Emissions from domestic solid fuel burning appliances (wood-heaters, open fireplaces)

Technical Report No. 5
J. Gras, C.Meyer, I. Weeks, R. Gillett, I. Galbally, J. Todd, F. Carnovale, R. Joynt, A. Hinwood, H. Berko and S. Brown.
Environment Australia, March 2002
ISBN 0 6425 4867 6

10 . Summary and conclusions

Review: The objective of this component was to review and consolidate the current state of knowledge on emissions from domestic solid-fuel-burning appliances used in Australia.

The review component of the study was conducted by the Centre for Environmental Studies at the University of Tasmania, and is included as Technical Report No. 4: Review of Literature on Residential Firewood Use, Wood-Smoke and Air Toxics. The main findings from the study are included in this section.

The literature review provides detailed coverage of published Australian research, information databases and government controls on wood-heaters and open fireplaces. The international literature has been selectively reviewed with greater focus on toxic components of wood-smoke and observed health impacts.

A good information base exists on the proportion (and recent trends) of Australian households in each state and territory using firewood for heating. The proportion and number of households using firewood as their main heating fuel are well established (15.5%, 1.1 million households), while the number using firewood as a secondary or occasional source of heating is subject to greater sampling uncertainty (6%, 0.4 million households).

The quantity of firewood burnt each year is less accurately known because of less survey data, uncertainty in householders' estimates of how much firewood they use, and the possibility of widespread underweight delivery of firewood. Estimates of annual consumption in Australia range from 3 to 5.5 million tonnes (air-dry weight).

A good information base exists on the aerosol mass emission factors for post-1993 wood-heaters when tested under laboratory conditions in accordance with Australian Standard AS4013. However, no in-situ measurements of emission factors have been made in Australia. Estimates of average emission factors for wood-heaters and open fireplaces suggest 10 to 15 grams of particles per oven-dry kilogram of wood burnt.

Very few measurements of gaseous emissions (e.g. CO, NOX) from wood-heaters or open fireplaces have been made in Australia. Nor have many emission studies of toxic components of wood-smoke been carried out prior to the current study.

Ambient air measurements, using chemical tracer species, have confirmed the anticipated significant contribution of wood-smoke to PM10 and PM2.5 concentrations in winter in many urban areas of Australia.

International studies, notably in the United States, provide a substantial data base on emission factors for wood-heaters and open fireplaces operated in homes. However, differences in firewood species and heater design lead to uncertainty in applying these data in Australia. Similarly, international data on emission factors for air toxics are available, but are subject to considerable uncertainty due to large variations in test procedures.

International and Australian epidemiological studies demonstrate short-term links between concentrations of PM10 and human health (i.e. observed increases in health impacts within a day or two of increased PM10 concentrations). It is reasonable to assume that wood-smoke contributes to this observed effect. Longer-term health impacts of wood-smoke have been observed in developing countries where women (in particular) are exposed to high concentrations of wood-smoke when cooking.

Survey: The objective of the survey component of the study was 'to obtain a sound national survey of the number and types of wood-heaters (including open fireplaces) and wood fuels currently in use throughout (southern) Australia'.

The main telephone survey, developed in conjunction with Environment Australia and Strahan Research, focused on the winter of 2000. This established that during the winter season, wood was burnt by 28.2% of the population across southern Australia (southern W.A., S.A., Victoria, Tasmania, and N.S.W. (including A.C.T). This was determined from 1007 positive respondents of the 3567 households contacted. This sampling gives 95% confidence that the response will be within 3.1% of national population values and within 6.9% of state population values.

Controlled combustion heaters were used by 78.6% of respondents. Open fireplaces were used by 19.3% of respondents. The majority of combustion heaters (57%) predate Australian standard AS4013 (seven or more years old). Only 2.4% were less than one year old (purchased since the strengthening of emission limits in AS4013:1999).

Overnight burning was practiced by 34.3% of respondents, including the 28.3% that burnt 24 hours per day.

Most operators described their loading as either about half-filling the combustion chamber (36.2%) or a few big logs with enough kindling to ignite them (44.1%). This is consistent with the most efficient burning practice as described for example by Shelton (1983).

Most respondents (77.3%) burnt hardwood, with the actual species varying from state-to-state. Further telephone surveying of wood merchants by CSIRO Atmospheric Research showed that for Melbourne, Adelaide and Canberra the main species were eucalypt, comprising redgum, followed by box species, for Perth, jarrah was dominant and with more mixed eucalypt species reported for Hobart (dependent on current supply).

