5.1 Testing programme
Test sites were selected by considering both the number of sites processing various metals and the total quantity of each metal melted. As discussed in section 4, a reasonable number of iron and steel sites were included in the testing programme because there is a large number of small and small-to-medium sites producing secondary iron and steel.
Unfortunately, the largest site, responsible for the majority of copper production, was unwilling to participate in the test programme. One other large site (Site F) also processes copper, and this site was included in the study; but Site F primarily processes aluminium, and emissions from melting copper and aluminium are vented to the same air pollution control equipment. It was not possible to differentiate between metals in this case, and emissions from aluminium processing are likely to have dominated the results.
Some smaller sites were included to investigate whether these were likely to be disproportionately large dischargers of PCDD and PCDF relative to their production because of, for example, poor process control and/or a lack of emission control.
There are many variables, and coverage was achieved as broad as possible within the resources available. The study has not been able to derive emission factors with a specified statistical level of confidence. Nevertheless, the data is considered to be representative of New Zealand emissions because only a few large sites are responsible for the majority of secondary metals production.
Emission testing was carried out at 12 sites throughout the North and South Islands of New Zealand. The testing programme is summarised in Table 6.
The 12 sites represent 10% of the total number of sites in New Zealand, but in terms of metal melted there is a considerably greater degree of representativeness. All four of the largest aluminium processing sites were included, accounting for 95% of the total aluminium production; 75% of the iron and steel production and 23% of the copper production were accounted for in the test programme.
Twenty-four results for total PCDD and PCDF were obtained from 30 analyses, including four gas-phase and particle-phase PCDD and PCDF analyses, and two samples collected by concurrently running two high-volume samplers. The sample train used is able to split the sample into particulate and gas phases, and this was used to investigate possible emissions controls via particulate removal. The sites expected to have the highest particulate matter emissions were used for the splits.
5.2 Test and analytical methods
Stack emission testing was carried out using US EPA Method 0023A for determining polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans and total particulate from stationary sources. US EPA Methods 1 and 2 were used to select sample point locations and traverses. The number of sampling points was determined in accordance with ISO 9096:1992(E).
Sample times and volumes for each site are reported in Volume II, with other emission test data. A minimum sample time of 2 hours was used. Sample volumes were greater than 1.5 m3 at standard conditions and most samples were 3 to 5 m3. The sample time and/or gas volume may have been limited in some cases due to relatively short processing times and/or low flue gas flows.
Details of the sample locations are provided in Volume II.
High-volume samplers with PUF attachments were used to measure PCDD and PCDF at Site L. US EPA Reference method T-09A was used. Site L has general dilution ventilation via two roof extracter fans. Two samplers were used during each test run; one was located near the furnace operations (S26 and S28+S29) and the other nearer to the casting operations (S27 and S30). The mean concentration of the results from the High-volume samplers was used to determine the emission. The building has good-quality ventilation systems with limited escape of fugitive discharges through doors. Air concentrations within the process building were measured with all doors closed to avoid the risk of emissions escaping undetected. Building ventilation air flows were measured at the same time to give as accurate a measure as possible of building ventilation rates and to allow calculation of mass emission rates.
All XAD-2 resin cartridges were spiked with a range of isotopically labelled PCDD and PCDF standards prior to the collection of the sample. All sample collection media (XAD-2, filters and glassware) were pre-cleaned prior to use.
Following receipt at the laboratory post sampling, samples were stored at 4°C pending analysis. Each sample consisted of an XAD-2 cartridge, filters (Whatman GFC) and solvent rinses. The particulate content for each sample was determined by desiccating then weighing each filter.
Sample extraction, purification and extract work-up steps were based on US EPA 1613B methodology. [Method 1613B is an equivalent method to Method 8290, which is referenced in the test method US EPA 0023.] Extracts were analysed by isotope dilution using high-resolution gas chromatography - high-resolution mass spectrometry (Micromass Autospec Ultima). The laboratory detection limit for air emission samples is reported at 1-10 pg/m3.
Mass of particulate was estimated by gravimetric determination. Filters were weighed in the laboratory before sampling. After sampling, filters were brought back to room temperature in a desiccator and reweighed to determine the mass of particulate. Particulate analysis results are reported to two significant figures.
