Aarhus Universitets segl

No. 777: Air pollution from residential wood combustion in a Danish village

Wåhlin, P., Olesen, H.R., Bossi, R. & Stubkjær, J., 2010: Air pollution from residential wood combustion in a Danish village. Measuring campaign and analysis of results. National Environmental Research Institute, Aarhus University. 49 pp. – NERI Technical Report No. 777.


WOODUSE is a comprehensive research project, which covers a wide range of issues related to residential wood combustion. It includes investigations of emissions, air pollution level in outdoor and indoor air, health effects and social aspects. The project has had participation by four Danish institutions (National Environmental Research Institute at Aarhus University, National Research Centre for the Working Environment, Department of Public Health, University of Copenhagen, Department of Civil Engineering, Technical University of Denmark). The project has received funding from the Strategic Research Council (grant ENMI-2104-05-0010, period 2006-2009).

In the winter 2006/7 a three month measuring campaign was conducted as part of the WOODUSE project. The campaign was targeted at studying the effect of residential wood combustion to air quality, and it took place in the village of Slagslunde in the countryside North of Copenhagen. The present report describes the campaign and its results.

The report is intended for a technically minded audience. It is supplemented by a more easily accessible report in Danish (Olesen et al., 2010b), which presents central results and extends them with the aid of model calculations.

The purpose of the campaign was to assess the effect of local residential wood combustion on air quality. Therefore, it was important to identify other potential sources and to quantify or eliminate their contribution. Slagslunde was selected as a suitable site because it is confined in extent, and there are only few sources other than wood combustion. The most important other sources are local traffic and a small combined power and district heating station using natural gas. This station has a 20 meter high stack and is positioned at the grounds of the public school in the western part of the village. The traffic near the sampling sites is modest. The effect of these other sources was expected to be small.

All air pollution monitoring instruments were placed in two mobile monitoring stations, one exposed to smoke from woodstoves at a site in the centre of the southern part of the village, and the other for background reference 500 meters outside the village to the WNW. This setup makes it possible to eliminate the influence from long range transport and from regional sources further away from the sites, by calculating the difference between concentrations measured in Slagslunde and outside of the village.

Main results

A central result of the study is a so-called 'wood smoke source profile', which relates several measures of wood smoke pollution to each other. Wood smoke pollution results in an increased concentration of fine particle mass (PM2.5), but also by an increased level of soot, by increased particle volume, and by an increase in various other compounds. The source profile is a useful tool for interpretation of future measurements.

The contribution of wood smoke to the PM2.5 level at the central measuring site in Slagslunde amounts to 2.0 mg/m3 as an average for a period of approximately nine weeks during a mild winter.

The report demonstrates how measurements of the particle size distribution are useful for identifying the contribution from various sources, while also describing the influence of wood smoke on the particle size distribution.

During ten days measurements of PAH (Polycyclic Aromatic Hydrocarbons) were carried out. The measurements showed a substantial increase in the level of PAH in the wood stove district compared with the background site.  The number of samples is too small to draw conclusions about the level of PAH on a yearly basis.

In various other studies levoglucosan and mannosan are used as indicators for wood smoke. However, the work on the wood smoke source profile presented here demonstrates that measurements of these two substances are not adequate for quantification of the contribution of residential wood combustion to PM2.5.

The measurement campaign
Slagslunde is a small village with approximately 400 detached houses, almost all single-storied, surrounded by agricultural areas, and with no local industry. Approximately half of the houses are equipped with wood stoves.

Measurements were conducted at a site centrally in Slagslunde and at a background site approximately 500 m outside the village. Simultaneous measurements with all the equipment running at both sites took place from 23 December 2006 until 26 February 2007. The consumption of wood for combustion in each house was mapped in detail by use of a questionnaire.

The village was situated at a gentle hill. Meteorological measurements were carried out at a mast placed 500 m NW of the village on top of the hill.

During the experimental campaign continuous measurements of a number of air pollution components were conducted: PM2.5, particle number and size distribution, CO, NO and NOx, soot particulates.

In addition to the continuous measurements, during 10 non-consecutive days, 24-hour high volume samples of particles and semi-volatile compounds were collected in order to measure PAH, levoglucosan and mannosan. PAH is carcinogenic, while levoglucosan and mannosan are of interest because they are characteristic for biomass combustion.

In a two-week campaign the relation between indoor and outdoor pollution was examined in two homes: one with, and the other without a woodstove.

Data analysis
An important step in the data analysis was to identify other potential sources than wood combustion.

The purpose of the NOx (and NO) measurements was to make corrections for the traffic, which is the main source of NOx, but may also be a disturbing local source of particle number (N) and PM2.5. From other studies the N/ NOx ratio and the PM2.5/ NOx ratio for traffic emissions are approximately known. This characteristic of traffic emissions can be used to assess the traffic contribution to the observed increments. Quality control of the NOx data after the campaign unfortunately showed that the monitors at both sites had not worked properly except for a period of 23 days. Therefore, it was not possible to make the corrections. Using the NOx increments measured in the 23 day period it was estimated that a small, but not insignificant part (10-20 %), of the particle number increments can be due to local traffic. For PM2.5 it was estimated that the local traffic contributions to the observed increments are almost insignificant (< 5%).

