Review of literature on residential firewood use, wood-smoke and air toxics

Technical Report No. 4
Environment Australia, June 2002
ISBN 0 6425 4868 4

International literature

3.1 General – International

This section provides a review of international scientific and technical literature on wood-smoke and residential wood combustion technology. Epidemiological studies of fine particle impacts on human health are included. As was the case in Australia (see Section 2.1) several significant conferences and workshops first raised/reinforced concerns amongst government and research organizations about the potential hazards of smoke from residential firewood use. For example, the US EPA sponsored conference 'Wood Combustion Environmental Assessment' (Ayer 1981) included 29 papers and workshop outcomes addressing the potential wood-smoke problem. Many similar conferences and workshops were held in the following years (e.g. Wood Stove Emissions Seminar, Omni 1985, International Conference on Residential Wood Energy, WEI 1986).

Also, as in Australia, the initial concern about wood-smoke led to broader evaluations of residential use of firewood (Core et al. 1984), community education campaigns and suggestions for mitigating the problem (see for example, Environmental Consequences of Wood and Other Biomass Sources of Energy: Coates et al. 1982; Residential Wood Combustion Emissions and Safety Guidebook: Becker et al. 1985). However, in the United States, the early government concern (e.g. Cannon 1984) led to immediate widespread research on emissions, including studies of toxic compounds in wood-smoke, studies of human health impacts of wood-smoke, and state and federal regulations to control wood-smoke. Similar activity only occurred 10 to 15 years later in Australia. The State of Oregon in the US was the first to introduce compulsory testing of emissions from new woodheaters (Kowalczyk and Tombleson 1985), developing their own test procedure and setting maximum allowable emission rates of 15 g/h (for 1986) reducing to 9 g/h by 1988. A few years later Jaasma et al. (1990) developed a method for measuring particulate and CO measurements for woodheaters operating in people's homes.

The research into wood-smoke emissions and impacts on human health has continued. For example, Shelton (1987) prepared a detailed review of regulations and emission test methods. In 1997, OMNI Environmental Services, Inc. published a list of 398 articles and reports on residential wood combustion, prepared for the US EPA (Fisher and Houck 1997).

New Zealand, in the Canterbury Region in particular, has also been active in investigation and regulation of wood-smoke. The New Zealand research parallels Australian research in many ways, with efforts directed at establishing the magnitude of the particulate emission problem and examining regulatory options for reducing the impact on air quality. In Christchurch, a high proportion (44%) of households use solid-fuel heaters (wood and coal) leading to high PM10 concentrations (Fisher et al. 1997). Average PM10 for the whole month of June (winter) is about 190 µ g/m³ (Fisher et al. 1997, Fisher 2000), far higher than any measured monthly averages in Australia.

3.1.1 Combustion

Several general texts exist on basic wood chemistry including, for example, Sjostrom's (1981) reference work 'Wood Chemistry Fundamentals and Applications'.

When thermal analysis of wood is carried out it behaves approximately as if it was a mixture of cellulose, hemicellulose and lignin. Nassar and MacKay (1984) have shown that lignin pyrolyses more slowly than other wood components and it is the lignin that supports the glowing combustion, as opposed to the flaming, phase of wood combustion. Lignin undergoes endothermic reactions (i.e. absorbs heat through chemical reactions) in the temperature range 50 to 220°C. It undergoes exothermic reactions (i.e. produces heat) in three temperature bands: 240 to 330°C, 360 to 370°C and 430°C.

3.1.2 Appliance design

3.1.2(a) General

The burning of wood is the oldest source of energy for cooking and heating, and as wood is a renewable resource, the burning of firewood should be a more sustainable form of domestic heating than the burning of oil, gas or electricity generated by coal-fired power stations. Most houses built in Australia prior to 1950 incorporated one or more fireplaces, in which wood (and sometimes coal) was burnt to heat the room by direct radiant heat from the flames and glowing coals. This was a particularly inefficient method of heating and required large quantities of firewood in the cooler areas of the country. The chimney draught generated by a traditional open fire can generate up to 15 changes of room air each hour (Hayden 2000) carrying much of the heat from the fire and the heated air in the room up the chimney.

Open fires were a major source of internal air pollution, as 'down draughts' often blew smoke back into the room under low fire conditions and this situation was often exacerbated during windy weather. This pollution was obvious in the blackening of mantelpieces, wall and ceilings of rooms where fires were regularly lit.

Wood fires have enjoyed a renaissance as the preferred method of domestic heating in Australia and many other industrialised countries, despite the availability of more convenient and sometimes cheaper alternatives. One of the factors influencing this trend has been the availability of a wide range of wood burning heaters which provide greater thermal efficiency than the traditional open fireplace and effective control of the combustion rate and heat output. This is particularly important in the cooler areas of Australia, where there is a demand for a heating system that can operate at low levels for prolonged periods, with the capacity for high heat outputs when required.

The combustion of wood is a complex process, which starts with heating the outer layers of the wood until pyrolysis of its volatile components produces sufficient flammable vapour for ignition. These vapours burn with a bright flame until the pyrolysis has been completed, when the residual charcoal starts to burn with little or no flame. Because a typical wood heater operates in a batch mode, all three of these processes may be operating at the same time in different parts of the fire.

Efficient combustion in a wood-burning heater requires an excess of air (typically twice that required for combustion), a reaction temperature exceeding 600 °C and turbulent airflow to provide good mixing with the fuel gases (Tiegs 1995b, Hayden 2000). A modest amount of excess air, as in a simple unsealed wood stove, can provide good combustion with reasonable thermal efficiency. However, in an open fire, the large quantity of bypass air leads to high heat losses in the flue gases as illustrated in Table 3.1.1. The very high airflow through an open fire also increases the total smoke emission by lowering the flame temperature and sweeping away unburned fuel gases before they can be heated to their ignition temperature.

Table 3.1.1: Effect of excess air supply on the thermal efficiency of a wood heater
Excess Air Supply 100% 500% 1000% 1500%
(open fire)
Air supply/Air required for combustion 2 x 6 x 11 x 16 x
Fraction of heat lost in flue gases 10 % 29 % 48 % 73 %
Fraction of heat delivered to room 78 % 59 % 40 % 15 %

Source: adapted from Hayden 2000

Wood burning heaters in Europe and North America during the eighteenth and nineteenth centuries were primarily developed for high thermal efficiency, where their burning rate and the heat output was controlled by limiting the air supply or reducing the draught by means of a damper in the flue. These thermally efficient 'slow combustion' woodheaters provide efficient combustion of the wood fuel under high draught conditions, but when operated with a restricted air supply, the incomplete combustion of the fuel produces copious quantities of smoke containing a noxious cocktail of toxic and carcinogenic organic compounds.

Faced with a common problem of wood-smoke pollution and environmental restrictions on the use of woodheaters, manufacturers in Australia, Europe and North America have been active in developing a new generation of cleaner burning wood stoves and heaters. These have incorporated a range of innovations to improve the combustion efficiency (particularly at low burn rates) and reduce the emission of harmful pollutants. Any informed debate regarding the future of wood fired heating in Australia's towns and cities requires a basic understanding of the operation of woodheaters, and particularly those design improvements that can reduce their contribution to urban air pollution. These improvements are reviewed in this section.

Kowalczyk et al. (1981) reported that, while good combustion principles were understood, applying them in practice was not straightforward. They pointed out in 1981 that heater manufacturers still had considerable development work to do before a practical solution to serious air pollution problems was solved. This relatively early research identified a number of general parameters influencing emissions.

Burnet et al. (1986) reported that small firebox volumes (23L) resulted in emission factors roughly half that of larger fireboxes (57 to 100L) for similar burn rates. They also noted that higher fuel loading densities (more wood in a given volume of firebox) led to higher emissions.

Design features such as pre-heated combustion air and good gas retention times in the flame zone are known to reduce smoke emissions (Todd 1981c, 1994). Other factors such as flame quenching on cool surfaces, lack of ignition points and poor mixing also contribute to higher emissions (Todd 1990b).

Zeedijk (1986) has attempted a computer model of the combustion process in a woodheater. The model simulates the pyrolysis stages of batch-loaded woodheaters and predicts the observed high emissions following addition of new wood.

3.1.2(b) Catalysts

When an airtight heater is operating with insufficient oxygen to completely burn the vapours being emitted from the main mass of burning wood, these vapours enter the flue, where they condense to form smoke or creosote deposits. If a secondary source of air is added to the exhaust gases, these vapours will burn provided the mixture is above their ignition temperature of 500°C–600°C (Tiegs 1995b, Pilcher and Ghajar 1985).

The ignition temperature of the vapour/air mixture can be reduced to around 300 °C in the presence of a platinum or palladium oxide catalyst, similar to those used to enhance the oxidation of unburned fuel vapours in car exhaust emissions (Tiegs 1995b). Experiments in the early 1980s indicated that a woodheater fitted with a catalytic converter could achieve an 80% reduction in particulate emissions relative to a conventional sealed stove (Shelton 1984a, 1984b) and 50% relative to an open fire. Allen et al. (1983) confirmed that the temperature of the catalyst had to be maintained in order to achieve reductions of emissions. They found the catalyst was only active above 425°C.

