Dynamic Characterization of a Ducted Inverse Diffusion Flame Using Recurrence Analysis

Abstract

Normal diffusion flame or partially premixed flame is used in many applications such as aviation engines, tanks, ocean vessels, and industrial furnaces because of its high flame stability and relatively low susceptibility to dynamic instabilities compared to lean premixed flames, which give lower emissions. However, associated with such flames are high NO x and soot emissions, which are particularly high for heavier hydrocarbon fuels.Increasingly stringent environmental norms have thus dictated the search for alternate approaches; one such being the inverse diffusion flame, which is currently being used in rocket motors, for staged combustion in gas turbine combustors, and furnaces. However, the dynamic response of such a flame, particularly in ducted applications where a coupling between unsteady heat release rate and duct acoustics may occur, is relatively less explored.The present work aims to plug that knowledge gap through an experimental investigation of a laboratory-scale ducted inverse diffusion flame. Two parametric variations of the flame were performed -variation of the flame position and variation of the air flow rate. Using tools of nonlinear dynamics such as phase space reconstruction and recurrence quantification, several interesting dynamic characteristics were observed such as limit cycles, intermittency, and homoclinic orbits. For a constant air flow rate, the system was observed to transition from a type-II intermittency regime to a limit cycle and then again to intermittent behavior as the position of the flame within the duct was varied. A similar trend was observed when the air flow rate was varied at a fixed flame position.

Effects of operating parameters on the combustion oscillation behaviour in a lean premixed CH4 combustor

“…Moon et al [29,30] studied the coupling and decoupling characteristics of thermoacoustic oscillations under different lengths and cross-talk interactions, and they found the mode clustering phenomenon during experiments. Sen et al [31] explored the dynamic attributes of combustion instability with recurrence analysis methods, and the air flow rate and flame position were studied to see the flame behaviour. Gopalakrishnan et al [32] illustrated the tipping point of warning signals during the thermoacoustic instability transition, and they revealed that the early warning signals can serve as indicators of thermoacoustic oscillations.…”

“…Although there are some studies on the influence of burner sizes on the dynamic characteristics of thermoacoustic oscillation [27][28][29][30][31][32]. At present, there is still no research investigates the nonlinear effect of combustor length and fuel flow rate on thermoacoustic instability.…”

Cross-Modal Virtual Sensing for Combustion Instability Monitoring

“…However, these approaches may require simplifying assumptions and face difficulty in achieving validation. An alternative is to implement data-driven methods utilizing acoustic pressure time-series [13,20,26]. These datadriven methods, based only on acoustic pressure time series, can sometimes be inaccurate due to interference from broadband background noise.…”

Data-driven Detection and Early Prediction of Thermoacoustic Instability in a Multi-nozzle Combustor

“…This method is further expanded upon by Bhattacharya, De, Mukhopadhyay, Sen and Ray (2020) where the method was shown to be effective at discriminating between lean blow-out, thermoacoustic instability, and stable operation by using a very simple scalar metric based approach. Recurrence analysis (Sen et al, 2018) has also been used as a means to distinguish stable signals from unstable ones. However, methods like recurrence analysis and complex networks are too slow to be used in an online detection or analysis setting; therefore, other faster and more accurate data-driven techniques have been explored.…”