Abstract
Premixed or partially premixed swirling flames are widely used in gas turbine applications because of their compactness, high ignition efficiency, low NOx emissions and flame stability. A typical annular combustor consists of about twenty swirling flames, which interact (directly or indirectly) with their immediate neighbors even during stable operation. These interactions significantly alter the flow and flame topologies thereby bringing in some discrepancies between the single nozzle (SN) and multinozzle (MN), ignition, emission, pattern factor and flame transfer function (FTF) characteristics. For example, in MN configurations, application of a model based on SN FTF data could lead to erroneous conclusions. Due to the complexities involved in this problem in terms of size, thermal power, cost, optical accessibility etc., a limited amount of experimental studies has been reported, that too on scaled down models with reduced number of nozzles. Here, we present a detailed experimental study on the behavior of three interacting swirl premixed flames, arranged in-line in an optically accessible hollow cuboid test section, which closely resembles a three-cup sector of an annular gas turbine combustor with very large radius. Multiple configurations with various combinations of swirl levels between the adjacent nozzles and the associated flame and flow topologies have been studied. Spatio-temporal information of the heat release rate obtained from OH* chemiluminescence imaging is used along with the acoustic pressure signatures to compute the Rayleigh index (RI) so as to identify the regions within the flame that pumps energy into the self-excited thermoacoustic instability modes. It is found that the structure of the flame–flame interaction regions plays a dominant role in the resulting thermoacoustic instability. To resolve the flow and reactive species distributions in the interacting flames, two-dimensional (2D), three component stereoscopic particle image velocimetry (SPIV) and planar laser-induced fluorescence (PLIF) of hydroxyl radical is applied to all the test conditions. Significant differences in the flow structures among the different configurations were observed. Simultaneous OH-PLIF and SPIV techniques were also utilized to track the flame front, from which the curvature and stretch rates were computed. Flame surface density (FSD) which is defined as the mean surface area of the reaction zone per unit volume, is also computed for all the test cases. These measurements and analyses elucidate the structure of the interaction regions, their unique characteristics, and possible role in thermoacoustic instability.