However, long-term field trials were less encouraging, with reported reductions in particulate emissions from commercial catalytic woodheaters in the range of 25% to 50% (Anon 1986a,b, Perry 1988, Hayden 2000). These and other reports (Burnett 1988a, Simons and Tiegs 1988) also indicated that the positioning of the catalyst within the heater was a critical factor in its effectiveness, and they were easily poisoned by chemicals released from burning timber treated with preservatives. In all cases, the activity of the catalysts was observed to decrease within a year of installation.

Experiments with retrofitted and add-on catalytic converters showed them to be of very limited value in reducing particulate emissions (Anon 1986b, Perry 1988), probably because the gases were cooler at entrance to the flue. One trial (Perry 1988) indicated add-on catalytic converters actually increased particulate emission levels by 3% by interfering with the gas flow in the flue. Quraishi (1987) also found increased emissions from a heater fitted with a catalyst (see Section 2.6.2).

3.1.2(c) High technology heaters

The most effective method for ensuring efficient combustion within a woodheater has been to raise the overall temperatures within the combustion zones to around 800°C, to ensure the gases leaving the main firebox are above the ignition temperatures of the fuel vapours. This has been achieved through a combination of several design changes, including:

Reports of the field trials of these 'high technology' heaters indicate that they consistently emitted the lowest levels of particulates and unburned wood gases of all the heater types tested. Reported reductions in particulate emissions range from 80% (Hayden 2000) to 50% (Tiegs 1995b, Burnett 1988a). Hayden (2000) reports that the successful operation of these heaters requires a flue with a strong draught to ensure an adequate air flow through the convoluted passageways of the heaters. In practice, this was achieved by a well-insulated flue with a minimum length of 6 metres. Heaters operated with shorter, or poorly insulated flues, failed to achieve their potential reductions in particulate emissions.

3.1.2(d) High thermal mass heaters

Originally developed in Eastern Europe, the thermal mass, or masonry stove is the traditional heating appliance in Finland, from where most of the design and performance information has been derived (TST-Group 1982, Senf 1994). The traditional masonry stove is constructed from multiple layers of fire-resistant bricks and refractory cement which provides a typical heat storage of 50kWh and a thermal delay of up to 12 hours between the fire box and the external surface (TST-Group 1982).

The stove is heated once each day, by burning 12–16kg of wood in a high intensity fire (20–40 kW) for two or three hours. When the fire has gone out, the flue is closed with a damper to reduce convection heat losses, and the mass will continue release 2–3 kW of heat over the next 24 hours. Although the internal layer of bricks forming the firebox and combustion chambers may reach temperatures of 300–500°C, the high mass of the stove and slow thermal diffusion limit the maximum external temperature to 60–80°C (TST-Group 1982, Pilcher and Ghajar 1985). In very cold weather, the stove can be fired every 12 hours, doubling the average heating power.

The masonry stove has a number of advantages as a central heating appliance when compared with continuous burning wood heaters, namely:

  1. the air pollution problems associated with slow-burning heaters are avoided by burning the day's ration of fuel under near-optimum conditions.
  2. the external surfaces of the heater never reach a high temperature and cannot present a risk of personal injury through contact burns, or a fire risk through heating adjacent flammable materials.

It is recommended that the combustion air for a masonry stove be obtained from an external source (TST-Group 1982, Senf 1994), as the air requirements of 30 L/s can produce significant room depressurisation in a modern dwelling. This can lead to chimney back flows during the die-down phase of the firing cycle, with the risk of carbon monoxide build-up.

The high thermal mass and long time constant of the traditional masonry stove would probably limit its application in Australia to the cooler regions where continuous heating of dwellings is considered desirable for a significant proportion of the year. However, the short-term heating capacity of the conventional woodheater can be combined with some of the heat-storage and safety features of the masonry stove, by constructing a close-fitting masonry wall on three sides of the heater (Pilcher and Ghajar 1985). The heat storage capacity of the masonry wall would reduce the need for the heater to be operated in the air-starved low burn mode, which is considered responsible for the majority of wood-smoke pollution.

3.1.2(e) Flue gas remediation (soil filtration)

Central power plants and other large-scale industrial processes that burn coal and other solid fuels treat their flue gases to remove particulate matter and major pollutants gases such as sulphur dioxide (SO2). While it is impracticable to install flue gas scrubbers, filters or precipitators on domestic wood heaters, similar results can be achieved by discharging the flue gases through perforated pipes in earth covered absorption trenches (Duncan et al.1982).

In the reported field trials, the soil filtration removed all visible particulate matter and 99 % of the sulphur dioxide (SO2) emissions from the flue gases produced by burning wood and high sulphur coal. The gas absorption worked best in damp conditions with no apparent disruption of the soil biota and it was proposed that the flue pipes could be laid in the seepage trenches used for domestic wastewater. However, the long-term effects of the continued absorption of toxic material on the health of the soil biota, still needs to be determined. Obviously, this method for wood smoke pollution requires forced extraction of the flue gases, which could pose a serious internal air pollution hazard if the fan failed while the heater was alight. This approach is considered more of an interesting trial than a practical solution to wood-smoke.

3.1.2(f) Pellet fuelled heaters

One of the disadvantages of woodheaters is they require regular operator intervention to load new fuel and adjust the air supply to maintain the desired heat output or optimum combustion conditions as the combustion progresses through its various stages. Automatic heat output is possible by incorporating thermostatically controlled air valves, but these tend to produce high levels of air pollution as they cycle between full burn and minimal burn conditions.

The use of wood pellets (or wood chips) as fuel allows automatic stoking of a furnace, changing the batch mode of operation to a continuous feed, where the fuel feed rate and the combustion air flow can be controlled to provide the desired thermal output, whilst maintaining efficient combustion conditions. The operation of a wood pellet fired furnace can be automated in the same manner as a coal, oil or gas fired unit, with better combustion control and lower emissions than a furnace fired by larger wood pieces. Brandon and Murray (1981), for example, measured emissions from five models of pellet fuelled heaters and found an average emission factor of 0.37 g/kg, or roughly one 20th the emission factor of a typical batch fuelled woodheater.

Wood pellets are effectively a renewable substitute for coal or fuel oil in areas where there is a plentiful supply of waste wood suitable for chipping or sawdust for processing into pellets. They are best suited to heating boilers and space heating in large buildings, such as hospitals, schools and apartment buildings. Boilers and furnaces fuelled by wood pellets require more frequent cleaning to remove ash from boiler tubes and flues, and there is a higher fire risk from ignition of wood dusts by burning embers.

3.2 Emissions – International data

The US EPA has collected emission factor data for various categories of woodheaters and open fireplaces (1996a). This database represents a collection of much of the US emission measurement research (e.g. Pacific Environmental Services 1983). A table of emission figures is presented in the summary (section 4) of this review.

3.2.1 Particulates

Particulate emissions from woodheaters and open fireplaces are measured using a variety of approaches (Shelton 1985, Shelton and Jaasma 1985, Quraishi 1985a, Quraishi 1985b) including

In all these methods, it is important to cool the flue gas to near ambient temperature in order to condense volatiles for filter capture.

The US Federal Register details the emission measurement methods and emission limits for woodheaters (US EPA 1988) with background information provided in US EPA (1987).

Butcher and Sorenson (1979) used a total smoke capture method to measure particles from two heater models burning Red Oak and Eastern White Pine with variable moisture content. The overall average emission factor was 9.2 g/kg, but wet pine (42% moisture) averaged 12.9 g/kg compared to an average of 8.4 g/kg for dry oak (9 to 24% moisture). In subsequent testing, Butcher and Ellenbecker (1982) measured particulate and carbon monoxide emissions from a single model of woodheater using a modified dilution tunnel approach. Ten measurements were made with Red Oak and Eastern Hemlock. Emission factors ranged from 1.6 to 6.4 g/kg for particles and 100 g/kg for CO. Higher particle emissions were obtained at slower burn rates.

Cooper (1980) reviewed early (pre-1979) measurements of woodheater and open fireplace emissions. Particulate emission factors for woodheaters ranged from 1 to 24 g/kg with an average of 8.5 g/kg. For open fireplaces the average emission factor was 9.1 g/kg with a range of 7.2 to 12 g/kg. Hall and DeAngelis (1980) reported on some early work on wood-smoke emissions carried out for the US EPA.

Early studies of particulate emissions from open fireplaces (US EPA 1975) using four wood species indicated an average emission rate of 76 g/h and an emission factor of 10.4 g/kg. The average particle size was reported as 3 micrometres (measured using a cascade impactor).

Kowalczyk et al. (1981) measured emissions from seven woodheaters burning air-dry Douglas fir using the US EPA Method V sample technique. The average emission factor from 13 tests was 24.6 g/kg, with the highest single test giving a value of 74.6 g/kg. These extremely high emission factors probably reflect the emissions from 'conventional' (i.e. non-certified) woodheaters, despite the claim that these were low emission designs.

Dasch (1982) measured particulate emissions from open fireplaces burning hardwoods, softwoods and synthetic logs in the laboratory and in people's homes. The averaged results indicated emissions factors of 9 ± 4 g/kg for softwoods, 10 ± 5 g/kg for hardwoods and 15 ± 8 g/kg for synthetic logs.

3.2.1(a) Size distribution

Dasch (1982) found the mass median diameter of wood-smoke particles from open fireplaces was 0.17µ m.

