction zones normal to the extensive strain then causes the observed broadening.
Vorticity contours (b) illustrate a general correlation of vorticity with reaction zone location, although significantly lower values of vorticity occur in the thinnest OH regions. It appears that the presence of the flame produces the strong correlation, since nonreacting flow DNS has shown that there is not a strong correlation between regions of high scalar dissipation and vorticity. The reason for the lower vorticity in the thin regions of high principal compressive strain (c) is probably due to the realignment of the vorticity.
A convincing example of the relationship between compressive strain and reaction zone thickness is shown in (c), where the highest levels of compressive strain lie directly over the thinnest OH regions. The increased strain results in a decrease in OH mole fraction and the region labeled as (i) has undergone local extinction. This region is shown enlarged (d), with the superimposed compressive strain direction.
Ongoing Work
The focus of the current work is to quantify the observations described above using statistical analysis. For example, the relationship between compressive strain magnitude (negative strain within the OH zone) and the alignment of the reaction zone surface to the principal strain direction is illustrated with the joint PDF shown above. The data represent over 8000 points inside the OH zones. The angle shown (alpha) is the angle between the minimum principal strain direction and the unit normal to the reaction surface. Several distinct trends are immediately obvious. First, the majority of the minimum principal strain within the reaction zone is compressive (negative), which seems to suggest that the flame surface may generally have a sheet-like topology. The joint PDF indicates that the median strain within the reaction zone is around the value of the outer scale strain regardless of reaction zone alignment. However, when compressive strain rate is high, the reaction surfaces tend to be aligned normal to the strain direction (cos alpha ~ 1) . This alignment produces the observed thinning of the OH regions, sometimes resulting in local extinciton.
>> back
Prashanth S. Kothnur, Jason E. Rehm and Noel T. Clemens
Theoretical studies of turbulent nonpremixed flames have suggested that the structure of the reaction zone is strongly coupled to the underlying strain rate field, owing to the influence of fluctuating strain on the scalar dissipation rate. However, few experimental studies exist of the strain/reaction zone relationship in turbulent reacting flows due to the difficulty of making such measurements. Most of what is known about this relationship is inferred from computational studies and measurements in nonreacting turbulent flows. Measurements of the strain/reaction zone relationship require the simultaneous measurement of the velocity field and a suitable flame radical.
The objective of the present study is to directly investigate the relationship between the reaction zone structure and the underlying strain and vorticity fields in planar turbulent nonpremixed jet flames using simultaneous PIV and PLIF of the OH radical. The PIV provides planar fields of two components of velocity, while the OH PLIF provides an approximate marker of the instantaneous reaction zone.
The magnitude and direction of the minimum principal strain rates (S') are also shown above (d and e). Note that the majority of the strain field is compressive (negative). It is seen that along the reaction zones, the principal compressive strain rates are primarily aligned at 45� to the flow direction. The preferred direction of principal strain is expected given the high shear across the reaction zone (as inferred from the vorticity contours of c). The reaction zone, which is primarily vertical, is not typically aligned normal to the principal compressive strain. This lack of alignment is a characteristic of laminar flows where the stabilizing effect of viscosity does not allow material elements to freely adjust to the applied strain.
The vorticity field (c) shows that vorticity is again fairly well correlated with the general reaction zone, although the correlation is not as strong as at the lower Reynolds number. Similar to the lower Reynolds number case, the maximum values of the vorticity are an order of magnitude larger than the outer-scale frequency. These large vorticity magnitude regions are more commonly associated with thick, rather than thin, OH zones.
Examination of the principal compressive strain rate field (d and e) reveals that, similar to the low Reynolds number case, the large shear across the reaction zone results in principal compressive strain axes that are primarily aligned at 45� to the flow direction. However, now the thinnest reaction zones (e.g. center-left) are now aligned normal to the direction of the principal compressive strain, in addition to being associated with high values of compressive strain. In general, the shapes of the reaction zones and high strain regions differ since strain rate variations occur throughout the turbulent flow field (regardless of stoichiometry) whereas the reaction zone location is restricted to lie along the stoichiometric surface. Although not shown, our results also indicate that the thin OH regions are under extensive strain in a direction parallel to the reaction sheet. Another interesting feature of the above figure is that the diffuse OH regions are often oriented at about 90� with respect to the thin regions, thus they are aligned parallel to the direction of the principal compressive strain or normal to the extensive strain. This extensive strain, diffusion and possibly small-scale turbulence, broaden the reaction zone.
Vorticity contours (b) illustrate a general correlation of vorticity with reaction zone location, although significantly lower values of vorticity occur in the thinnest OH regions. It appears that the presence of the flame produces the strong correlation, since nonreacting flow DNS has shown that there is not a strong correlation between regions of high scalar dissipation and vorticity. The reason for the lower vorticity in the thin regions of high principal compressive strain (c) is probably due to the realignment of the vorticity.
A convincing example of the relationship between compressive strain and reaction zone thickness is shown in (c), where the highest levels of compressive strain lie directly over the thinnest OH regions. The increased strain results in a decrease in OH mole fraction and the region labeled as (i) has undergone local extinction. This region is shown enlarged (d), with the superimposed compressive strain direction.
The focus of the current work is to quantify the observations described above using statistical analysis. For example, the relationship between compressive strain magnitude (negative strain within the OH zone) and the alignment of the reaction zone surface to the principal strain direction is illustrated with the joint PDF shown above. The data represent over 8000 points inside the OH zones. The angle shown (alpha) is the angle between the minimum principal strain direction and the unit normal to the reaction surface. Several distinct trends are immediately obvious. First, the majority of the minimum principal strain within the reaction zone is compressive (negative), which seems to suggest that the flame surface may generally have a sheet-like topology. The joint PDF indicates that the median strain within the reaction zone is around the value of the outer scale strain regardless of reaction zone alignment. However, when compressive strain rate is high, the reaction surfaces tend to be aligned normal to the strain direction (cos alpha ~ 1) . This alignment produces the observed thinning of the OH regions, sometimes resulting in local extinciton.
Related work involving the large-scale structure of planar jet flames is also described on this site.
Publications:
- The Association of Scalar Dissipation Rate Layers and OH Zones with Strain, Vorticity, and 2-D Dilatation Fields in Turbulent Nonpremixed Jets and Jet Flames , J. E. Rehm & N. T. Clemens, AIAA-99-0676, 37th Aerospace Sciences Meeting, 1999.
- The relationship between vorticity/strain and reaction zone structure in turbulent nonpremixed jet flames, Rehm, J. E. and Clemens, N.T., 27th International Symposium on Combustion, 1998.
- An improved method for enhancing resolution of conventional double-exposure single-frame particle image velocimetry, Rehm, J. E. and Clemens, N.T., Experiments in Fluids, Vol. 26, pp. 497-504, 1999.
- The Large-Scale Turbulent Structure of Nonpremixed Planar Jet Flames, Rehm, J. E. and Clemens, N.T., Combustion and Flame, Vol. 116, pp. 615-626, 1999.
- Effects of heat release on the large-scale turbulent structure of planar jet diffusion flames, Rehm, J.E. and Clemens, N.T., Paper WS-97S-053, Western States Section of the Combustion Institute, 1997.
- A PIV/PLIF investigation of turbulent planar non-premixed flames, Rehm, J.E. and Clemens, N.T., AIAA 97-0250, 35th Aerospace Sciences Meeting, 1997.
- The flow structure of planar hydrogen diffusion flames, Rehm, J.E. and Clemens, N.T., AIAA 96-0704, 34th Aerospace Sciences Meeting, 1996.
>> back