Predictive Models of Intake Valve and Combustion Chamber Deposit Formation Tendency Based on Gasoline Fuel Composition 2020-01-2097
Internal engine deposits are predominantly the product of incomplete combustion of fuel and have the potential to negatively impact engine performance and exhaust emissions. The ability to expeditiously screen fuels for their deposit formation tendency would be advantageous. This paper presents Partial Least Squares (PLS) computational models with the ability to predict the degree of deposit formation tendency from the hydrocarbon composition of gasoline fuels. Furthermore, the relationship between gasoline fuel and Intake Valve Deposits (IVD) and Combustion Chamber Deposits (CCD) that form in Port Fuel Injected (PFI) engines is explored. The models utilize detailed hydrocarbon analysis (DHA) to resolve the complex mixture of hydrocarbons for different model fuels, which are correlated to ten years of engine data collected from a port fuel injected engine. CCD and IVD models were built using DHA data from 17 model fuels and validated with an additional 20 validation fuels for the CCD model and 16 fuels for the IVD model. Robust statistical treatment of engine deposit measurements was done with median absolute deviation (MAD) to filter out outlier data and resulted in a 2-4 fold data set precision improvement. The degree of deposit formation tendency in 75% of the validation fuels were predicted successfully for CCD and 87.5% for IVD. Variable selection methods were used to evaluate the hydrocarbon compounds that correlate with deposit prediction in the CCD and IVD models, which aligned with previously published correlations between hydrocarbon class and deposit formation. The models represent an improvement in R2 when compared to previous publications.
Citation: Morlan, B., Smocha, R., and Lorenz, R., "Predictive Models of Intake Valve and Combustion Chamber Deposit Formation Tendency Based on Gasoline Fuel Composition," SAE Technical Paper 2020-01-2097, 2020, https://doi.org/10.4271/2020-01-2097. Download Citation
Brian Morlan, Ruth Smocha, Robert Lorenz