Hueglin et al. (1997) measured wood-smoke particle size distributions using a differential mobility analyser. They found that the size distribution changed as a batch of firewood went from the ignition phase (diameter peak at 0.25µm), to the steady flaming stage (diameter peak at 0.15µ m), and finally the charcoal-burning phase (diameter peak at 0.06µm).

Kleeman et al. (1999) measured the size distribution of wood-smoke particles from pine, oak and eucalyptus wood. Samples of smoke were obtained from a dilution tunnel and analysed using laser counting and differential mobility techniques as well as particle size impactors. The size distribution was similar for all wood species, showing a peak at 0.1 to 0.2µm diameter, with almost all particles less than 1µm diameter.

3.2.1(b) Field measurements

All the above measurements of particulates were carried out in laboratories, where operating conditions and fuel were closely controlled. Emission measurements determined in people's homes are of greater significance to regulators.

Burnet (1988 a,b) reports on measurements of particulate emissions made in 42 wood-burning households in Vermont and New York between 1985 and 1987. The emission rates are summarised in Table 3.3.1. Burnet notes that emission rates are higher than laboratory tests would suggest and that there were large variations from one household to another. The low emission 'high-tech' heaters performed the best. The data presented in the published reports did not allow conversion to emission factors.

Correll et al. (1997) monitored 13 woodheaters in people's homes over four years in Crested Butte, Colorado. Seven catalytic and six non-catalytic heaters were monitored. The results are shown in Table 3.2.1. The authors concluded that significant deterioration of the catalytic heaters led to their relatively poor performance. The US EPA (1996a), in their compilation of emission factors, uses values of 9.8 g/kg (certified low-emission), 15.3 g/kg (conventional uncertified), and 10.2 g/kg (catalytic).

McCrillis (1990) has reviewed the field measurements carried out between 1985 and 1987. He reported that average emission rates for low emission, non-catalytic woodheaters used in people's homes was 12.3g/h compared to 6 g/h in laboratory testing.

Table 3.2.1: Field testing of particulate emission rates and particulate emission factors
Heater type Emission rate (g/h)
Burnet (1998 a,b)
Emission factor (g/kg)
Correll et al. (1997)
Certified (low emission) woodheaters 13.4 (range 3 to 24) 9.8
Conventional woodheaters (i.e. non-certified) 20.1 (range 11 to 30) 22.1
Catalytic fitted woodheaters 16.4 (range 8 to 26) 22.8
Conventional woodheaters + retrofit catalysts 17.9 (range 9 to 27)  

3.2.2 Toxics

The gaseous and particulate components of wood-smoke have complex chemical structures. Primary emissions include volatiles driven from the wood as the wood is heated, secondary chemicals are formed through reactions in the hot combustion chamber and the flames, and further transformations take place when the emissions enter the atmosphere. One aspect of this complex mix of chemicals of particular interest to researchers is the presence of toxic components of the smoke. The presence of many toxic compounds in wood-smoke is now well established, however, the concentrations of individual compounds are highly variable so it is difficult to establish emission factors with any certainty. The United States Environmental Protection Agency (US EPA 1996a) publishes a list of selected emission factors (Table 3.2.1) with the warning that '[d]ata show a high degree of variability within the source population. Factors may not be accurate for individual sources' (US EPA 1996a, p1.10-5). The potential toxicity of wood-smoke was immediately apparent because of the quite extensive data on the toxicity of soot. Barfknecht (1983) prepared an extensive review of the toxicology of soot (from both fossil fuels and biomass), demonstrating the heath risk and toxic compounds present.

Salomaa et al. (1985), working in Finland, analysed the genotoxic potency of wood-smoke and compared it to cigarette smoke. Wood-smoke from a small, horizontal-baffle woodheater burning birch and spruce was shown to be in the same range (i.e. highly genotoxic) as condensate from cigarette smoke. Emission measurement of wood-smoke in Sweden indicated a high proportion of volatile hydrocarbons from woodheaters were carcinogenic, including benzene, 10 to 20% by weight of non-methane hydrocarbons; and 1,3-butadiene, 1 to 2% (Barrefors and Petersson 1995).

The toxic nature of wood-smoke has also been demonstrated in some animal studies. Kou et al. (1997), for example, in studies of wood-smoke inhalation by rats, showed an immediate change in respiration. The authors attributed this change to the increased presence of the hydroxyl radical in the wood-smoke.

Respiratory irritants, such as various aldehydes, are emitted from woodheaters and open fireplaces with total aldehyde emission factors in the range 0.6 to 2.3 g/kg from open fireplaces (Lipari et al. 1984).

Zhang and Smith (1999) have developed carbonyl emission factors for various fuels, including firewood, used for cooking in China. Biomass combustion produced a wide range of carbonyl compounds, sufficient to cause acute health effects in typical village use.

3.2.2(a) Mutagenicity

As part of the Integrated Air Cancer Project in the United States, the mutagenic properties of wood-smoke and motor vehicle emissions were investigated for ambient and indoor air. This project generated a number of the studies described below. Mutagenicity is usually measured as direct mutagenicity (without S9 activation) or indirect mutagenicity (with S9 activation, which causes the substance to metabolise). Two strains (TA98 and TA100) of Salmonella typhimurium bacteria are used in most of the reported testing, with mutagenicity reported in units of revertants/mg or revertants per m³ for air samples.

McCrillis et al. (1992) report on a series of woodheater tests conducted to investigate wood species, wood moisture, burn rate and fuel load on mutagenicity and PAH (see 3.2.2(b)). Mutagenicity was measured using an Ames plate method. Based on 90 or 95% confidence intervals, they concluded that:

The authors conclude that 'The results also show that there is wide variability due to small, uncontrollable differences from one fire to the next and to [heater] design.' (McCrillis et al. 1992 p.691).

Kleindienst et al. (1986) performed experiments on dilute wood-smoke to compare the mutagenicity of the gas and particle phases of the smoke, with and without the presence of NOx. They concluded that while the gas and particle phase components of wood-smoke show little direct mutagenic activity (0 to 0.3 rev revertants/µg), after treatment with NOx mutagenicity increases considerably (0.2 to 8.5 revertants/µg). They also note that the gas phase products are 3 to 30 times more mutagenic than the particle phase acting on TA100 strains of Salmonella typhimurium, but this difference did not show up with the TA98 strain.

Table 3.2.2: Table 3.2.2 Emission factors for various compounds for woodheaters
Compound Emission factor (g/kg)
Ethane 0.735
Ethylene 2.245
Acetylene 0.562
Propane 0.179
Propene 0.622
i-Butane 0.014
n-Butane 0.028
Butenes 0.596
Pentenes 0.308
Benzene 0.969
Toluene 0.365
Furan 0.171
Methyl Ethyl Ketone 0.145
2-Methyl Furan 0.328
2,5-Dimethyl Furan 0.081
Furfural 0.243
o-Xylene 0.101
PAH total <0.250
Acenaphthylene 0.016
Anthracene 0.005
Benzo(a)Pyrene 0.003
Biphenyl 0.011
Chrysene 0.005
Fluoranthene 0.004
Fluorene 0.007
Naphthalene 0.072
Phenol <0.001
Pyrene 0.004

Note: The numbers are for non-catalytic or conventional woodheaters. A more detailed list is provided in the discussion Section 4.1.1.

Source: selected from US EPA (1996a).

Lewis et al. (1988) report on linear regression receptor modelling used to apportion mutagenicity in ambient fine particles in winter in Albuquerque. Wood-smoke was found to contribute more extractable organics and mutagenicity than motor vehicle emissions, however the mutagenic potency of motor vehicle sourced organics was three times greater than that of wood-smoke. In their discussion, Lewis et al. (1988) note that mutagenic potency of wood-smoke will alter in the atmosphere depending on ambient conditions, so direct emission measurements of mutagenic potency may differ from ambient measurements.

This change in mutagenicity of wood-smoke in outdoor air has been studied by Kamens et al. (1984, 1985, 1986, 1989, 1990) and Bell and Kamens (1986) using a retention chamber. The smoke was aged with and without sunlight, O3 and NO2. The mutagenicity increased 2 to 10 fold in the presence of O3 + NO2, but much less under other conditions. Before aging, mutagenicity ranged from 0.03 to 0.34 revertants/mg (direct: -S9) and 0.12 to 0.98 revertants/µg (indirect: +S9). After aging, mutagenicity ranged from 0.02 to 2.32 revertants/µg (-S9) and 0.18 to 1.1 revertants/µg (+S9).

Moller et al. (1985) report on mutagenicity of particles collected in a small Norwegian village where wood burning contributes significantly to winter particle levels. Relatively high winter average PM3 levels of 51 µg/m³ are observed (maximum 101 µg/m³). The ambient particle mutagenicity ranged from 0.2 to 1.3 revertants/µg.

Alfheim and Ramdahl (1984) have carried out indoor measurement of mutagenicity of particles. Sampling was conducted in an isolated (rural) house in Norway that was heated with an enclosed woodheater, an open fireplace and electricity. Mutagenicity was only detected indoors when the open fireplace was used, or when the residents smoked cigarettes. Mutagenicity was detected outside when the open fireplace or the woodheater was used, but not when electric heating was used. The authors conclude that the effects of using an open fireplace on indoor mutagenic activity are 'moderate' when compared to tobacco smoking.