Softwood use was relatively minor with 1.2% of the main survey respondents using only softwood and 13.2% using a mixture of softwood and hardwood.

Approximately one half of respondents (49.6%) to the main survey collected wood from private land and the next largest group (41.2%) purchased wood from a wood merchant. Most respondents (86%) reported that they stored their wood in a covered area before burning.

Overall the surveys indicate a tendency for heaters to be aged, with 57.1% of heaters seven or more years old and thus potentially not meeting the current, or previous emission standards. Another 38% of heaters should have met the previous standards when they were installed – at least in those states where these are mandated. This suggests a considerable scope for reduction of emissions by encouragement to move to current standard heaters.

From the responses there is an implicit acceptance or awareness of the need to not overload heaters and to burn dry fuel. The demand for high-density fuels, as indicated in fuel merchant surveying, also indicates an awareness of the longer burning properties of these fuels.

Wood-heater emissions and their controlling factors: The main component of the study involved a series of controlled burns with an objective of gaining an understanding of the emissions from a range of types of domestic solid-fuel-burning appliances using a variety of fuels under conditions of maximum and minimum (as permitted by the appliance).

Four appliances, two AS4013:1999-compliant controlled-combustion heaters, a non-AS4103-compliant heater and a fireplace insert, were used to burn a range of Australian fuel types. This included three hardwoods (eucalypts – jarrah, bluegum and redgum), softwood (Pinus radiata), both types with a range of fuel moisture (seasoning), also a manufactured fuel produced by high-pressure extrusion of mixed softwood and hardwood chips. A range of burning conditions was included, resulting in 45 separate burns, in a controlled AS4013 laboratory setting. This is the most extensive study of its type undertaken in Australia.

Concentrations of an extensive series of gas and aerosol species, primarily the toxic species on the Commonwealth's Living Cities Air Toxics priority list, and several gaseous pollutants, were determined along with supporting combustion, gas and aerosol parameters.

As well as incorporating burns closely linked to AS4013 standard procedures, which allow some links to the extensive compliance-testing database, the study incorporated a range of variations from 'standard' operating procedures. For example burns with an overloaded combustion chamber and running on a low setting – simulating overnight burns. Also burns of very green (freshly felled) timber and oven-dried low-density fuel (Pinus). The time evolution of selected emissions was also investigated.

The two AS4013:1999-compliant heaters had nominal, or type, mass emission factors of 0.9 g/kg, and 3.7 g/kg. These heaters were preconditioned first by a series of graded-intensity burns. The non-compliant heater was manufactured in 1985; its condition is best described as 'well-used'. The fourth appliance was a new open fireplace insert, which was operated within a thermally-insulating bricked surround to simulate a residential setting. This appliance was also pre-conditioned with graded-intensity burns before it was used for testing.

The study was not designed as an 'audit' of the AS4013 status of the tested heaters. Complete AS4013 tests including all flow settings and repetitions were not conducted as part of the study. The test protocol used involved only high or low airflow setting burns and the number of repetitions, along with other parameters, including the amount of fuel burnt-off to establish combustion, was varied from the AS4013 protocol. But, some general conclusions can be drawn about the relative mass emissions from the tested appliances.

The low emission (0.9 g/kg) compliant heater was operated only using seasoned eucalypt (redgum) with a low flow setting and performed better than 'rated' over three burns 0.5 ± 0.2 g/kg (dry fuel basis, values given are the mean and uncertainty ranges are ± 1 standard error of the mean). The open fireplace insert was tested for three burns using seasoned eucalypt (redgum), this also performed within the 'nominal' emission guideline. The mean emission factor for the fireplace insert was 2.3 ± 0.4 g/kg dry basis. The non-compliant heater was tested over a slightly wider range of conditions using seasoned eucalypt (redgum). This included three low flow burns, one high flow and two overloaded low burns. Taken together these six burns resulted in mean mass emission factor of 3.5 ± 0.6 g/kg dry basis. Although this nominally appears to satisfy AS4013 emission levels the efficiency of this heater was relatively low (38 ± 2%), consistent with excessive bypass or leakage.

