Review of Tier 1 Workplace Exposure Estimates for Petroleum Substances in REACH dossiers
For the exposure assessment in the 2010 REACH dossiers of petroleum substances, Concawe has used the Tier 1 exposure model ECETOC TRA v.2. In order to account for the heavier, less volatile and more complex petroleum substances and the corresponding exposure situations, several modifications not originally within the scope of the ECETOC TRA were developed. These modifications include an approach to estimate liquid aerosol along with some risk management measures describing the use and handling of petroleum substances commonly in use in the European oil refining industry.
In this project, Chemical Safety Assessments (CSAs) on these petroleum substances were evaluated concerning relevant industry areas and included scenarios. Measured data were collated in order to evaluate the exposure estimates in general and the modifications made by Concawe.
Data were collected from the database built in the course of the Evaluation of Tier 1 Exposure Assessment Models (eteam) project (www.eteam-project.eu) undertaken by IOM and ITEM, from Concawe reports and via literature search. These include kerosines (mainly as jet fuel), heavy fuel oils (HFOs), two naphtha groups (0-1% and 1-5% benzene, essentially corresponding to gasoline after and before the year 2000), gas oils (vacuum, hydrocracked gas oils and distillate fuels; i.e. mainly diesel fuels) and other lubricant base oils (OLBOs). The quantity and quality of datasets, however, varies between the groups and there are still a number of substance groups and scenarios which could not be evaluated due to lack of suitable data (e.g. gas oils, foots oils, other naphthas). This was partly due to changes brought about by the REACH process in how exposures are assessed against a reference value. For instance, it was customary to assess inhalation exposure to diesel fuel as vapour, but the REACH process resulted in a Derived No Effect Level (DNEL) expressed as aerosol.
The comparison exercise showed some discrepancies depending on substance group and the specific scenarios. These discrepancies may be partly attributed to new modifiers or other changes of the ECETOC standard algorithm (e.g. concentration modifier in case of naphthas). In general, for most measures such as draining of equipment or training of operatives, both under- and overestimations can be found; therefore, it is difficult to reach a final conclusion concerning their applicability.. Other possible reasons for the observed underestimations were variations within an exposure scenario (e.g. Research and Development laboratory activity vs. production laboratory activity, bottom loading vs. top loading) or the age of datasets (e.g. in case of naphthas).
Concerning aerosol exposure, measured data for OLBOs and HFOs could be identified. No significant underestimations were found for the evaluated scenario in case of OLBOs while in the case of HFOs results were inconclusive (partly underestimations but only few data points). Overall it is recognised that available sampling methods for liquid aerosol often tend to give biased or at least variable exposure results and this has to be taken into account for future investigations concerning risk assessment of petroleum substances or validation of the existing CSAs. Measurements made for HFOs show higher overall and vapour concentrations compared to the aerosol values which may suggest that either vapour may be more relevant than previously assumed for high boiling petroleum substances or the corresponding aerosol measurements may not be suitable for a comparison with DNELs or model estimates. Comparable difficulties will probably exist for other semi- or low volatile substances which tend to form aerosols.
Although some underestimations have been observed, there are also cases where clear overestimations were observed and thus, a further refinement with higher Tier tools may be possible. Two possible tools, STOFFENMANAGER© and ART were discussed and illustrated with an example scenario. Petroleum substances and the resulting exposure types (vapour and aerosols) are within the scope of both models; however, the new modifiers introduced by Concawe are only implemented to a limited extent (vapour recovery in the case of ART).
A qualitative evaluation of the updates made when changing from ECETOC TRA v.2 to v.3 suggested that inhalation exposure estimates will probably be lower if the more recent version is used. This is partly due to newly introduced or changed measures or operational conditions and partly due to modified initial exposure estimates.
Overall, there are a number of situations where the comparison of measurements and estimates suggests reasonable results and a controlled risk. There are other situations, however, where, due to different reasons, the contrary is observed. A particular problem seems to be the lack of high quality aerosol data.