To summarize, the mutagenic potential of wood-smoke in ambient air is reported (Lewis et al. 1988) as 0.12 to 1.3 revertants/µg in the US, with some higher results in Scandinavia (0.2 to 4.4 revertants/µg). The gas phase emissions may be more mutagenic than the particle phase. Presence of NOx significantly increases mutagenicity. Use of an open fireplace (but not an enclosed woodheater) increased indoor mutagenicity to levels comparable to indoor cigarette smoking, although this conclusion is based on measurements in just one house.

3.2.2(b) PAH

Polycyclic aromatic hydrocarbons (PAHs) result from incomplete combustion of fuel and are almost entirely anthropocentric in origin in urban atmospheres. They are considered hazardous air pollutants. There are many PAH compounds including: benzo[a]anthracene, benzo[a,h]anthracene, benzo[b]fluoranthene, benzo[k]fluoroanthene, benzo[ghi]perylene, benzo[a]pyrene, chrysene, fluoranthene, indeno[c,d]pyrene, and pyrene.

Moller et al. (1985) measured ambient concentrations of PAH in a small town with winter wood-smoke problems. Relatively high average PAH concentrations of 158 ng/m³ were measured in winter, with a maximum of 497 ng/m³.

A small Norwegian town with some wood used for heating was observed to suffer high PM3 concentrations in winter (average 51 µg/m³, range 31 to 101) together with high PAH (average 158 ng/m³, range 24 to 494) (Ramdahl et al. 1984).

Lee et al. 1977 compared PAH emissions from burning coal, wood and kerosene. They found greater concentrations of alkylated PAH from coal than the other two fuels, but lower concentrations of high molecular weight species from coal than the other two fuels.

Kamens et al. (1986) have demonstrated that PAH in wood-smoke decays in the presence of sunlight, but the decay rate is slower at low temperatures. At 20°C the half-life of various PAH species is 30 to 60 minutes, while it drops to 'many hours and even days' at -7°C. The authors conclude that there will be some decay in mild winter regions depending on the available sunlight.

McCrillis et al. (1992) found higher emissions of PAH when burning softwood (southern yellow pine) than hardwood (oak). They found no correlation between the PAH emission factor (g PAH per kg wood burned) and burn rate. They also noted a decrease in PAH emissions at high altitude (825m).

Knight et al. (1982) conducted emission tests on 5 woodheaters burning hardwood (oak) under various burn conditions. The PAH emission factor for the five heaters ranged from 47 µg/kg to 251 µg/kg (overall average 148 µg/kg). The PAH emission factors were higher when larger loads of fuel were burnt and lower in catalytic heaters than conventional heaters.

Alfheim and Ramdahl (1984) measured indoor PAH levels in a house in Norway with a woodheater and an open fireplace. There was a very slight increase in indoor PAH when the woodheater was operating (5 to 16 ng/m³), which they suggested might be associated with infiltration from outside. When the open fireplace was in operation, much higher PAH concentrations were measured (150 to 206 ng/m³). Cigarette smoking indoors led to PAH concentrations of 37 to 192 ng/m³.

3.2.2(c) Benzo(a)pyrene

Harkov and Greenberg (1985) reviewed emissions of benzo(a)pyrene from residential firewood use and found a range of 27 to 6300 ng/Btu (0.026 to 6.0 mg/MJ) with an average value of 227 ng/Btu (0.215 mg/MJ), which converts to an average emission factor of 0.0034 ng/kg.

3.2.2(d) Dioxins

A Danish study of four woodheaters, including three commercial models and one experimental model, measured an average dioxin emission factor of 1.9 ng TEQ/kg (Vikelsoe et al. 1994). The study(24 experiments, 3 or 4 burn cycles each) measured dioxin using the TEQ2 (Nordic) method. This research investigated the four heaters operated under 'normal' (slow to medium burn rate) and 'optimal' (minimum CO emissions) firing conditions and with three firewood species. The authors concluded that there was a statistically different emission of dioxin as a function of heater model, wood species and operating condition. Interestingly, they did not find any correlation between dioxin emission and total hydrocarbon emissions. Also of interest was the observation that one heater showed much higher dioxin emissions when operated in the optimal mode rather than normal.

Bumb et al. (1980), in their pioneering work on dioxins, found TCDD in the soot/creosote collected from two residential open fireplaces where only wood had been burnt. The TCDD concentration in the soot was up to 0.4 ng/g.

The Inventory of Sources of Dioxin in the United States adopted an emission factor of 2ng TEQ/kg, apparently largely on the basis of the above Danish study. Houck (1998) was critical of the US Inventory figure on the grounds of its uncertainty. He pointed out that the Danish study found significant variation in emission factors as a function of heater design and wood species and so, he argued, it was inappropriate to use this number for emissions in the United States.

3.2.2(e) Chemically treated wood

Several studies of firewood from chemically treated trees have been carried out in the United States (e.g. Bush et al. 1987a,b, Newton 1986, Woolston 1986). All authors concluded that the use of herbicides to kill trees did not present a health risk when the wood from the treated trees was burnt. Organic herbicides are degraded in the combustion process. Arsenic based herbicides do not spread through the tree and Woolston (1986) concluded that arsenate was present in particulate emissions but if wood more than 0.3 metres from the point of injection was burnt, the health risk was minimal.

These tests of treated wood do not apply to CCA (copper, chrome, arsenic) treated timber where much higher concentrations are present.

3.3 Ambient air quality

Two main approaches have been used to identify the contribution of residential wood-smoke to total particulates in ambient air. One is the emissions inventory approach, where particulate emission factors for all known, significant sources within an air shed are estimated and used to attribute the measured particulate concentrations to each source. The other is to use unique chemical markers associated with individual sources, including residential burning of firewood, to estimate the contribution from each source. Both approaches have limitations.

A study in Eagle River, Alaska, a community of 20 000 with about 10% of its heating needs met by firewood, compared an emissions inventory approach to estimating wood-smoke with a chemical source apportionment based on measured particulates (Myers 1985). The emissions inventory, using an emission factor of 10 g/kg, overestimated wood-smoke by about a factor of two when compared to actual observations and source apportionment. The study was based on TSP rather than PM10 measurement, but illustrates that emission inventory analysis of wood-smoke may be subject to significant error.

Tombleson et al. (1983), however, found good agreement between chemical source apportionment and emission modelling approaches for Medford, Oregon, USA where the average annual contribution to respirable particulates from woodheaters was calculated as 19.4 µg/m³ using the emission model and 21.5 µg/m³ using the chemical marker approach.

More recently, Chow et al. (1995) compared chemical mass balance and inventory approaches to apportion sources of PM10 in San Jose (which includes San Francisco). Agreement between the methods was poor with the inventory suggesting about 10% of PM10 was from residential wood combustion and the chemical mass balance approach indicating about 45%. PM10 concentrations were quite high at most of the 26 monitoring sites (four year averages in the range 30 to 55 µg/m³). The highest 24-hour PM10 concentrations were 165 µg/m³ at two sites, 155 and 153 µg/m³ at two other sites.

3.3.1 Chemical tracers for wood-smoke

Hough and Kowalczyk (1983), using chemical mass balance techniques, attributed 66% of respirable particles in Medford, Oregon, USA to residential firewood use. The average annual respirable particle concentration was 46µg/m³. About two thirds of residents used firewood for heating, each burning about 3 cords per year (approximately 5t/y). Tombleson et al. (1983) report on a study conducted in Portland, Oregon using a similar chemical marker approach where residential wood combustion is estimated to contribute 75% of respirable particles on a high pollution day in winter.

Benner et al. (1995) used 1,7-dimethylphenanthrene (a PAH emitted when burning softwoods) as a tracer to distinguish between wood-smoke and motor vehicle emissions in Boise, Idaho. They also used a radiocarbon measurement approach to distinguish these two possible sources, obtaining good correlations between the two methods. Potassium has also been used as a tracer to identify wood-smoke, using the property that wood-smoke potassium is more water-soluble than potassium from soil (Calloway et al. 1989). Khalil et al. (1983) and Edgerton et al. (1986) consider methyl chloride a good indicator of wood-smoke from residential woodheaters and open fires.

Hawthorne et al. (1988, 1989, 1992) used methoxylated phenols to characterise wood-smoke from residential sources. Khalili et al. (1995) consider a mix of PAH species can be used to identify wood-smoke.

Simoneit et al. (1993) suggest that for poor combustion conditions, certain biomarkers could be used to identify the type of biomass burnt (e.g. hardwood or softwood) as some higher molecular weight compounds escape unaltered from the combustion stage. Using this technique, it was possible to show that conifer emissions dominated suburban smoke aerosols in three towns in Oregon (Standley and Simoneit 1990).

Zeedijk and Bubbert (1986) suggest that potassium might be used as a tracer for smoke from residential wood combustion.

3.3.2 Ambient wood-smoke in developed countries

Burton and Senzel (1984) report on studies in eight cities in north-west United States where residential wood use was estimated to contribute up to 93% (Yakima, Washington), with an average of about 67%, of PM2.5. This study included measurement of several PAH compounds. Table 3.3.1 summarises the measured PAH concentrations for residential monitoring sites over the winter of 1980/81.

