Technical Report No. 5
J. Gras, C.Meyer, I. Weeks, R. Gillett, I. Galbally, J. Todd, F. Carnovale, R. Joynt, A. Hinwood, H. Berko and S. Brown.
Environment Australia, March 2002
ISBN 0 6425 4867 6
4. Wood-heater emissions – and controlling factors
The study objective relevant to this section is 'to gain an understanding of the emissions from a range of types of domestic solid-fuel-burning appliances using a variety of fuels under conditions of maximum and minimum (as permitted by the appliance)'.
Effective sampling of smoke from wood fires is a challenging technical task. Along with the copious particulate emissions there are substantial emissions of water from both free water in the fuel and combustion water. This water, together with the tarry condensates, can readily block filters and any unprotected instruments can rapidly become contaminated. One generally successful strategy for sampling these emissions is the use of a dilution tunnel, as for example in the AS4013 heater test methodology (AS/NZS4013, 1999). In this procedure all the emissions are captured by entrainment into a much larger volume of clean, flowing air. Typically this is larger by a factor of around 10 to 20, diluting the emissions substantially and reducing the relative humidity. In practice, because of the very concentrated particulate emissions from wood heaters, room air is used for dilution and this procedure was followed for this study. In addition to the wanted smoke samples, a series of four room, or background, samples were also taken during operation of the heaters to determine typical concentrations of all analysed species in the dilution air.
Standard procedures, including AS/NZS 4013, attempt to minimise variance in emissions by controlling several burn variables within relatively narrow limits. The present study specifically involves selected variations from the standard procedures, but otherwise, wherever possible the standard procedure has been followed. To summarise, the AS/NZS 4013 procedure includes three test burns for each of three flow settings (high, medium and low) and includes a conditioning burn for each change of conditions. Fuel moisture and density are prescribed within narrow limits, as is the volume that fuel takes up in the combustion chamber. For each burn a single uniform charge of fuel is placed onto a prepared coal bed, with emission factors determined as an integral over combustion of this fuel charge. For burns on low flow settings, 20% of the fuel mass is consumed in establishing sustainable combustion before filter collections are initiated. The major variations for the present study include use of selected high or low flow settings only, specific variations in fuel type, moisture content and density and a varied number of repetitions. Test burn conditions and their effects are discussed more fully in later sections.
For the test burns, the NATA-certified AS4013 test rig operated by HRL Technology Pty. Ltd. in Morwell Victoria, was used in essentially standard operating conditions, with the addition of extra manifolds, and where required additional dilution, to accommodate the sampling train and the large array of samplers and instruments. A schematic of the dilution tunnel system as specified in AS4013 is given in Fig. 1. Photographs of the test rig and samplers are given in Appendix 4.
As an adjunct to the HRL facilities at Morwell, a CSIRO Atmospheric Research portable laboratory caravan was set up adjacent to the HRL heater test facility to provide a clean temperature-controlled working area for loading and unloading samples before and after exposure.
Source: Taken from AS/NZ4013(1999)
The sampling train for this study comprised a new section of stainless steel flue pipe installed in the AS4013 dilution tunnel into which were inserted sub-sampling stainless-steel probes that connected to two primary, stainless-steel manifolds. These are given names here of aerosol inlet and the gas inlet. The aerosol inlet was operated at near isokinetic conditions facing upstream and the gas inlet was operated anisokinetically (facing downstream) to minimise aerosol interference. The scheme is depicted in Fig. 2 along with the other components in the aerosol sample line.
The aerosol line comprised the inlet sub-sampling jet and approximately one metre of 12.5 mm diameter stainless steel tubing passing to a primary aerosol manifold, also constructed from stainless steel, 13.5 cm diameter and 32 cm long. This provided 10 outlets distributing sample air via 6 mm diameter stainless steel tubing to the various aerosol and some gas sample lines. As well, some air was passed to a dilution device, comprising a switched volume diluter and actively controlled filtered air source with a secondary manifold–delay tank (stainless steel, 13.5 cm diameter, 92 cm long, with 10 outlets). This also provided a means to convert from the negative pressure in the ventilated dilution tunnel to a positive pressure more suitable for operation of the continuous aerosol sampling instruments. Typical dilution ratios for this device were around 2000:1. Aerosol filter samplers and traps were mounted on a support rack. Each line included an air drier, or desiccator, and active mass flow control and all flow rates were logged every ten seconds.
