| The research hypothesis of this study is: "In a Life Cycle Impact Assessment (LCIA) context and for the aquatic ecotoxicity category, the hydrocarbon complex mixture theoretical characterization factors (CFs), developed by using the mixture principal fraction properties, have values which are not significantly different compared to the CF obtained from experimental measurements, which take into account potential interactions between co-contaminants.";As petroleum hydrocarbons form mixtures whose constituents have a common toxic mode of action, the pole-treating oil was chosen. Firstly, methods had to be determined in order to calculate CFs for complex mixtures. The CFs, and consequently the fate factors (FFs) and effect factors (EFs), for the aquatic ecotoxicity category were calculated by using the model USEtox version 1.01. Ecotoxicological FFs, EFs and CFs were then obtained for 1) each potential constituents, 2) each fractions according to the HBM, 3) each original fractions from the TPHWG as given in their study, 4) each TPHWG refined fractions and 5) the pole-treating oil as a whole component. The factors according to the HBM were calculated by taking for each fraction the arithmetic mean of the FFs, EFs and CFs of the substances contained in it. After that, the minimum, the maximum and the geometric mean were determined for the calculations 1, 2, 3 and 4, and were compared with each other and with the oil experimental ones. Finally, as the definition of the properties required for the calculations is a key part of the project, a sensitivity analysis was conducted in order to determine to which ones the final results are sensitive.;Six hundred and seventy seven pole-treating oil potential constituents were identified. They lead to the formation of 13 original TPHWG fractions, 18 refined ones according to the TPHWG method (four chemical classes and eight ECN ranges) and 18 fractions according to the HBM (four chemical classes and five boiling point ranges). Generally, the properties experimentally measured were different from the ones from literature. The properties of the oil as a whole component were either lower (one to three orders of magnitude) or higher (one or two orders of magnitude) than the potential constituent and TPHWG fraction geometric mean ones, except for the molecular weight, the octanol-water partition coefficient and the degradation constant in water. However, the property values were included in the value range created by the minimum and the maximum for each property. Then, the FFs, EFs and CFs in freshwater were compared for the calculations 1, 2, 3, 4 and 5 for a pollutant emission to urban air and then on a continental scale to rural air, freshwater, seawater, natural soil and agricultural soil. The results show that the FF, EF and CF ranges obtained by using the fractioning methods are smaller than the potential constituent ones. As a matter of fact, fractions reduce the obtained factor variability from one to three order of magnitude compared to the potential components.;The sensitivity analysis of the properties underlined those which have to be defines cautiously. The toxicity data used as input in USEtox (EC 50), the octanol-water partition coefficient, the Henry law constant and the degradation constant in the air are the parameters which influences the most the final factors. A change in their value of more and less 50% leaded to high FF, EF and CF variations (approximately 30% to 200%). The EC50 data strongly depend on Kow, which is the result of theoretical algorithms and experimental measurements. The degradation constant in the air is a default value taken in the USEtox model. Consequently, it is not representative of the studied mixture, but has an influence on the obtained results.;CFs were also calculated for the pole-treating oil mixture. The theoretical methods seem to underestimate the factors obtained from experimental measurements, which take into account potential interactions between co-contaminants. As a matter of fact, for an emission to freshwater theoretical FFs and CFs are generally lower than the experimental ones of two orders of magnitude (10 +02). However, the experimental CFs are contained within the CF ranges obtained by using the theoretical methods and are close to the geometric means. Consequently, the methods taken from literature give a good approximation of the CFs for hydrocarbon complex mixtures and can be used in order to characterize this type of pollutant in LCA studies. (Abstract shortened by UMI.)... |
