We present a data-driven method for the early detection of thermoacoustic instabilities. Recurrence quantification analysis is used to calculate characteristic combustion features from short-length time series of dynamic pressure sensor data. Features like recurrence rate are used to train support vector machines to detect the onset of instability a few hundred milliseconds in advance. The performance of the proposed method is investigated on experimental data from a representative LOX/H${}_2$ research thrust chamber. In most cases, the method is able to timely predict two types of thermoacoustic instabilities on test data not used for training. The results are compared with state-of-the-art early warning indicators.

Two types of combustion instabilities were observed in a multi-injector research combustor with liquid oxygen and liquefied natural gas (LOX/LNG) propellants operating at conditions similar to an u...

Hot fire tests were performed using a single-injector research combustor featuring a large optical access window ([ ] mm) for flame visualisation. Three test campaigns were conducted with the propellant combination of liquid oxygen and hydrogen, liquid oxygen and compressed-natural-gas, as well as liquid oxygen and liquefied-natural-gas at conditions relevant for main- and upper-stage rocket engines. The large optical access enabled synchronised flame imaging using ultraviolet and blue radiation wavelengths covering a large portion of the combustion chamber for various sets of sub- and supercritical operating conditions. Combined with temperature, pressure and unsteady pressure measurements, this data provides a high-quality basis for the validation of numerical modelling. Flame width, length and opening angle as features describing the flame topology were extracted from the imaging. The suitability of flame imaging using ultraviolet and blue radiation wavelengths as qualitative markers of heat release was evaluated. Two-dimensional distributions of the Rayleigh Index were calculated for intervals with and without high-amplitude, self-excited oscillations of the longitudinal acoustic resonance modes. The calculated Rayleigh Index values properly reflect the thermoacoustic state of the chamber, indicating that both types of imaging may be used for qualitative study of thermoacoustic coupling of high-pressure cryogenic flames.

Hot-fire tests were performed with a single-injector research combustor featuring a large optical access (255 × 38 mm) for flame imaging. These tests were conducted with the propellant combination of liquid oxygen and compressed natural gas (LOX/CNG) at conditions relevant for main- and upper-stage engines. The large optical access enabled synchronized flame imaging using OH* and CH* radiation wavelengths covering an area of the combustion chamber from the injection plane to shortly before the contraction section of the nozzle for two sets of operating conditions. Combined with temperature, pressure and unsteady pressure measurements, these data provide a high-quality basis for validation of numerical modeling. Flame width and opening angle were extracted from the imaging in order to determine the flame topology. A two dimensional Rayleigh Index was calculated for an acoustically unexcited and excited interval. These Rayleigh Indices are in good agreement with the thermoacoustic state of the chamber.

Hot fire tests of a multi-injector research combustor were performed with liquid-oxygen and liquefied-natural-gas (LOX/LNG) propellants at chamber pressures from 30 up to 67 bar, hence at conditions similar to an upper stage rocket engine. Within these tests shear coaxial injectors were tested with and without a recessed LOX post. In both configurations, operating conditions with flames anchored at the LOX post tip and thus, if available, pre-combustion in the recess volume as well as lifted flames were observed. Flame anchoring was indirectly detected via acoustic measurements, using mean speed of sound to indicate the presence of flame in the head end of the combustion chamber. While the injector without recess showed only stable combustion irrespective of the flame anchoring behavior, the recessed injector featured short-lived bursts of oscillatory combustion and sustained combustion instabilities. Analysis of the test data showed that stable flame anchoring could not be ensured at momentum flux ratios below 20 for a non-recessed and below 45 for a recessed injector.