Table 3.3.1: Average PAH concentrations in ambient air at residential sampling sites in NW United States in winter based on fifteen 24-hour samples
Compound Mean (ng/m³) Standard deviation Maximum (ng/m³)
Benzo(a) anthracene 19.5 17.3 50.7
Benzofluoranthenes 7.4 5.2 18.1
Benzo(a)pyrene 4.1 3.1 11.1
Fluoranthene 10.6 13.2 42.2
Pyrene 14.4 18.9 69.9
Dibenzanthracenes 0.3 0.4 1.4
Benzo(g,h,i)perylene 2.7 1.9 6.6

Source: Burton and Senzel 1984

Murphy et al. (1984) observed relatively high concentrations of TSP and benzo(a)pyrene in a mountain resort in Colorado where woodheaters were prevalent. At the worst of four sites, TSP averaged 61 µg/m³ and BaP averaged 7.4 ng/m³. Sexton et al. (1984a, 1985) measured respirable particles and PAH in a community in Vermont where half the households used firewood and found relatively low concentrations of respirable particles (average 24 µg/m³). BaP showed a mean concentration of 0.8ng/m³ (range 0.2 to 2.4).

Harkov and Greenberg (1985) measured benzo(a)pyrene at 27 locations in New Jersey. They found average annual concentrations of 0.6 and 0.3 ng/m³ in urban and rural areas respectively. Winter B(a)P concentrations were roughly 10 times greater than summer concentrations. The authors attributed most of the B(a)P to residential wood combustion.

3.3.3 Indoor wood-smoke in developed countries

Moschandreas et al. (1980) measured indoor particles (TSP and PM3.5), benzo(a)pyrene and carbon monoxide in three homes with woodheaters in Boston, USA. This study found very high levels of PM3.5 (24 hour) when woodheaters were in use (range 14 to 160 µg/m³, average 85 µg/m³). On days when woodheaters were not used the indoor PM3.5 concentrations were lower (6 to 70 µg/m³, average 30 µg/m³). Outdoor concentrations ranged from 8 to 54 µg/m³. Benzo(a)pyrene concentrations reached 8 ng/m³ (24 hour). No information was presented on the types of woodheaters in use in the homes. The authors concluded that woodheaters may lead to increased levels of respirable particles and B(a)P indoors and may present a health risk.

Traynor et al. (1987) measured indoor TSP, CO and PAH for airtight (3) and non-airtight (1) woodheaters. They found minor emissions of particulates and CO during door opening for the airtight heaters. The non-airtight heater showed continuous emissions to indoors under certain operating conditions. For the airtight heaters indoor CO concentrations reached 4 ppm, TSP ranged from 24 to 71µg/m³. Under normal operating conditions, the non-airtight heater led to indoor CO up to 8ppm and TSP ranged from 30 to 650 µg/m³. The non-airtight heater contributed high PAH levels indoors.

Mutagenic activity in homes heated by woodheaters, open fires and central heating was measured in The Netherlands (van Houdt et al. 1986). The results were inconclusive for woodheaters but an open fire consistently led to increased mutagenicity in indoor particulates.

Sexton et al. (1984b) measured indoor respirable particle concentrations in homes with and without woodheaters and found no statistical difference. They concluded that airtight woodheaters can be installed, operated and maintained with negligible release of particles to the indoor environment. Sexton et al. (1984c) also report that personal exposure to respirable particles was higher than either indoor or outdoor particle concentrations.

Quraishi and Todd (1987) reviewed the effects of domestic firewood use on indoor air quality and concluded that there was strong evidence that use of open fireplaces did contribute to indoor air pollution, but the evidence was contradictory in the case of enclosed woodheaters.

3.4 Human health

The impact of wood-smoke on human health can be measured or inferred in various ways. Direct observation of morbidity and morality of humans exposed to measured or inferred concentrations of wood-smoke have been carried out, mainly in indoor environments in developing countries where smoke levels are high and overwhelm possible confounding factors (see section 3.4.4). Some indoor observations in industrial countries have also been reported (see section 3.4.3). A few studies of observed health impacts from outdoor wood-smoke are also reported (see section 3.4.2).

The majority of studies linking wood-smoke to human health are based on indirect or inferred relationships such as

3.4.1 Epidemiology – particles

Air quality management is founded on the principles of protection of human health, property and visual amenity. It is not surprising, therefore, that periods of very poor air quality can be linked to increases in morbidity and mortality. However, the question of whether or not a lower cut-off value for pollutant concentrations exists, below which no harmful effects occur remains a topic of debate. Large epidemiological studies offer one means of trying to establish correlations between human health and measured air pollutants. In the United States, such studies indicate 1.6 million people live in areas where air pollution has a significant impact on health (Neher and Koenig 1994).

Over the past decade, special attention has been paid to correlations between health and respirable particles. Many studies have found significant correlation between health and the concentration of particles with diameters less than 10 micrometres (PM10). Since wood-smoke is comprised of fine particles and is known to be a significant source in urban areas where firewood is used for domestic heating, the concern about PM10 has focussed attention on this source of pollution. For this reason, a selection of the many published studies demonstrating the health impacts of respirable particles is included in this review.

Despite the number of studies that demonstrate a relationship between particulate concentration and health, there is some concern that confounding factors may have not been adequately dealt with (Vedal 1997). Vedal also urges caution because of difficulties in measuring actual exposure of individuals.

Much of the predicted health impact of wood-smoke is based on the epidemiology studies reviewed below. For example, Foster (1996) estimates the number of deaths and hospitalisations attributed to wood and coal smoke in Christchurch, New Zealand (population 250 000) on the basis of PM10 concentrations. She concludes that high PM10 concentrations are responsible for 21 to 29 extra deaths per year and 35 to 45 extra hospitalisations. Such estimates give strong support to programs aimed at reducing particulate emissions from domestic heating sources.

A large body of evidence has established a strong association between the prevalence of airborne particles and increased levels of human morbidity and mortality (US EPA, 1996b)3 . As the Final Impact Assessment for the Air NEPM notes (NEPC 1998, p. 120):

Over the past decade, evidence that human exposure to inhalable particles can result in significant increases in both morbidity and mortality has become overwhelming…

Table 3.4.1 provides a summary of the results of several studies that have been conducted on the relationship between particulate matter and adverse health effects.

Several important points can be gleaned from the results of these studies.

  1. Particulate pollution has been associated with a range of adverse health affects. Epidemiological research conducted over the past 25 years has consistently demonstrated that increased levels of airborne particles are statistically related to increases in various respiratory, cardiopulmonary, and cardiovascular diseases and mortality from a variety of causes. Recent studies have suggested that particle pollution may cause cell injury and apoptosis (i.e. programmed cell death) of human alveolar macrophages and may have genotoxic effects (Lewtas, 1993; US EPA, 1996a,b).
  2. Elevated particulate concentrations have been associated with:
    • increased total mortality;
    • increased respiratory deaths;
    • increased cardiovascular deaths;
    • increased cancer deaths;
    • increased risk of premature births and infant mortality;
    • increased risk of pneumonia;
    • increased risk of postneonatal mortality from respiratory disease and sudden infant death syndrome;
    • increased hospital admissions and emergency room visits;
    • increased hospital admissions, emergency room visits and surgery for respiratory and cardiovascular conditions;
    • exacerbation of asthma attacks, increased bronchodilator use and increased hospital admissions associated with asthma attacks;
    • increased pneumonia, bronchitis and chronic obstructive pulmonary disease;
    • increased respiratory symptoms in both the lower and upper respiratory tract;
    • decreased lung function;
    • increased incidences of rhinitis;
    • increased absenteeism; and
    • increased number of days of restricted activity.
  3. There is currently no evidence that there is a threshold level at which particulate pollution is not associated with adverse health effects. Low concentrations of particulate pollution have been linked to mortality and morbidity from a number of causes.
  4. Despite the wealth of evidence concerning the strength of the statistical relationship between increases in particulates and detrimental health effects, the precise nature of the biological cause-and-effect relationship between these factors is currently unknown (Dockery and Pope, 1994; Simpson et al., 1997; Lippmann et al., 2000). Some researchers have suggested that particulates may not be directly responsible for the observed health effects (Simpson et al., 1997) and that there is a possibility that particulates may be a proxy for another form of unidentified pollution. Further, there is conjecture as to whether one or more components of particulate matter may be responsible for the observed health effects (Lippmann et al., 2000). However, the weight of evidence supports the notion that there is a direct relationship between particulate pollution and adverse human health effects.