PCDD and PCDF analysis results are reported to three significant figures as total mass of PCDD and PCDF in both international toxic equivalencies (I-TEQs) and World Health Organisation toxic equivalencies (WHO-TEQs).
5.3 Quality assurance
The Ministry for the Environment has previously signalled in proposed regulations that PCDD and PCDF sampling should be undertaken by IANZ- or NATA-accredited organisations (Ministry for the Environment, 2001). An IANZ-registered laboratory was used for the analysis, but no accredited stack-testing companies were available in New Zealand, nor were two prominent Australian contractors (at the time). The stack-testing company selected had, however, commenced the process of IANZ accreditation and their application has since been lodged for assessment.
One field duplicate sample was collected by running two sample trains simultaneously in the stack at Site A. The results agreed within a factor of two (S1 and S2). The congener details for all samples are provided in Appendix A of Volume II.
Laboratory methods were quality assured as follows.
- A laboratory glassware and reagent blank was analysed with each batch of samples and assessed for background levels.
- A matrix spike was analysed with each batch of samples to assess method precision and recovery. The matrix spike was spiked into cleaned XAD-2 resin prior to extraction.
- The high-resolution mass spectrometer (HRMS), performance, sensitivity and resolution were established for each instrumental run.
- The recoveries of all isotopically labelled surrogate standards were calculated and reported. The quality control acceptance guideline for surrogate standard recovery was 40-135 % recovery.
A variety of checking procedures were in place to ensure the integrity of all laboratory processes. These included:
- sample receipt checks
- laboratory checks
- HRMS analysis checks
- analytical data and report check
- final report check.
The analytical laboratory used regularly participates in Interlaboratory Comparison Programmes for PCDD and PCDF analysis.
No significant quality issues arose from the laboratory analysis. Background levels in laboratory blanks were insignificant with respect to levels in the samples. Matrix spike recoveries were generally within acceptable limits. In the few cases where the matrix spike recoveries of some analytes were outside the limits specified in US EPA Method 1613B, professional judgement was used in assessing and reporting the analytical data. The matrix spike recovery data for this project is reported in the laboratory data sheets provided in Volume II, Appendix A.
Some isotopically labelled surrogate standard recoveries for samples S3, S7, S8, S12, S14 and S18 were outside the method guidelines of 40-135%. The lowest recovery was 16% (OCDD for sample S7), and the highest was 140% (12378-PeCDF for sample S3). Surrogate recoveries can be influenced by a variety of factors (including the effect of co-extractives in the sample matrix on chromatographic and other laboratory processes), and are often unavoidable.
The US EPA has indicated that even where standard recoveries are outside the method guidelines accurate quantification can still be achieved. Method CARB 428 (California Air Resources Board 1990), an equivalent of US EPA Method 23A, states the following about internal standard recoveries outside the method guidelines:
This criterion is used to assess method performance. As this is an isotope dilution technique, it is, when properly applied, independent of internal standard recovery. Lower recoveries do not necessarily invalidate the analytical results for native PCDD/PCDF or PCB, but may result in higher detection limits than are desired.
The technique employed is an isotope dilution technique, so the levels of PCDD and PCDF congeners are independent of internal standard recoveries. The effect on the reported levels, and in particular the total toxic equivalence for the affected samples, is therefore deemed minimal.
Sinclair Knight Merz testing specialists checked the results and calculations from the testing in detail.
5.4 PCDD and PCDF results reporting
PCDD and PCDF are reported according to varying conventions. Weightings are given to the individual congeners depending on the convention selected to calculate the total toxic equivalence of a sample. The results in this report are in both WHO (Van den Berg et al, 1998) and International (Kutz et al, 1990) TEQs (WHO-TEQ and I-TEQ). The I-TEQ has been used for the majority of the analysis to allow comparison with measurements and emission factors from other international studies.
TEQs are also reported three ways: with congeners below the limit of detection (LOD) assumed to be zero, half the LOD, and at the LOD. Common convention, particularly in New Zealand, is to include the half LOD in the TEQ calculations, and this was the proposed approach in the Ministry for Environment's dioxin action plan (Ministry for the Environment, 2001).
