This paper develops a novel strategy for prediction of lean blowout in gas turbine combustors based on symbolic analysis of time series data from optical sensors, where the range of instantaneous data is partitioned into a finite number of cells and a symbol is assigned to each cell. Depending on the cell to which a data point belongs, a symbolic value is assigned to the data point. Thus, the set of time series data is converted to a symbol string. The (estimated) state probability vector is computed based on the number of occurrence of each symbol over a given time span. For the purpose of detecting lean blowout in gas turbine combustors, the state probability vector obtained at a condition sufficiently away from lean blowout (reference state) is considered as the reference vector. The deviation of the current state vector from the reference state vector is used as an anomaly measure for early detection of lean blowout. The results showed that the rate of change of the anomaly measure with equivalence ratio changed significantly as the system approached lean blowout. This change in slope of the curve was observed approximately at a similar proximity to lean blowout for different operating conditions and, hence, could be used as an early lean blowout precursor. The actual location of the change of slope depended primarily on the choice of reference state. This technique performed satisfactorily over a wide range of premixing.

No abstract

The study characterizes the behavior of a premixed swirl stabilized dump plane combustor flame near its lean blow-out (LBO) limit in terms of CH* chemiluminiscence intensity and observable flame color variations for a wide range of equivalence ratio, flow rates and degree of premixing (characterized by premixing length, L fuel ). LPG and pure methane are used as fuel. We propose a novel LBO prediction strategy based solely on the flame color. It is observed that as the flame approaches LBO, its color changes from reddish to blue. This observation is found to be valid for different levels of fuel-air premixing achieved by changing the available mixing length of the air and the fuel upstream of the dump plane although the flame dynamics were significantly different. Based on this observation, the ratio of the intensities of red and blue components of the flame as captured by a color CCD camera was used as a metric for detecting the proximity of the flame to LBO. Tests were carried out for a wide range of air flow rates and using LPG and CH 4 as fuel. For all the operating conditions and both fuels tested, this ratio was found to monotonically decrease as LBO was approached. Moreover, the value of this ratio was within a small range close to LBO for all the cases investigated. This makes the ratio suitable as a metric for LBO detection at all levels of premixing.

Lean or ultralean combustion is one of the popular strategies to achieve very low emission levels. However, it is extremely susceptible to lean blow-out (LBO). The present work explores a Crosswavelet transform (XWT) aided rule based scheme for early prediction of lean blowout. XWT can be considered as an advancement of wavelet analysis which gives correlation between two waveforms in time-frequency space. In the present scheme a swirl-stabilized dump combustor is used as a laboratory-scale model of a generic gas turbine combustor with LPG as fuel. Various time series data of CH chemiluminescence signal are recorded for different flame conditions by varying equivalence ratio, flow rate and level of air-fuel premixing. Some features are extracted from the cross-wavelet spectrum of the recorded waveforms and a reference wave. The extracted features are observed to classify the flame condition into three major classes: near LBO, moderate and healthy. Moreover, a Rough Set based technique is also applied on the extracted features to generate a rule base so that it can be fed to a real time controller or expert system to take necessary control action to prevent LBO. Results show that the proposed methodology performs with an acceptable degree of accuracy.

No abstract