    It is generally understood that hair follicles and mucous membranes located in the extrathoracic airways (nose, mouth, and larynx) are capable of capturing and filtering larger (greater than 10µm) particles. However, fine (less than 10µm) particles are able to pass through these protective devices and penetrate into lower parts of the respiratory system. This can result in the deposition of these particles in the airways of the tracheobronchial and alveolar regions of the respiratory system. At this point, there is a range of hypotheses on the mechanisms that produce the observed health effects. These include the following.
    • Particulates (particularly those with a high content of soluble metals or organic matter) can cause inflammation, increase permeability and cell damage in the lung (Lay et al. 1999). This can inhibit the operation of the respiratory system and may induce oxidant production and the release of intercellular signalling molecules (cytokines) that trigger observed adverse health effects (Lay et al. 1999). It may also trigger the release of natural chemicals into the blood stream that could result in the coagulation of the blood (Montague, 1995). This would lead to an increase in the risk of heart disease and strokes.
    • The presence of particulates in the respiratory system can trigger cells (particularly in epithelial cells) to synthesise new DNA and produce oxidants (Raloff 1998, Lay et al. 1999). This process can involve the activation of the c-jun gene, which can act as a proto-oncogene. In certain circumstances, proto-oncogenes can foster the growth of tumour cells and cancer (Raloff 1998).
    • Diseased lungs may alter dosimetry and lead to concentrated deposition of particles in the lung. This may stifle the operation of the lung and result in localised inflammation (Bennett et al. 1996).
    • Particulates may augment responses to antigens in allergic humans (US EPA 1996b). This may exacerbate asthma and produce other respiratory responses.
    • Condensed organic matter found in certain particles can be carcinogenic and mutagenic in short-term bioassays (US EPA 1996a,b).
    • Swelling and increased permeability in the respiratory tract may result in immune deficiencies and stifle the ability of the body to respond to bacterial and viral respiratory infections (Montague 1995).
    • Ultra-fine particles are able to penetrate deep into the tracheobronchial and alveolar regions of the lung. Where particles containing toxins are able to penetrate into the alveolar, they are more easily transferred into the blood stream (US EPA 1996b).
  5. There is strong evidence that certain groups within society are particularly susceptible to the detrimental health effects associated with particulate pollution. Subpopulations that have been identified as being particularly vulnerable include the elderly, the young, the unborn and those with pre-existing respiratory or cardiopulmonary diseases (Boezen et al. 1999; Simpson et al. 1997; Saldiva et al. 1995; Xu et al. 1995).
  6. There is evidence that the chemical composition and the manner in which the particles are formed may influence their effects on human health (US EPA 1996b). A number of studies have suggested that particles with a high content of soluble metals and organic matter have higher toxicity than other particles (Prichard et al. 1996). Attempts to measure the association between various components of particulates and observed health effects have been stifled by confounding between different pollutants (Lippmann et al. 2000).
  7. The observed statistical association between particulate concentrations and human health do not appear to be the result of the presence of other pollutants.
  8. There is considerable support for the hypothesis that fine particles (those with a diameter of less than 2.5 µm) may be more deleterious than course particles (US EPA 1996a,b). This is thought to be associated with the tendency for fine particles to be deposited over a greater surface area of the lung than larger particles and to be retained for longer periods of time in pulmonary tissue (US EPA 1996b). Recent research in the US on the relationship between PM10, PM10–2.5 and PM2.5 and various health effects found that all three PM mass indices had similar relative risk factors and that PM10–2.5 and PM2.5 were not highly correlated in either correlation coefficient or factor analysis (Lippmann et al. 2000). These results suggest that there may be merit in monitoring the levels of both coarse and fine particles.
Table 3.4.1: Summary of selected international published papers dealing with epidemiological studies linking particulate air pollution to human health impacts
Authors Key Findings
Lave and Seskin (1973)
  • Using data from 117 metropolitan areas in the USA in 1960 and 1961, estimated that increase of 10 µg/m³ in total suspended particulate (TSP) concentrations associated with increase of 0.41 per 10 000 in the total mortality rate.
  • Increased TSP concentrations were also associated with infant mortality, tuberculosis deaths and asthma deaths.
Dockery et al. (1989)
  • Studied effects of elevated particulate concentrations (measured using TSP, PM15, PM2.5 and fine fraction aerosol sulphate ('FSO4')) on 10–12 year old white children in 6 eastern USA cities.
  • Found that reported rates of chronic cough, bronchitis and chest illnesses were positively associated with all measures of particulate exposure.
  • No significant associations were found between particulate concentrations and the reported rates of asthma, persistent wheezing, hay fever, or non-respiratory illness.
  • No significant associations were found between particulate concentrations and pulmonary function as measured by forced vital capacity (FVC), forced expiratory volume at 0.75 of a second (FEV0.75), forced expiratory volume at one second (FEV1) and maximal midexpiratory flow (MMEF).
Chestnut et al. (1991)
  • Elevated particulate levels linked to days of restricted activity, work loss, daily emergency room visits, emergency hospital admissions and mortality.
  • Found statistically significant relationship between TSP and two measures of lung function: FVC and FEV1. Increases in TSP were linked to falls in both FVC and FEV1. Stronger association between TSP and FVC measures than between TSP and FEV1.
  • No threshold level was found at which no association could be detected between TSP and lung function.
  • No clear evidence on the precise biological cause of the link between decreased lung function and TSP exposure.
Penna and Duchiade (1991)
  • In Rio de Janeiro, a statistical association was found between average annual levels of particulates and infant mortality from pneumonia.
Pope et al. (1991)
  • Based on sample of fourth and fifth grade elementary students, found that elevated PM10 concentrations associated with significant decline in lung function as measured by peak expiratory flow (PEF). PM10 concentrations of 150 µg/m³ over three consecutive days associated with estimated average reduction of PEF of 15.5L/min (i.e. between 3%–6%).
  • Elevated PM10 concentrations also significantly associated with an increase in reported symptoms of upper and lower respiratory disease and the use of asthma medication.
Schwartz (1991)
  • In Detriot (USA), found that a 100 µg/m³ increase in average TSP levels was associated with a 6% increase in all-cause mortality.
Bobak and Leon (1992)
  • Found a statistically significant association between neonatal mortality, postneonatal mortality and postneonatal respiratory mortality and the average daily levels of TSPs.
  • Difficulties encountered with socio-economic confounding.
Dockery et al. (1992)
  • In St. Louis (USA), found that a 100 µg/m³ increase in average PM10 levels was associated with a 16% increase in all-cause mortality.
  • In Kingston (USA), found that a 100 µg/m³ increase in average PM10 levels was associated with a 17% increase in all-cause mortality.
Pope et al. (1992)
  • In the Utah Valley (USA), found that a 100 µg/m³ increase in the 5-day moving average PM10 levels was associated with a 16% increase in all-cause mortality.
  • Mortality averaged approximately 6-8% higher for each 50 µg/m³ incremental increase of 5-day-lagged moving average PM10 levels. On the 6.8% of days when 5-day-lagged moving average PM10 levels exceeded 100 µg/m³, mortality counts averaged approximately 19% higher than on days when these levels were less than 50 µg/m³.
  • An increase in PM10 levels equal to 100 µg/m³ was associated with an increase in respiratory and cardiovascular mortality of 43% and 20% respectively.
Schwartz and Dockery (1992a)
  • In Philadelphia (USA) between 1973 and 1980, found that 100 µg/m³ increase in daily average TSP levels was associated with a 19% increase in chronic obstructive pulmonary disease (COPD) mortality, 11% increase in pneumonia mortality, 10% increase in cardiovascular mortality and a 7% increase in all-cause mortality.
  • The relationship between TSP and health effects was stronger amongst persons >65 years old (100 µg/m³ increase in TSP associated with 10% increase in all-cause mortality).
  • Consistent dose-dependent relationship across broad range of TSP concentrations.
  • No significant association between sulphur dioxide concentrations and mortality after adjustment for TSP.
Schwartz and Dockery (1992b)
  • In Stuebenville, Ohio (USA), found that a 100 µg/m³ increase in previous day TSP levels was associated with a 4% increase in all-cause mortality.
Abbey et al. (1993)
  • A statistically significant relationship was found between TSP and ozone and TSPs and several respiratory diseases (including airway obstructive disease, chronic bronchitis and asthma) amongst non-smoking Seventh Day Adventists.
  • Increased TSP concentrations linked to both new cases of respiratory diseases and increased severity of pre-existing respiratory ailments.
  • No evidence of the interaction between TSP and ozone having an effect upon the relationship between particulate levels and adverse health effects.
Dockery et al. (1993)
  • Particulate pollution (PM10s and TSPs) was associated significantly with mortality from lung cancer and cardiopulmonary disease.
  • Concluded that particulate pollution contributes to excess mortality in the six US cities that were the subject of the review.
Schwartz (1993)
  • In Birmingham, Alabama (USA), found that 100 µg/m³ increase in 3-day average PM10 levels was associated with all-cause mortality (relative risk = 1.11), COPD deaths (relative risk = 1.16) and cardiovascular disease deaths (relative risk = 1.17).
  • Temperature had little impact on the estimated effects of PM10 concentrations.
  • No evidence of a threshold was found.
Schwartz et al. (1993)
  • Found a highly significant association between previous day and 4-day average PM10 concentrations and hospital emergency room visits for asthma amongst persons <65 years old in Seattle, Washington (USA). A 10 µg/m³ increase in previous day PM10 concentration was associated with relative risk of asthma emergency visits of 1.037. 30 µg/m³ increase in 4-day average PM10 concentrations associated with relative risk of 1.12.
  • No evidence of association between asthma emergency visits and sulphur dioxide or ozone.
  • No evidence of a threshold was found.
Dockery and Pope (1994)
  • US studies have found that all-cause mortality increases between 0.7% and 1.6% for each 10 µg/m³ increase in the daily average PM10 levels.
  • A strong association has been found between PM10 levels and cardiovascular and respiratory mortality.
  • Recognised that there is no clear understanding of the biological processes that are the cause of the statistical association between PM10 and adverse health affects.
Schwartz (1994)
  • Significant associations were found between PM10 levels and hospital admissions for pneumonia and COPD amongst persons >65 years old in Minneapolis-St. Paul, Minnesota (USA).
  • A 100 µg/m³ increase in PM10 levels was associated with a relative risk of hospitalisation from pneumonia and chronic obstructive pulmonary disease of 1.17 and 1.57 respectively.
Saldiva et al. (1995)
  • In Sao Paulo (Brazil), found that a 100 µg/m³ increase in daily average PM10 levels was associated with a 13% increase in all-cause mortality amongst people >65 years old.
  • The dose-response relationship between mortality and respirable particulate pollution was almost linear, with no evidence of a 'safe' threshold level.
  • The relationship between PM10 and mortality was not very sensitive to weather and seasonal variables.
  • A strong correlation was found between the PM10 levels and the levels of nitrogen dioxide and sulphur dioxide. However, found that mortality was still attributable to PM10.
  • Correlation between PM10 concentrations and all-cause mortality was strongest in summer.
  • PM10 levels in Sao Paulo mostly attributed to automotive exhaust emissions.
Xu et al. (1995)
  • In Beijing (China), found a statistically significant dose-dependant relationship between gestation age and TSPs. For each increase of 100 µg/m³ in the daily average TSP levels, gestation age fell by 0.042 of a week (i.e. 7.1 hours).
  • Concluded that high levels of TSP contributed to excess risk of premature delivery.
Pearce and Crowards (1996)
  • This study did not derive new exposure/mortality results, rather it applied average figures to UK urban PM10 data.
  • Using an estimated dose-response function of 0.096% increase in all-cause mortality for every 1 µg/m³ increase in PM10 concentrations, estimated that between 6668 and 10314 deaths in urban areas in England and Wales in 1993 were attributable to total PM10 exposure. Assuming 70% of PM10 concentrations have an anthropogenic source, the authors estimate that between 4667 and 7220 deaths in England and Wales were attributable to PM10 pollution from anthropogenic sources.
  • Estimated that PM10 pollution has premature mortality cost of approximately 11.1 billion (urban and rural exposure) and £6.3 billion for urban morbidity.
Schwartz et al. (1996)
  • Studied six eastern US cities over 8-year period. With the exception of one city (Topeka, Kansas), found that daily mortality increased by between 0.7% and 1.2% with each 10µg/m³increase in two-day mean PM10 concentration. The combined average for all six cities was a 0.8% increase in daily mortality for each 10µg/m³ increase in PM10 concentrations.
  • Found a strong association between fine particulates (PM2.5) and several health indices. In combined analysis of all 6 cities, found that 10 µg/m³ increase in two-day mean PM2.5 concentration was associated with a 1.5% increase in all cause mortality, 1.7% increase in mortality amongst people >65 years old, 2.1% increase in deaths from ischaemic heart disease, 3.3% increase in COPD deaths and a 4.0% increase in pneumonia deaths.
  • Inconsistent and unstable associations were found between coarse mass particulates (PM10–2.5) and mortality.
  • Sulphate particulates were significantly associated with elevated risk of mortality (5th to 95th percentile increase in sulphate particulate concentration associated with 3.8% increase in mortality). No significant associations were found with aerosol acidity (hydrogen ion) concentrations.
Hoek et al. (1997)
  • In Rotterdam (Netherlands), found that daily mortality was consistently associated with previous-day concentrations of TSP (relative risk = 1.05 for a change of 91µg/m³ ).
  • The relationship between mortality and TSP was linear below 100µg/m³ and levelled off at higher concentrations. Evidence suggested that if a threshold level exists for mortality effects of particles, it is at very low levels.
  • Relationship between TSP and mortality was not effected by levels of carbon monoxide or sulphur dioxide.
Woodruff et al. (1997)
  • Found a statistically significant relationship between average concentrations of TSP and postneonatal mortality from respiratory disease and sudden infant death syndrome in the USA.
Burnett et al. (1998)
  • In Toronto (Canada), found that 1% of non-accidental deaths could be attributed to particulate concentrations.
  • Observed a 1.9% and 0.7% increase in non-accidental mortality for each 10µg/m³ increase in PM2.5 and PM10 concentrations.
Ponka et al. (1998)
  • In Helsinki, found a strong association between PM10s and cardiovascular mortality and PM10s and all-cause mortality amongst persons <65 yrs old.
  • For persons <65 years old, each increase of 10 <µg/m³ in the average daily PM10 levels was associated with an increase of all-cause mortality and cardiovascular mortality of 3.5% and 4.1% respectively.
  • Strong correlation found between higher average daily concentrations of PM10s and respiratory disease mortality.
  • Did not find a statistically significant relationship between average daily PM10 levels and all-cause mortality or cardiovascular mortality for persons >65 years old.
  • PM10 found to have stronger correlation to adverse health effects than TSP.
  • No evidence of a threshold at which PM10 were not associated with mortality.
  • There was evidence that the interaction of PM10, ozone and nitrogen dioxide at high concentrations can have adverse health effects.
  • Suggested that findings support hypothesis that PM10s from soil have same effects on health as PM10 from combustion.
Atkinson et al. (1999)
  • In London, found there was a 2.98% increase in emergency respiratory admissions for a 10th to 90th percentile increase in PM10 concentration. This association was found across all age groups and was particularly strong in children (i.e. 4.85% increase).
  • PM10 concentrations were also associated significantly with increases in hospital admissions amongst persons >65 years old for COPD and asthma.
Boezen et al.(1999)
  • Found that in children with bronchial hyper-responsiveness and relatively high serum concentrations of IgE, the prevalence of lower respiratory symptoms increased significantly by between 32% and 139% for each 100µg/m³ increase in particulate matter (Netherlands).
Keles et al. (1999)
  • In Istanbul, found evidence that particulate pollution was associated with incidence of rhinitis.
  • Until 1994, coal widely used for industrial purposes and domestic heating. Natural gas network installed in certain areas in 1994. Incidence of rhinitis amongst sample high school population fell from 62.5% in 1994 to 51% in 1996. Fall attributed to reduction in TSP.
Rossi et al. (1999)
  • In Milan (Italy), found a 3% increase in all-cause mortality for every 100µg/m³ increase in daily average levels of TSP.
  • Results consistent with other Europe based studies that have found that the TSP-mortality link in Europe is less than that found in USA. Generally, in USA, studies have found that for every 100µg/m³ increase in average daily levels of TSP there is a 4%-6% increase in all-cause mortality. Suggested that this difference is attributable to different particulate aerosols in Europe and USA.
  • When restricted to days when TSP levels below 200µg/m³, every 100µg/m³ increase in average daily TSP levels associated with an 18% increase in COPD mortality, 20% increase in respiratory infection related mortality, 21% increase in myocardial infarction mortality, 10% increase in heart failure mortality and a 7% increase in all-cause mortality.
  • No evidence of a threshold level at which TSP have no effect upon mortality. In fact, as above data suggests, evidence that the dose-response relationship was higher at lower TSP concentrations.
Lippmann et al. (2000)
  • Studied relationship between different types of particulates and daily deaths and hospital admissions in Detroit, Michigan (USA). They found that PM10, PM10–2.5 and PM2.5 all had similar relative risk factors.
  • Found that particulate concentrations had effect on human health notwithstanding presence of other pollutants (sulphur dioxide, nitrogen dioxide, ozone and carbon monoxide).
  • Unable to determine the relative effects of different components of particulate matter.
Xu et al. (2000)
  • In Shenyang (China), the risk of all-cause mortality increased by an estimated 1.7% with a 100 µg/m³ concomitant increase in average TSP levels.
  • A 100 µg/m³ increase in average TSP levels was associated with an increase in the reported incidence of cardiovascular disease and chronic obstructive pulmonary disease of 2.1% and 2.6% respectively.
  • No statistically significant link was established between cancer mortality and TSP levels.