5.5 Dioxin measurement results
Table 7 reports the mass emissions of PCDD and PCDF per tonne of metal melted (i.e. emission factors for each process). The results are in µg TEQ per tonne of metal, with half the LOD included as per common convention in New Zealand and overseas. There is generally little difference between the I-TEQ and the WHO-TEQ results reported.
Due to the good coverage achieved from the test programme, the emission factor data can be assigned a medium-to-high certainty ranking, in accordance with the certainty rankings for emission factors used in the 1998 dioxin inventory (Ministry for the Environment, 2000). The samples with reference numbers S13+S14, S16+S17, S18+S19 and S29+S30 were separately analysed for particle and gas phase PCDD and PCDF. The results have been added together to give total PCDD and PCDF for the sample.
A summary of the results from the testing programme are presented in Appendix C of this report and full results are presented in Volume II.
The results for Sites F and H (samples S11, S12 and S18+S19) are substantially higher than those measured at other sites. Most sites have fabric filtration and reasonably clean scrap. The higher test results correlate to sites where significantly contaminated scrap was involved. In one case, the scrap had a high proportion of material coated in cutting oils; [Personal comment, Stuart Keer-Kerr, K2 Environmental Ltd, 2003.] in the other case there was a large amount of mixed material. The highest result (S18+S19) may also correlate with poor fabric filter performance,although the particulate test result for this site was not unusually high.
The first sample for Site H (S16+S17) is significantly lower than the second sample (S18+S19). This was due to an error in the sampling procedure, which meant that the sampling time did not cover the main melting operations (i.e. was not representative of melting). Therefore this result has not been considered in the development of emission factors for New Zealand metal melting, although the result is reported in Table 7.
Table 8 gives the results for the emission factors by plant size and Table 9 gives results by metal.
Measured emission concentrations are summarised in Table 10. Data could not be normalised by correcting to 11% O2 as per the normal convention, because the oxygen concentration at the sampling points was essentially ambient for all the sites. This is because many of the ventilation systems are open systems designed to draw in air from the general workspace, rather than directly ventilating the furnaces as an enclosed system. The air extracted from the furnace is therefore diluted by the workspace air. The precision of the oxygen measurements are such that corrections to 11% O2 are meaningless when O2 concentrations approach 21%.
The low PCDD and PCDF concentrations measured are due to dilution from workspace air. Consequently, it is not advisable to compare the results with concentration-based discharge standards.
The individual O2 concentrations for all sites are given in the Volume II report.
5.6 Particulate matter measurement results
Table 11 presents particulate matter results. All sites tested other than Site L had fabric filtration. Although the stack testers reported that the fabric filter for Site H was not functioning properly, the results are still within the range for the sites with fabric filters. Particulate matter concentrations are not necessarily an indication of pollution control performance because of dilution due to ventilation of general workplace air. This is indicated by the oxygen concentrations measured at the sampling points, which were at or approaching 21%. For Site L the results are from the monitoring workplace atmosphere using the high-volume samplers and are representative of building air.
5.7 Relationship of PCDD and PCDF to particulate matter
It is possible to obtain a general indication of the distribution of PCDD and PCDF between the particulate and gas phase by analysing the filter material and XAD traps separately. Table 12 gives the results of the four samples for which this was done.
The results indicate that the majority of PCDD and PCDF is associated with the particulate phase. It should be recognised, however, that split analysis of the filters and XAD traps may only give a crude indication of the proportion of PCDD and PCDF found in particulate. This is because the results can be affected by the characteristics of the sampling method rather than the true distribution within the flue gas.
Figure 5 is a log-log graph showing the correlation between measured levels of particulate matter and PCDD and PCDF for each sample. There is a reasonable degree of scatter in the relationship. This is not unexpected because of the range of variables and the complex factors associated with PCDD and PCDF formation. Nevertheless, low PCDD and PCDF results tend to correlate with low particulate matter results.
The results in Table 12 and the correlation in Figure 5 generally confirm that a high level of particulate control will contribute to minimising PCDD and PCDF emissions in New Zealand, although other variables such as the condition of scrap may have a larger influence.