4.1 Emission factor
Emission factors are mainly used for calculating emissions inventories. Emissions inventories tend to calculate total emissions as emissions per kilogram of fuel multiplied by total kilograms of fuel burned per time period, multiplied by the number of such wood burners in an air shed. This calculation is then summed with the emissions from all other types of wood burners, and other sources.
Appendix 1 gives the raw data from the 60 test runs. This data is presented in the form of a histogram in Figure 1, which shows that the data is positively skewed with a long tail to the right-hand side.
Figure 1: Distribution of emissions
Previous researchers (Environment Waikato, 2006; Scott, 2005) have reported the median as their preferred measure of central tendency. This may have been due to the skew of the data that they collected for their research. However, while the median may be useful as a descriptive indicator of central tendency, it is not the appropriate figure to report if one wishes to use the results to develop an emission factor, for the following reasons.
The long right-hand tail seems to be a feature of emissions from wood burners under real-life conditions. The events which contribute to the tail tend to be operator behaviours which contribute to a smouldering burn such as using wet fuel, long times between reload, or turning the heater down before a flaming burn is established. When calculating an emission factor and extrapolating to the whole air shed, we don’t want to ignore this long tail by using a median and possibly under-reporting emissions as a whole. Therefore, only the mean is reported here, specifically, the mean of the mean of each individual wood burner. This estimation method avoids biasing the overall mean estimate towards the mean of the burner with the most observations.
The mean in this study was 4.6 g/kg. Using Student’s t-distribution, the 95% confidence interval around the mean is found to be 4.6 ± 2.0 (see Appendix 2).
4.2 Comparison with pre-existing burners
The Warm Homes Project involved replacing pre-1994 wood-burning heaters with NES-compliant wood burners. The average emissions rate for the pre-1994 appliances installed in Tokoroa was 14.0 g/kg. This contrasts with the average emissions rate measured for the NES-compliant burners in this study, which was 4.6 g/kg. The p-value is 0.003 for the unequal variances t-test. A 95% confidence interval for the difference in means is 3.9 14.8 g/kg (see Appendix 2). The results imply that replacing pre-1994 burners with NES-compliant burners can contribute to a significant reduction in air pollution.
4.3 Moisture content of fuel
Households participating in the study burned fuel with a range of moisture content. It can be seen from Figure 2 that the moisture content of the fuel appears to be a major contributing factor to the observed emission rates.
Although these results support the use of fuel below a moisture content of 16%, ‘dry’ wood does not always burn with lower emissions. AS/NZS 4014.2 specifies that pine with a moisture content of 16 to 20% be used when carrying out AS/NZS 4013. As a result woodburners in New Zealand are developed to provide optimum combustion when the fuel has a moisture content in this range. This phenomenon is observed by Environment Waikato (2006) and an explanation is provided by Jay Shelton (1983).
Figure 2: Effect of fuel moisture content on emissions
4.4 Variability of results
Even under laboratory test conditions the results of emission tests of wood burners vary from run to run. This is caused by variations in factors such as the way the fuel pieces are loaded, variations in the wood, and variations in the way the logs burn in a particular test. A statistical analysis was carried out on 387 laboratory test results. When normalised to an emissions rate of 1 g/kg, they showed a standard deviation of 0.28 (see Appendix 4). This gives a measure of the variability of the results under laboratory conditions.
The results are normalised to the average emissions rate for a given heater and control setting. The corresponding standard deviation for normalised emission rates from in-home tests was 0.60 for pre-1994 burners and 0.67 for the wood burners in this study. The higher value compared to the laboratory tests results from variations in factors such as control settings, fuel, and operator behaviour that happen in real life but not in the laboratory. For the individual heaters tested in the present study, where seven runs were carried out, the result indicates that there is a 95% probability that the average result for a particular heater will lie within 50% of the measured average value.
A similar analysis can be applied to the study as a whole to assess the extent to which the average emission figure measured for all heaters tested is likely to represent the true average of a large number of NES-compliant heaters (averaging over the full range of heater types and operator behaviours). The standard deviation for results from the 60 tests reported in the current study is 0.89. This indicates that there is a 95% probability that the measured average of the 60 runs will lie within 23% of the true average.