For further investigation of possible disturbing local particle sources the DMPS size spectrum data turned out to be useful. A peak in the increment spectrum around a particle size of 13 nm was sometimes present, but only in periods with certain wind directions. The peak had a positive sign at wind directions around 280°, and a negative sign at wind directions around 140°. These are exactly the directions to the district heating station from the exposed site and the background site, respectively. With other wind directions and when the heating station was turned off in the late night there was no 13 nm peak.  Also some episodes with high concentrations of traffic generated particles could be identified by the presence of a peak in the size spectrum around 26 nm.

Based on the analyses of the DMPS size spectrum data it was possible to construct a 'clean data set', for which the increments represent almost only wood combustion; this data set was used for the further analyses. Data measured during the strong traffic episodes and with wind directions coming from the central heating station were excluded from the cleaned data set. The cleaned data set includes 62% of the half-hours compared to the full set. A significant difference between the entire data set and the cleaned data set of the increment is found only for the particle number concentration. The reason is that the traffic source and the district heating station both produce many particles, but very little mass, particle volume and soot compared with the wood stoves.

The diurnal variation (hourly averages using the entire cleaned data set) shows elevated concentrations in the late afternoon and evening for the increments of particle volume, soot, PM2.5 and particle number. This fits well with the data from the questionnaire study, where the residential wood combustion in Slagslunde was mapped in detail.

The fact that the diurnal variations of PM2.5, V and Soot are similar indicates that only one major source (the wood stoves) is the cause of the increments at the site in the village. The diurnal variation of the particle number is slightly different with relatively increased levels in the middle of the day, which may be due to traffic emissions.

Source profile for wood smoke
Using the receptor model COPREM a constant 'wood smoke profile' was determined, which gives good predictions for PM2.5, V and Soot. The prediction of the particle number N was less satisfying, thus supporting the hypothesis that other local sources than wood smoke contribute to the particle number.

The increments of the PAHs were calculated on the basis of seven 24h samples which were collected simultaneously at the two sites. For these samples all measured values at the background site and at the exposed site were above the detection limit, and data for PM2.5, V and Soot were available. A regression analysis was made for the PAH increments as function of the source strengths found by the COPREM model. The correlations between the PAH increments and the increments of PM2.5, V and Soot were poor, which is reflected in the high uncertainties on the PAH values in the resulting source profile.

In other studies of pollution from wood combustion the monosaccharide anhydrides levoglucosan and mannosan have been used as indicators for wood smoke, because they are formed precisely by biomass combustion. Increments of levoglucosan and mannosan for nine 24h samples were studied. A woodstove source profile including levoglucosan, mannosan, PM2.5, V and Soot was determined. However, levoglucosan and mannosan showed poor correlation with PM2.5, V and soot, which is reflected by high uncertainties for levoglucosan and mannosan in the source profile.

A central and useful result of the study is a 'wood smoke source profile' - i.e. a vector which links the three measures PM2.5, V and soot to each other, so that from knowledge of the increment of one of these, one can estimate the value of the other two - provided that the source responsible for the increment is wood smoke. However, the increments of the parameters particle number N, PAH and monosaccharide anhydrides do not have a close link to these three measures – they can only be calculated with large uncertainty. Thus, the study indicates that although levoglucosan and mannosan are specific markers for biomass combustion, they are not adequate for quantification of the contribution of PM2.5.

The resulting source profile for wood smoke, which indicates average increments due to the wood smoke source in the period corresponding to the cleaned data set were:

PM2.5          1,91 ± 0.05     µgm-3

V                1,78 ± 0.07     µm3cm-3

Soot           2,01 ± 0.08     Mm-1


The level of PAH at the exposed site was clearly elevated compared to the background site. For benz(a)pyrene (BaP) the concentration was increased by a factor of 2.3. The average BaP concentration at the exposed site was 1.68 ng/m3. The available data material is much too limited to allow any conclusions about the annual average, but for comparison it can be noted that the EU target value for BaP is 1 ng/m3 as an annual average.

Indoor air pollution
Results from the minor campaign where indoor concentrations were measured in two homes are not treated in detail in the present report. However, the following observations can be noted:

  • In a house without wood stove the level of soot was substantially smaller than the outdoor level.

  • In a house with woodstove the level of soot was approximately at the same level as outdoor when averaged over a week. However, there is a strong time variation. Whenever a cold stove is lit a very pronounced peak far above outdoor levels occurs. Subsequently, with an active stove, the soot level gradually decreases over 5-7 hours and approximately reaches the outdoor level. When the stove becomes inactive, the level falls below the outdoor level.

Full report in pdf (2,21 MB)