Application assessment of UV-vis and NIR spectroscopy for the quantification of fuel dilution problems on used engine oils

Fuel dilution in engine oil is a frequent problem in internal combustion engines (ICE); its main consequence is the reduction of the oil viscosity, decreasing lubrication film strength, and causing a negative impact on friction and wear. The standard and more precise methods for assessing fuel content in oil are based on chromatographic analysis (e.g., ASTM D3524, ASTM D7593), requiring high-cost equipment and highly qualified personnel. This work performed a study to propose an alternative method for quantifying diesel fuel dilution in used engine oil by UV-vis and NIR spectroscopy. The samples for the study were prepared from used oil from six different ve-hicles with various mileages. According to the results obtained in this study, NIR spectroscopy proved to be the most suitable method for the quantification of diesel fuel in used engine oils. Furthermore, the use of NIR spectroscopy combined with multivariate calibration methods could predict the fuel concentration of the samples used for validating the model. The best predictive model for the quantification was obtained by Partial Least Squares Regression, which achieved a Root Mean Squared Error of prediction of 0.436% and a coefficient of determination of 0.9435. In comparison, the parameters for Principal Component Regression were 1.049% and 0.8441, respectively.

Automated summary

Fuel dilution in engine oil is a common problem in internal combustion engines, leading to negative effects on oil performance. Traditional detection methods require expensive equipment and specialized personnel. This study proposed an alternative method using UV-vis and NIR spectroscopy for quantifying diesel fuel dilution, and demonstrated that NIR spectroscopy is the most suitable method. Additionally, multivariate calibration methods combined with NIR spectroscopy can predict fuel concentration, with the best predictive model obtained using Partial Least Squares Regression.