The National Jet Fuels Combustion Program, focused on reducing the carbon intensive evaluation and approval process of alternative jet fuels, has found a first order dependence on derived cetane number in predicting lean blowout for most of the combustor rigs in the program. Additionally, it has been observed that at least one rig has shown no correlation with the derived cetane number and lean blowout. These observations, moreover, have been hypothesized to be explained via timescale analysis considering fuel evaporative, combustor mixing, and chemical reactivity timescales. This paper combines this timescale theory with reduced order random forest regression properties to represent each of the previously identified timescales. Evaporative timescales are estimated via a fuel's density and 20% recovered temperature to represent spray quality/atomization and evaporative potential, respectively. Autoignition and extinction chemical timescales are estimated by a fuel's derived cetane number and radical index, respectively. Random forest regressions with only these four properties are able to account for better than 89% of experimental variance in four diverse rigs and hundreds of data points. Applying these results to timescale theory corroborates previous observations. Rigs with minimal fuel atomization spray differences are minimally sensitive to evaporative timescales. The rig with the most reported sensitivity to viscosity, a Honeywell Auxiliary Power Unit type rig, also shows the most sensitivity to atomization and evaporative timescales. The fourth rig studied from the University of Sheffield shows a near equivalent dependency on chemical and evaporative timescales. These simplified regressions illustrate how four fuel properties, representative of each timescale, are able to accurately capture the physics dominating LBO for each rig, and in turn reduce the carbon intensive evaluation and approval process of alternative jet fuels.
展开▼
