Assessment and prediction of exposure to benzene of filling station employees

In the present study, the exposure to benzene of employees working in two filling stations (one urban and one rural) was estimated, through the method of passive sampling. Additional data (30′ measurements of benzene exposure through active sampling to employees dealing with different activities, meteorological and traffic data) were collected. The measurements campaign was performed in both summer and wintertime to determine the seasonal variation of the exposure pattern.In addition, a set of artificial neural networks (ANNs) was developed to predict benzene exposure pattern for the filling station employees based on active sampling data and the parameters related to the employees’ exposure. The quantification of the contribution of each parameter to the overall exposure pattern was also attempted.The results showed that although vapour recovery technologies are installed in the refuelling systems and benzene emissions are significantly reduced compared to the past, filling station employees are still highly exposed to benzene (52–15 μg m−3). Benzene exposure is strongly correlated to car refuelling (exposure levels up to 85 μg m−3), while activities like car washing or working in cash machine inside an office contribute to lower exposure levels (up to 44 and 24 μg m−3 respectively). In rural filling station, exposure levels were in general lower compared to the urban ones, due to the smaller amount of gasoline that was traded and the absence of any significant traffic effect or urban background concentration. The developed ANN seemed to be a promising technique in the prediction of the exposure pattern giving very good results, and the quantification of the parameters affirmed the importance of the refueling procedure to the exposure levels.