2. Experimental methods
2.1. Nozzle, model combustor, operational conditions
The generic nozzle shown in Fig. 1 was based on a design by TURBOMECA. An axi-symmetric
gaseous fuel jet (Ø = 4 mm) penetrated into a con- centric 30 mm wide swirling combustion air flow that was preheated electrically (623 K). The rounding-off at the outer rim of the original nozzle was replaced by right-angled edges as obvious from Fig. 1. Although this measure changes flu- id-dynamics significantly, computational meshes underlying any numerical simulation can reproduce this simplified geometry more precisely. Swirl in the air flow was generated by tangential vanes. The theoretical swirl number using geometrical design parameters was 1.2. The nozzle was operated close to the lean blow-off limit at three different combustor pressures as reported in Table 1. Although desirable, operation at 1 bar was impeded due to too large flame lift-off heights resulting in extinction. Air mass flow rates were scaled according to a pressure drop of Dp/p = 3% across the nozzle resulting in identical bulk velocities for all operational conditions. Residence times in the region of interest were in the order of a few ms which is comparable to realistic conditions. More experimental details and corresponding non-reacting flow conditions are reported in [34,35].
The optically accessible modular combustor provided boundary conditions similar to gas tur- bines. As detailed in [34,35], it consisted of a dou- ble-walled air cooled flame tube (Ø = 120 mm) encased in a pressure vessel. No film cooling inside the flame tube was necessary that may have dis- turbed flow and scalar fields. Dried and filtered air from two screw compressors (Atlas Copco) was split into combustion and cooling air at a ratio of 1:3. Cooling air was necessary to chill the quartz plates used for the optical access. Downstream it was mixed into the exhaust. Using a back pressure valve in the exhaust duct the