3.4.2 Risk assessment of wood-smoke and health

There have been many studies that deduce the impact wood-smoke may have on people in industrial countries. In one of the earlier studies Travis et al. (1985) analyse both safety risks and air pollution risks. They point out that safety, particularly house fires associated with poorly installed appliances, was the cause of almost 300 deaths in the United States in 1982. This risk was well documented. They note, however, that it was difficult to establish a causal relationship between wood-smoke and health. Using a model based on emissions of benzo(a)pyrene, they estimated the air pollution health risk of delivering heat from woodheaters was 100 times greater than delivering the same amount of heat from coal fired power stations. They urged the introduction of woodheater certification programs for cleaner burning appliances. Another early US study (Tombleson et al. 1983) also demonstrated the need for control measures to reduce the rapid growth in particulate concentrations in communities with significant wood-burning appliances.

Calle and Zeighami (1984) develop the health risk assessment approach to wood-smoke for indoor air using benzo(a)pyrene (B(a)P) as an indicator. They review several studies of B(a)P levels inside and outside households with woodheaters. Their conclusion, based on some relatively high indoor smoke levels (B(a)P levels of 5 to 100 ng/m³), suggest that a community with a high proportion of wood-heated homes would have an excess risk of lung cancer of 0.1 to 25%. The large range of excess risk results from use of two different risk assessment methods and uncertainty in the relative potency factors.

Concern about health impacts of wood-smoke with many published papers such as 'Potential Adverse Health Effects of Wood Smoke' (Pierson et al. 1989) appearing in the environmental and medical literature. In this paper the authors note the emissions of carbon monoxide, nitrogen oxides and sulfur oxides in addition to the particulate emissions. They stress the need for use of dry wood and correct operation of heaters in order to reduce health risks.

