4. Conclusions
This study reports on optical measurements in a lifted confined natural gasair flame. A basic intention is to generate a data base of a turbulent flame much closer to practical conditions than
laboratory flames like piloted jet or opposed jet flames. Aside of a general understanding this data is suitable to validate numerical simulations. Underlying models are often tested just for ‘‘simple’’ turbulent flames but their applicability to practical conditions needs to be confirmed.
Preheating of the combustion air, nozzle dimensions and geometry mimic gas turbine con- ditions. Striking differences to a real combustor are nevertheless present. They are caused by con- straints of the laser diagnostics and by the degree of measurement accuracy that needs to be met to be useful for validation purposes.
The turbulent flow field was investigated by means of LDV. 1st and 2nd statistical moments identify the locations of recirculation zones, stag- nation point, and intense shear layers. Additional measurement of autocorrelation functions reveal coherent structures superimposed to turbulent fluctuations. These coherent motions can be inter- preted as vortex filaments and precessing vortex cores. Their amplitudes depend strongly on the location in the flow field. Because the amplitudes are very high at locations that are important for fuel–air mixing and flame stabilization, any numerical simulation needs to reproduce these coherent motions suitably.
The flame structure is investigated in terms of planar OH LIF. The liftoff height at centreline coincides with the location of the mean stagnation point where stoichiometric conditions are very probable. OH contours are deduced from their planar distributions and are used to obtain a mean reaction progress and Rxy. In the limit of the finite spatial resolution in the LIF experiments, the increase of Rxy with increasing combustor pressure is caused here by the flame brush shape and not by increased wrinkling. It is important to note that this is not a general statement as the spatial resolution is much larger than Batchelor scales. The shift of the mean flame stabilization towards the nozzle with increasing pressure is explained by the higher Rxy causing a larger reaction rate.