(Georgia Institute of Technology, 2019-07-26)
Adusumilli, Sampath
Conducting full-scale experiments as part of the design process of jet engine combustors is a costly and time-consuming process. Therefore engine developers have been increasingly using numerical modeling approaches to assess new designs or design changes. The reaction chemistry, which is dependent on the flow conditions, the fuel composition, and the oxidizer composition, plays an important role in the accuracy of these simulations. The kinetic mechanisms that describe this chemistry need to be validated. Various global combustion characteristics are used to validate mechanisms against experimental data; one of these is laminar flame speed (SL). In this work, laminar flame speeds of various fuels relevant to jet engine combustion are measured using a previously developed, modified Bunsen Flame Technique (BFT). The accuracy of the BFT is examined here, both through a comparison to experimental results from other standard approaches for a range of fuels and through a detailed analysis of the impact of flame stretch. The measured flame speeds are also used to test leading chemical kinetic mechanisms, primarily the NUI and USC models. Laminar flame speeds of n-decane, ethylene and propylene are measured at conditions relevant to jet engine main combustors and afterburners. The experimental conditions include high preheat temperatures (up to 650 K) and reduced O2 levels (down to 15% mole fraction in the oxidizer); the latter is relevant to vitiation, where there is partial pre-burning of the oxidizing flow. Furthermore, vitiation introduces combustion products such as CO2 into the reactant stream that can participate in the combustion chemistry. Therefore, flame speeds are measured using dilution with both CO2 and N2 (considered non-reactive) to study these effects. SL measurements for alkenes using BFT are within 10% of measurements from literature and chemical kinetic mechanism predictions at 300 K and atmospheric pressure. At high preheat temperatures, the mechanisms accurately predict SL for ethylene mixtures, while they over predict SL of propylene mixtures at 650 K. Vitiation studies at 650 K preheat show that for N2 dilution and ethylene, the reduction in flame speed is mostly due to thermal effects. Some chemical effects were observed when the O2 level in the oxidizer was reduced to 15% (vol.). For propylene, reducing O2 had a larger impact on flame speed than that predicted by the mechanisms. With CO2 as a diluent, the mechanisms over predicted the flame speed, and the prediction error increased with higher levels of CO2. Reactions involving the allyl (C3H5-A) radical were identified as a likely source of the propylene flame speed errors, increase in the pre-exponential rate factor of the allyl-H recombination reaction improved the predictive capability of the mechanism at high preheat temperatures. Similarly, analysis of different sources of errors with CO2 dilution suggest the third-body efficiency of CO2 is underestimated in a three-body association (such as H + O2 (+M) <=> HO2 (+M)) type of reactions.