The higher emission-compliant heater (rated 3.7 g/kg) was used for the majority of the emission testing. For seasoned eucalypt (redgum) on high flow, this gave a mean emission factor of 1.8 ± 0.6 g/kg dry basis (from four burns) and 1.4 ± 0.3 g/kg dry basis for all 10 burns of seasoned eucalypts (three species). On low flow setting this heater produced quite variable emission factors from 1.8–21 g/kg, overall averaging 12.6 ± 5.6 g/kg dry basis for three burns of seasoned eucalypt (redgum). Combining the seasoned redgum burns for high and low flow gives a mean emission factor of 7.1 ± 4 g/kg dry basis, which for one standard error falls within the 'nominal' rating. The high variability in mass emissions for the low flow settings should not be considered unusual; burning includes a number of inherently 'random' or 'chaotic' processes.

For the conditions of this study, mass emission factors were found to be consistently greater for burning softwood than eucalypt, expressed either as g/kg or g/MJ basis, (all values for Pinus radiata exceeded 7 g/kg). No outstanding difference was observed in mass emissions from the tested eucalypt species, using the same heater with high airflow (jarrah 1.15 ± 0.06 g/kg, bluegum 1.2 ± 0.5 g/kg, and redgum 1.8 ± 0.6 g/kg). The mass emission factor for the compressed manufactured fuel logs was intermediate between that of softwood and eucalypt.

Superficially, mass emissions showed a broad relationship with fuel moisture. Minimum emission factors were observed for fuel moisture content in the range of about 20–30%. But care with interpretation is necessary because burn parameters and the number of samples varied widely across the range of fuel moisture values. Strongest emissions were associated with the green softwood (54% water, wet fuel basis). A second group of consistently strong emissions occurred with dried softwood, which at 11–14% moisture was at the lower end of, or just outside, the AS4014 moisture range of 12–16%. Some of the seasoned eucalypt samples also showed high emissions for low air-flow-rate burns. Mass emission factors tended to be greatest for lower power burns, with dry softwood producing the marked exception. The most consistent relationship was between mass emission factor and combustion efficiency (C[CO2]/ΣC). Greatest mass emissions occurred for burns with the lowest combustion efficiencies and vice versa. All elevated emissions occurred when combustion efficiency was less than 0.82. No consistent relationship was evident between combustion efficiency and fuel moisture, relative to other causes of variance (such as fuel type, burning condition etc).

This work clearly confirms existing knowledge, that practices which result in a low combustion efficiency, also produce elevated mass emissions. This includes, for example, overloading the heater and not allowing sufficient time for the fuel to establish pyrolysis before reducing the airflow, also the use of very wet fuel with low airflow.

Concentrations were determined for more than 120 chemical species. These were used to derive the emission factors that are tabulated as part of this report. Summary data were derived on the basis of fuel type, irrespective of the other burn conditions and also a best estimate was derived using a weighted average to relate the experimental conditions to real-world burning conditions. Weightings for this were derived from the survey of heating appliances and their operation that was conducted as part of the study. There are inescapable limitations to this approach, primarily governed by the limited number of appliances tested and uncertainties in operating parameters of installed residential heaters. Despite these limitations the compilations provide extensive new data on mass emission factors for toxic species that can be scaled to more commonly measured factors such as the mass, CO or VOC emission factors. This is the first derivation of its type for Australian heaters and Australian fuels.

Most of the composition differences between smoke from eucalypts and softwood burns can be related to the overall mass emission factors 4.5 g/kg (for eucalypt) and 15.8 g/kg (for softwood) (and hence back to combustion efficiency). Notable exceptions include chloromethane, NO, NO2, NOX, sulfuric and acetic acids, aerosol phthalate, S and Cl, where emissions relative to aerosol mass were significantly greater for eucalypt than softwood burns. For some species the difference is more than an order of magnitude. No detectable PCDD/F was detected for the softwood burns to within the background uncertainty (1 ng/kg), whereas significant levels were detected in hardwood burns (overall, 7.5 ng/kg). PAHs were significantly greater for the softwood burns, typically by a factor of about six.

The present study is unique in its range of fuels and heater operating conditions. Nevertheless, many of the species emission factors determined in the study are comparable to those typically reported in the literature for modern heaters in other jurisdictions This includes the BTEX species benzene, toluene, ethylbenzene and xylenes for which the best estimate emission factors from this study are comparable with or less than those reported for modern north-American non-catalyst and catalyst-equipped heaters and significantly less than those from older conventional heaters. Unspeciated VOC emission factors are equivalent to values reported for modern non-catalyst and catalyst-equipped north-American heaters and CO emission factors are about double recently reported values. The formaldehyde emission factor is typical of values reported for wood-smoke. For the sixteen US EPA priority PAHs the combined gas and aerosol emission factor for PAH-16 in the present study is around one half or less that reported for modern US and non-catalyst and catalyst-equipped heaters and for the seven most toxic species PAH-7 the present study best estimate emission factor is typical of other reported values. For PCDD/F the best-estimate emission factors in this study fall within the overall range of values reported for wood heaters and open fireplaces (0.2–28.5 ng/kg) but are around a factor of 2 greater than values currently used in emission inventories in some other jurisdictions. At least part of this difference may be because of the unique, wide-ranging combustion parameters in this study. Also, much of the reported data on PCDD/F is derived from only a very small number of independent determinations.