Note: Also used for some gas species e.g. PAH, PCDD/F, HF, ketones and aldehydes
The emissions of the gaseous pollutants carbon monoxide (CO), methane (CH4), nitrogen oxide (NOx) and sulfur dioxide (SO2) were measured in real time from undiluted gas sampled directly from the dilution tunnel. In addition, integrated bulk gas samples were collected for speciated volatile organic carbon (VOC) analysis. Air was withdrawn via the gas inlet, which pointed downstream into the centre of the dilution tunnel, and was then split into four streams as shown in Fig. 3. Each stream was filtered by a 47-mm diameter PTFE 1 µm membrane filter (Type FA, Millipore, USA). With the exception of the SO2 sample line, the gas also was pre-filtered by a pair of 2 µm sintered stainless steel filters (Swagelok, USA) connected in parallel. This was to remove the larger particles and to reduce the risk of the PTFE filter becoming blocked during sampling. Pre-filtering was not appropriate for SO2 analysis due to the high reactivity of SO2 with stainless steel and the resultant loss of sample. All sample lines downstream of the PTFE filters were 6.35 mm (1/4') PFA (Swagelok, Solon OH USA).
The speciated VOC sample was withdrawn from the stainless steel manifold by a stainless steel/PTFE diaphragm pump (Model MPU487-NO5, KNF Neuberger, NJ, USA), and pumped via a rotameter-type flowmeter to an 80 L Tedlar bag. The sample flow rate was regulated with a fine metering needle valve and adjusted manually throughout each test burn to maintain constant flow. On completion of each test run, gas was decanted into dual-valve SUMMA canisters (Scientific Instrument Services, USA). One or two canisters were flushed by a minimum of four volumes of gas, using the PTFE diaphragm sample pump, after which the outlet valve was closed and the canisters filled to a pressure of approximately 250 kPa absolute.
A rigorous cleaning regime was followed for all filter holders, lines and sample tubes comprising multiple washing in Decon 90, Milli-Q water, toluene (pesticide grade), methylene chloride (pesticide grade) with oven or air drying. Quartz filters were all pre-baked at 400 °C. HPLC (high performance liquid chromatography) grade isopropyl alcohol was used for cleaning silicon O-rings in the filter holders.
A dedicated sample line from the primary aerosol manifold was used for collecting samples for determination of polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzo furans (PCDFs) and polychlorinated biphenyls (PCBs), see Fig. 2. Sampling for these species was based on US EPA Method 23 – 'Determination of Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzo furans (PCDFs) from municipal waste combustors' but with adaptation to suit the overall AS4013 dilution tunnel sampling regime.
Aerosol and gas-phase samples were collected from the same line at a nominal 5 L/min. Sample flow was actively controlled and logged with 10-s resolution. The PCDD/PCDF/PCB sample line comprised the main stainless steel aerosol isokinetic inlet, inlet line and primary manifold and secondary stainless steel distribution line. Aerosol phase samples were collected at the head of the sampling train using tandem ultra-pure quartz filters; the front filter was 47 mm diameter and the back-up 33 mm diameter (Gelman 2500 QAT-UP). Housings were metal-PTFE filter holders (anodised aluminium housing with stainless steel support screen and PTFE housing with stainless steel support screen; both employed dry ungreased viton O-rings). Gas-phase samples were collected after removal of particles from the gas stream, using a combination pre-cleaned polyurethane foam (PUF) plug (22 mm D x 38 mm PUF) – XAD-2 resin – PUF foam plug (22 x 38 mm) trap (Supelco, type ORBO-100) in Pyrex glass holders (a separate tube per sample). Although air temperatures in the dilution tunnel and primary aerosol manifold were generally low (T < 50 °C) and the laboratory temperature was controlled, generally in the range 20 °C–26 °C, the gas sampling collection traps (in particular the XAD-2 resin) were maintained at 20 °C during sampling using an actively-controlled thermo-electric (Peltier) cooling block. All sample and filter holders were prepared using the previously described cleaning protocol.
Prior to sampling, all filters were pre-baked at 400 °C for 12 hours and stored in pre-baked aluminium foil. PUF plugs were cleaned by Soxhlet extraction using hot toluene. XAD-2 resin was also cleaned in an extracting thimble by Soxhlet extraction. Probe housings were cleaned using a hot toluene wash and the probes baked at 200 °C to remove all remaining traces of organics. The probes were assembled then stored in toluene-washed aluminium foil. Sample tubes were sealed with PTFE caps and re-wrapped in toluene-washed foil. All samples were immediately refrigerated and were stored refrigerated until analysis.
