Monitoring fuel quality: a case study for quinizarin marker content of unleaded petrol marketed in Greece

Fuel adulteration and cross-contamination lead to low-quality fuel products, which may cause increased environmental pollution, loss of taxes and engine problems. An establishment of a quality monitoring mechanism based on laboratory measurements may reveal problematic areas of the fuel supply chain. For the purposes of this work, 97 unleaded petrol samples were measured in order to quantify mass concentration of quinizarin, a substance used in Greece to easily mark the presence of 95 Research Octane Number unleaded petrol in other types of automotive fuels. The samples were obtained from petroleum retail stations selling different brands of fuels and located in different geographic regions of Greece. Statistical analysis of the results revealed quinizarin mass concentrations below the 3 mg L^?1 legislation specification limit and significant differences between brands and geographic regions, which may attributed to the structure of the fuel supply chain in Greece in combination with quinizarin properties and way of handling. Moreover, certain approaches were used for the calculation of decision limits for assessing compliance or non-compliance. These approaches take measurement reproducibility or estimated in-house uncertainty into account, in order to minimize the probability of false rejection.     

Estimation of the standard uncertainty of a calibration curve: application to sulfur mass concentration determination in fuels

The construction of a calibration curve using least square linear regression is common in many analytical measurements, and it comprises an important uncertainty component of the whole analytical procedure uncertainty. In the present work, various methodologies are applied concerning the estimation of the standard uncertainty of a calibration curve used for the determination of sulfur mass concentration in fuels. The methodologies applied include the GUM uncertainty framework, the Kragten numerical method, the Monte Carlo method (MCM) as well as the approximate equation calculating the standard error of prediction. The standard uncertainty results obtained by all methodologies agree well (0.172?C0.175?ng???L^?1). Aspects of inappropriate use of the approximate equation of the standard error of prediction, which leads to overestimation or underestimation of calculated uncertainty, are discussed. Moreover, the importance of the correlation between calibration curve parameters (slope and intercept) within GUM, MCM and Kragten approaches is examined.