No abstract

The prognosis of thermo-acoustic/combustion instability is usually accomplished by applying a priori knowledge about features of unstable operation and measuring deviation from those features using point values. In the present work, we adopt a different methodology, whereby the presence and extent of the signature of unstable combustion are learnt as an anomaly from the distribution of pressure oscillations during stable operation across several protocols. The protocols involve a transition from stable to unstable operation in a swirl combustor. It is inferred that the stable combustion in the present case is stochastic noise with a normal distribution containing values comparable with root-mean-square values at unstable operation with a [Formula: see text] value 0.05–0.07. We exploit this feature to detect anomalies from flame intensity images, which represents the heat release rate fluctuations by manipulating their features to be a part of multivariate Gaussian distribution. To formulate this distribution, we employ a convolutional-neural-network-based variational auto-encoder (CNN-VAE) and express the associated reconstruction loss as an anomaly metric. The anomalies obtained through CNN-VAE and integrated intensity fluctuations are then evaluated for their sensitivity against the unsteady pressure data. The analysis reveals that the CNN-VAE metric performs better than the integrated intensity fluctuations for predominantly all [Formula: see text] values.

We perform lab scale experiments in a swirl combustor by reducing the equivalence ratio for two cases involving slightly different inlet air flow rates. Reduction in equivalence ratio at a constant air flow results in the combustor transiting from stable to unstable combustion. The transition passes through stable, type-2 intermittency and beat oscillations. The beat oscillations are seen to fluctuate at a single frequency, and hence suggestive of a process involving spatio-temporal variations in the phase of the driver (i.e., heat release rate fluctuations). This is understood in the manner of a reduced order model, where the driver is considered to an ensemble of phase oscillators with time delays and a probabilistic distribution of natural frequencies, similar to Kuramoto oscillator. The acoustic field is modeled as a Van-der-pol Duffing system, with natural frequency equal to the duct acoustic mode. The coupling between the oscillators is varied based on the physical premise of Rayleigh criterion. The coupled system is seen to qualitatively and quantitatively match the pressure data obtained from experiments. Insights into various conditions illustrate the role of mean and fluctuating instantaneous frequency amongst the phase oscillators in determining the modeled pressure oscillations. By quantifying the extent of fluctuating frequency coupling among the phase oscillators, it is observed that beat oscillations have high correlation with lower deviation compared to intermittency and stable oscillations.