3166 Ambient monitoring of benzo-a-pyrene (B[a]P) emissions at two oil refineries

Executive summary

This report describes work undertaken by the National Physical Laboratory, for the Energy Institute’s Air Emissions Group, to monitor boundary-fence concentrations of particulate phase benzo-[a]-pyrene (B[a]P) at two oil refineries in the UK. Monitoring took place at Refinery A in early November 2004, and at Refinery B in late November/ early December 2004.

NPL acquired five pairs of samples at each refinery; each sample pair consisting of samples taken at nominally upwind and downwind locations of the potential PAH sources at each refinery. Additionally, three pairs of vapour phase samples were acquired at each refinery.

Analysis of these samples revealed that there was no significant difference between the (particulate or vapour phase) B[a]P concentrations at the upwind and downwind sampling locations. In addition, the B[a]P concentrations measured at the refineries were comparable to the concentrations expected at urban sampling locations around the UK PAH Network during November and December.

The average measured concentrations at the refineries were well below the target value stated for PAHs in the European Commission’s Fourth Air Quality Daughter Directive.

2. Methodology

NPL visited the two oil refineries during November and December 2004 to conduct two separate monitoring exercises.

A minimum of three separate pairs of, simultaneously acquired, upwind and downwind samples were concurrently taken at each site. Samples were acquired using two Graseby Andersen High-Volume samplers (with a sampling rate 68 m³ h^-1) with a PM₁₀ size selective inlet, sampling in parallel. The particulate phase of the ambient air was collected on quartz filters within the samplers. For a selection of the samples, polyurethane foam inserts were used, in addition to the particulate filters, in order to sample B[a]P in the vapour phase. The particulate and vapour phase samples were analysed separately. The flow rate of the High-Volume Samplers was calibrated, traceable to national flow standards, before each sample was taken.

In addition, at each site, a field blank filter, and field blank polyurethane foam insert, was analysed to determine the blank levels to be subtracted from the measured results.

Initially sampling periods of 48 hours were envisaged. However, owing to the limited number of sampling sites available at each refinery coupled with the variability of the wind direction over relatively short periods, it was clear that a more robust study would be obtained with a greater number of shorter sampling periods.

During sample acquisition the samplers were placed in the locations which would be the closest to being directly upwind and downwind from the predicted centre of B[a]P emission. This siting was based on local Met Office weather forecasts for the forthcoming sampling periods. At Refinery B there were no options for sampler location, so in this instance the methodology was to wait until the wind was set to be in a beneficial direction for the duration of the sampling period. Once the samples started being collected, the samplers were then not moved during the sampling period.

During the sampling period real time local metrological data, such as wind direction, wind speed, temperature and pressure was gathered. Additionally, hourly meteorological observations of rainfall, visibility and cloud cover were also taken during the day and supplementary air pollution information was obtained for the sampling periods. These observations ensured that the samples were gathered under worst-case meteorological conditions; for example, there was no more than 30 minutes of rainfall during any sampling period, and during all sampling periods ozone concentrations (which can degrade sampled B[a]P) were low.

Emissions information, for the main sources of B[a]P on the refinery, such as source height, emission flux and emission temperature, was provided by the refinery.

After sampling the filters and polyurethane foam inserts were uniquely labelled and stored securely by NPL.

The extraction and analysis of the filters was performed using NPL’s fully validated methodology for PAH measurement in ambient air. The process is completely described by a full measurement equation and robust uncertainty budget, covering all aspects of sampling, extraction and analysis, in order to deliver concentrations values of PAHs in ambient air with valid uncertainty statements. This methodology has been developed to be fully compliant with ISO 12884 and the proposed analysis strategies currently being drafted by CEN/TC264/WG21.

Validation of the extraction procedure is carried out regularly by the extraction and analysis of a US National Institute of Standards and Technology (NIST) standard reference material (usually NIST SRM 1649a). This procedure is carried out regularly to ensure the accuracy and consistency of the results. In addition, before each extraction a surrogate standard, in this case per-deuterated perylene, is added to each sample in order to evaluate the efficiency of the extraction process and the variability in this extraction. Our procedures have been designed to ensure that this extraction efficiency is maximised and the uncertainty in this is maintained within acceptable limits.

NPL analysed the B[a]P extracts using GC-MS. The sample analysis is performed by a Hewlett-Packard 6890 GC system with 5973 Mass Selective Detector (MSD) and HP-5MS crosslinked PH-ME siloxane column (60m x 0.25mm x 0.25mm film thickness). The GC-MS is maintained according to the manufacturer’s instructions with air and water checks and a mass calibration being carried out before each analysis sequence.

Calibration is carried out using at least three gravimetrically prepared concentrations of NIST standard reference solutions containing all PAHs of interest and the internal standards to be used (d₁₂-B[a]P and d ₁₀-phenanthracene). The concentrations of the calibration solution are always chosen so as to bracket the likely value of the sample to be measured in order to ensure the best possible accuracy during calibration.

The corrected peak area ratios are analysed by a method of generalised least squares (GLS) using XGenline (an NPL-developed program), to construct a calibration curve and determine the mass fractions and uncertainties of the unknown samples.

The vapour phase samples on the polyurethane foam inserts were analysed by Marchwood Scientific Services (Marchwood, Southampton) who have UKAS accreditation for this procedure.

Results are quoted as the mass of B[a]P per cubic metre of ambient air with measurement uncertainties where applicable.

4. Conclusions

Analysis of the B[a]P concentration levels measured reveal that there is no significant difference between the (particulate or vapour phase) B[a]P concentrations measured at the upwind and downwind boundary-fence sampling locations, within the accuracy of the measurement methodology.

As expected, owing to its low volatility, the concentrations of B[a]P measured in the vapour phase was significantly lower than in the particulate phase.

Furthermore, the concentrations of particulate phase B[a]P in all the samples taken were relatively low. The B[a]P concentrations measured at the refineries are comparable to the concentrations expected at urban sampling locations around the UK PAH Network during the fourth quarter of the year. To illustrate this, the table below compares the average B[a]P concentrations recorded at the refineries with the average fourth quarter concentrations recorded at a selection of sites on the UK PAH monitoring Network.

Additionally, the average measured concentrations at the refineries were well below the target value stated for PAHs, of 1.0ng.m ^-3, in the European Commission’s Fourth Air Quality Daughter Directive.

Table 3: Average B[a]P concentrations recorded at the refineries during the sampling periods
(November and December 2004) together with the average fourth quarter concentrations recorded at a selection of sites on the UK PAH monitoring Network during October to December 2003

Acknowledgements

This Report was produced for the Environment Agency for England and Wales and the Energy Institute by the National Physical Laboratory. The EI wishes to thank NPL and the principal authors Richard J. C. Brown, Peter T. Woods, Andrew S. Brown and Roger S. Brown.

The project was steered by a joint EI and EA Steering Group, members of which included: