Comparison of Raman and IR spectroscopy for quantitative analysis of gasoline/ethanol blends

Abstract Ethanol is commonly admixed to petrochemical gasoline, and its amount in the fuel blend can influence the performance of an engine. The ethanol content in a commercial fuel can vary. To ensure reliable engine operation, control strategies based on a measurement of the composition need to be developed. Two possible methods to determine the ethanol content in ethanol/gasoline blends are Raman and IR spectroscopy. We compare both techniques for quantitative measurements in systematically varied blends of ethanol and a gasoline surrogate. For each method, two different approaches for data evaluation are tested and compared: Firstly, the calibration of the intensity ratio of characteristic peaks as function of composition; secondly, a principal component regression (PCR). Both methods are found to have comparable uncertainty. For the evaluation of the Raman spectra, the PCR method yielded better accuracy than the intensity ratio approach. In addition, a detailed investigation of the influence of noise in the signal is presented. When the full IR spectra were evaluated by PCR, even high noise levels did not reduce the measurement accuracy significantly.