In 1993 the Air Risk Information Support Center (Air RISC) in the United States Environmental Protection Agency commissioned a review of the non-cancer respiratory effects of wood-smoke (Larson and Koenig 1993). Based on this analysis of the literature, the authors concluded that it was biologically plausible that wood-smoke could cause adverse respiratory effects and that seven out of eight published epidemiological studies on residential wood-smoke found adverse effects. The adverse effects reported included increased respiratory symptoms, increased lower respiratory infection and decreased pulmonary function. However, a dose response (biological gradient) had not been shown, possibly because of the limited data.

Kyrklund (1994) outlines on a Swedish study addressing the lifetime risk of cancer in a hypothetical village of 100 000 where firewood was the only heating fuel. Depending on the risk assessment method chosen, the study predicted between 55 and 333 extra cancer deaths per 100 000 people. Kyrklund concluded 'In view of the decrease in ambient soot and PAH levels in recent decades, the health risks of wood-fire emissions should not be exaggerated. However, a situation in which domestic use of wood-fired boilers makes a substantial contribution to total emissions of harmful pollutants must be avoided.'

Larson and Koenig (1994) have reviewed literature on wood-smoke and its health impact. They conclude their review by noting that much is understood about the constituents and fate of wood-smoke in the atmosphere but not enough is known about its health impact. They argue that the weight of evidence suggests a causal relationship between elevated wood-smoke levels and respiratory problems in children. They do not draw conclusions on other health aspects of wood-smoke.

3.4.3 Indoor wood-smoke risk – developed countries

A few studies have suggested a relationship between respiratory problems in children and the use of woodheaters in the home. Honicky et al. (1983, 1985) report a significant correlation between symptoms of respiratory illness among pre-school children and the use of woodheaters in Lansing, Michigan USA. This was a relatively small study of 62 children (31 in study group, 31 in control group) and indoor air quality measurements were not made. A study by Daigler et al. (1991) of 508 children in Springville, New York, USA found no significant correlation between children with asthma and wood heating, but the study did show significant correlation between otitis (inflammation of the ear) and the use of a woodheater. No correlation was found between asthma or otitis and households with an open fireplace.

A relatively small (45 subjects, 45 controls) US study of Navajo children (<2 years old) with acute lower respiratory illnesses (ALRI) indicated a five-fold increase in the risk of ALRI for children from homes where wood was used for cooking (Robin et al. 1996). A seven-fold increase in risk occurred where average indoor PM10 exceeded 65 µg/m³. There was insignificant increase in risk for children from homes where wood was used for heating. A weak correlation was observed between indoor particulate levels and the use of wood for heating or cooking.

A quite different conclusion was reached in a study of 1287 children in a rural area of Germany (von Mutius et al. 1996). This study found much lower rates of hay fever, allergy to pollen and bronchial hyper-responsiveness in children living in homes heated with wood or coal than homes with other forms of heating.

Another study that failed to identify a link between firewood use and respiratory disease in children was carried out in Massachusetts, USA by Tuthill (1984). In this case 399 households (65% with wood-burning heaters) were asked to complete questionnaires on acute respiratory illnesses in children (kindergarten to 6th grade). No significant correlation between respiratory illness and firewood use was observed, however a four-fold risk of illness was observed in households with recent renovations or new furniture (taken as a surrogate for increased formaldehyde levels indoors).

The above studies are inconclusive. They do not establish or reject a clear correlation between the presence or absence of wood use for heating or cooking and children's respiratory illnesses. A possible reason is that escape of smoke into a house from a woodheater or cooking stove will depend on the model of appliance, the installation and maintenance of the appliance and the mode of operation. Thus, presence or absence of a woodheater is too simple an indicator of possible exposure to indoor wood-smoke.

3.4.4 Health risk from indoor wood-smoke – developing countries

Women and children in societies where it is still common to cook over an open fire or unflued cooking-stove, often indoors, are exposed to extremely high levels of wood-smoke (Smith 1986, 1987). For example, in villages in Nepal, indoor TSP levels of 3000 to 42000 µg/m³ have been measured and attributed to indoor cooking and heating with biomass fuels (Davidson et al. 1986). In many other regions unflued wood-fires are used for cooking and heating homes (e.g. New Guinea Highlands) leading to prolonged exposure to high concentrations of wood-smoke. Smith et al. (1983) observed indoor concentrations of TSP in the range 1100 to 56600 µg/m³ in four Indian villages (average 6900 µg/m³) and benzo(a)pyrene concentrations of 62 to 19284 ng/m³ (average 3900 ng/m³). Such communities provide clear evidence of the adverse health effects from very high wood-smoke exposure. Kirk Smith, who has been a leading researcher in the field of indoor air quality for the rural poor in developing countries, argues that despite greenhouse gas implications, a switch to fossil fuels might be justified to overcome the health consequences of high wood-smoke exposure (Smith 1994b). The World Resources Institute (WRI 1999) suggests that biomass use in developing countries is, in fact, decreasing, although it still poses a major health threat.

A wealth of scientific and medical evidence exists on this subject. A review by McCracken and Smith (1997) collected over 180 articles on acute respiratory infections and indoor air pollution, mostly examples of exposure to wood-smoke in developing countries. Other significant reviews on this topic, among many, include the World Health Organization reports 'Epidemiological, Social and Technical Aspects of Indoor Air Pollution from Biomass Fuel' (WHO 1992a) and 'Indoor Air Pollution from Biomass Fuel' (WHO (1992b), the US Agency for International Development study 'Air Pollution and Child Health: Priorities for Action' (Bendahmane 1997), and an earlier work of Kirk Smith's 'Cooks on the World Stage: the Forgotten Actresses/Actors' (Smith and Colfer 1983).

A selection from the hundreds of articles on the problem of exposure to wood-smoke in developing countries is presented below.

Perez-Padilla et al. (1996) conducted a study of women over 40 years of age in Mexico. Women with chronic bronchitis and chronic airway obstruction (CAO) were the study group (127 women) and 95 healthy women were used as a control group. The study estimated accumulated wood-smoke exposure in terms of average hours exposure per day multiplied by number of years of exposure (hour-year). The authors concluded that risk of chronic bronchitis and CAO increased linearly as a function of hour-years exposure. For 200+ hour-years exposure (e.g. 5 hours per day for 40 years) the odds ratio, compared to the non-exposed control group, for contracting chronic bronchitis was 15, and the odds ratio for chronic bronchitis plus CAO was 75.

Another study in Mexico (Sandoval et al. 1993) of long-term exposure (subjects over 60 years old) to wood-smoke concluded that the impact on subjects' lungs was similar in many aspects to interstitial lung disease caused by exposure to inorganic dust. Brauer et al. (1996) measured PM10 and PM2.5 concentrations (averaged over 9 hours) in seven traditional Mexican kitchens using biomass fuels and obtained mean values of 768 and 555 µg/m³ respectively.

Dennis et al. (1996) studied elderly women of low socio-economic status in Bogota, Colombia and found wood-smoke exposure (from cooking) was associated with obstructive airways disease (OAD). They also found association between OAD and smoking, passive smoking and use of gasoline for cooking.

Studies of children and adults in New Guinea by Anderson (1978, 1979a, 1979b) demonstrated mixed results in establishing correlation between wood-smoke and respiratory disease. A study of 1650 children (0 to 14 years old) found no difference in obstructive lung disease between two groups, one from the Highlands where nocturnal exposure to wood-smoke is very high (600 to 2000 µg/m³ particulates) and a coastal group where wood-smoke exposure is low (Anderson 1978). Another study (Anderson 1979b) of 1284 adults in the Highlands showed increased prevalence of chronic cough, shortness of breath on exertion and bronchial hypersecretion from middle life onwards. This was not statistically associated with chronic respiratory symptoms or reduction of ventilatory capacity but is was associated with recent cough and bronchial hypersecretion. The relevance of wood-smoke exposure was unclear from this study.

The health problems of high wood-smoke exposure are reflected in regular exposures to high concentrations of carcinogens, such as benzo(a)pyrene, in some population groups in developing countries. For example, Aggarwal et al. (1982) measured concentrations of benzo(a)pyrene in the breathing zone of women cooks in India using various local fuels as: 1270 ng/m³ (burning wood), 8248 ng/m³ (burning dung), and 4207 ng/m³ (burning coal).

3.5 Summary – International

The international literature provides information that makes the Australian research and data collection more useful. The US development of several emission measurement techniques for wood-smoke particulates made development of Australian standard methods more practical. Field measurement of particulate emission factors for woodheaters and open fireplaces in the United States provides reassurance that estimated real-world emission factors for Australia are of the right order of magnitude. Measurement of emission factors for many air toxic species in the United States and Europe provide useful background for the present Australian test program.

Large-scale epidemiological studies in several countries have demonstrated significant correlations between mortality and morbidity and PM10 and/or PM2.5. This has had a significant impact on air quality goals for particulates in Australia and stimulated policies aimed at reducing wood-smoke in all Australian states and territories.

International studies have demonstrated the mutagenicity of wood-smoke. Studies in developing countries have demonstrated the health impact of long-term exposure to high concentrations of wood-smoke.

2 TEQ toxicity-equivalent
3 A comprehensive review of research published on the association between adverse health effects and particulate concentrations can be found in US EPA (1996b), Vol II, Table 6-24, available at 

Before you download

Most publications are available as PDF files. Adobe Acrobat Reader  is required to view PDF files.

If you are unable to access a publication, please contact us to organise a suitable alternative format.


   Links to another web site
   Opens a pop-up window