Although not specifically designed as a process study, this work does show that PCDD/F emissions are not controlled solely by the pyrolysis–distillation mechanisms controlling the organic condensates that represent the bulk of mass emissions. There is insufficient evidence to conclude de novo production, but clear evidence that the aerosol mass emission is not a good indicator for PCDD/F emission.

Many other constituent emissions show reasonably consistent functional dependence with aerosol mass emission. This includes for example CO, unspeciated VOCs and some individual species such as toluene and xylenes as well as some aldehydes and ketones, for example acrolein and MIBK. PAH emissions overall do not correlate particularly well with aerosol mass emissions but high PAH emissions tend to occur with high mass emissions. Production of nitrogen oxides is greater for burning conditions where the aerosol mass emissions are lower and vice versa.

An overall conclusion is that control of aerosol mass emissions through AS4013 should control emissions of most of the species considered. Two clear exceptions are nitrogen oxides and dioxins, where it appears that heater design or operation parameters that reduce aerosol mass emissions work to increase the emissions of these species.

Although the actual process of combustion in a controlled-combustion heater is complex, or even chaotic, the temporal pattern of emissions observed using a series of real-time measurements tended to be reasonably well organized. In particular the degree of smoldering before widespread pyrolysis, was observed to be of particular importance for both particulate and gaseous emissions.

All of the observed species' emissions initially rose rapidly with placement of a new fuel charge onto the hot coal bed. The first species to peak was SO2, followed by VOC and CO, then CO2 and NOX. Unlike the other species, NOX doesn't follow the initial pyrolysis-induced fuel peak but rather the chamber temperature or power output. Decreased chamber temperature and reduced NOX emissions and the greatest NOX emissions were observed for burns with the highest combustion efficiencies.

Differences in time to reach peak emissions, for example between dry pine, dry redgum, green redgum, in that order, reflect the time taken for heat to penetrate the fuel, a factor related to density and moisture content.

Mass emissions also follow the pyrolysis fuel peak. For dry or seasoned fuel and high flow settings mass emissions effectively last only around 10–20 minutes. Peak mass emissions for pine (low density) were significantly greater than for more dense fuels (e.g. redgum). Low airflow rates extended the smoldering period out to one or two hours. For green eucalypt with a high flow setting the burn pattern was very similar to that with dry eucalypt. Switching from an initially high setting to a low setting with green eucalypt caused quenching followed by a slow smoldering recovery lasting around two hours. For very green pine there was extended smoldering for around four hours.

Very high ultrafine particle number concentrations were observed during the entire burn, generally being greatest in the early part of the burn. The particle size distribution typically appears lognormal. Particle sizes systematically reduced during the burn, dependent on burn type and reflecting the availability of condensates. Initially the geometric mean diameter was around 0.2 µm, progressively falling to around 0.05 µm in the latter part of the burn.

Emission guidelines and standards worldwide, whilst achieving similar goals of limiting or controlling the emission of smoke from residential wood combustion, have a low degree of compatibility in a quantitative sense. This is due to different local perceptions, political requirements, commercial considerations and development history. Current practice involves different fuels, methodologies, test species and emission goals. In general, with suitable standards most methods that are practiced can achieve the same fundamental objective of controlling smoke emission and will also largely control most of the emissions of minor species. Notable conflicts appear to be associated with nitrogen oxides and dioxins, where greater emissions may result as a consequence of 'tuning' heaters to reduce, for example, mass emissions.

Selection of emission standards should be rigorously based on a risk-benefit weighting of wood-smoke exposure and the benefits to the community of wood heating.

Overall the existing Australian standards and procedures are relevant tools for reducing exposure to toxic components in wood-smoke. Major deficiencies in the control strategy are the lack of a comprehensive risk analysis of wood-smoke constituents and the absence of any programs to assess the effects, if any, of the introduction or tightening of emission standards on exposure and risk. Several potentially effective control strategies have not been fully implemented.