Analysis of polychlorinated-para-dioxins (PCDDs) and polychlorinated-para-dibenzofurans (PCDFs) was carried out by the Australian Government Analytical Laboratories (AGAL), Sydney. The analysis procedure follows US EPA method 23. A short summary of the method is given here.
On return to AGAL the probe components were spiked with 13C labelled internal standards (filters, probe housing, PUFs and XAD-2 resin) and were extracted together to remove the collected PCDD/F. This used a hot soxhlet extraction in toluene. Clean-up involved partitioning with sulphuric acid then distilled water, followed by column chromatography on acid and base modified silica gels, neutral alumina and carbon dispersed on celite.
Analysis for PCDD/Fs was by high resolution Gas Chromatograph–Mass Spectrometer GC-MS (ThermoQuest Finnigan MAT95XL) that has a sensitivity for 2,3,7,8-TCDD of better than 10 fg with a signal to noise ratio of 10:1. Individual congeners are identified using the GC retention time and ion abundance ratios, with reference to internal standards. Quantitative analysis is performed using selected ion current profiles following a variety of methodologies that are congener specific and requiring calibration as described for example in USEPA Method 1613. There are 210 possible PCDD/F congeners arising from different chlorine substitutions. Of these, 17 are considered to be of concern with the most toxic being 2,3,7,8-TCDD. For this study the 17 congeners considered to be of concern were analysed, these are:
All values of concentrations for PCDD/F are reported here using the WHO-98 Toxic Equivalent (TEQ) scale, based on a toxicity rating of 1 for 2,3,7,8-TCDD. Other congener concentrations are weighted by their toxicity relative to 2,3,7,8-TCDD. For PCDD/F data given in this report, values below the minimum detectable level are assigned a value of one half of the minimum detectable level.
Polychlorinated biphenyls (PCBs) were determined from the same samples collected for analysis of PCDD/F. The samples were extracted with toluene and analysed by High Resolution Gas Chromatography / Low Resolution Mass Spectrometry with calibration against individual PCB congener standards (using a mixture of 41 non-coeluting congeners). Twenty-seven congeners were determined, comprising those generally considered to be toxic. This includes the four coplanar congeners 77, 81, 126 and 169, which are considered most toxic; it also includes the commonly measured marker PCBs. The full set analysed includes PCB numbers 8, 18, 28, 44, 52, 77, 81, 101, 105, 114, 118, 123, 126, 128, 138, 153, 156, 157, 167, 169, 170,180, 187, 189, 195, 206 and 209.
Aerosol carbon can be present as either organic species, which are relatively volatile, or more refractory (non-volatile) black carbon. Samples were collected for determination of aerosol carbon using a dedicated sampling line from the primary aerosol manifold, operating at a nominal flow rate of 5 L/min, with active flow control and with the flow rate logged at 10-s intervals. All sample lines were stainless steel and the filter holders used for the collection were anodised aluminium with stainless steel support screens. Two filters were operated in tandem; the front filter was used for collection of the semi-volatile aerosol sample and the second filter as a gas blank. Filters used were 47-mm diameter ultra pure quartz (Gelman 2500 QAT-UP) and all filters were pre-cleaned before use by baking at 400 °C for a minimum of 12 hours. Pre-cleaned filters were stored in baked aluminium foil before use and in cleaned glass petri dishes after exposure. These were also wrapped in baked aluminium foil. All filters were refrigerated immediately after collection and were kept frozen until analysis. On return to the laboratory, the filters and segments of the exposed filters for analysis were weighed. Total carbon in the aerosol on the selected filter segments was determined using thermal decomposition to CO2 in CO2-free air in a tube furnace, with a copper catalyst to oxidise any CO produced by partial oxidation. A Licor 6262 analyser was used for determination of the evolved CO2. The carbon analyser was calibrated by injection of known quantities of CO2 into the flowing air stream. Samples for determination of black carbon were subjected to an additional stage of combustion at 350 °C in filtered air for 30 minutes, removing the semi-volatile carbon. These samples were then analysed for carbon content in the same way as the total carbon samples.
Aerosol and gas-phase samples were collected via a dedicated line from the aerosol primary manifold. The sampling train used tandem filters comprising a front filter 47 mm ultra-pure quartz and back-up 33 mm quartz (Gelman 2500QAT-UP) in a combined metal-PTFE filter holders (anodised aluminium housing with stainless steel support screen and PTFE housing with stainless steel support screen). A pre-cleaned polyurethane foam plug (PUF) from Supelco, type ORBO-100 22 mm x 76 mm, mounted in a pyrex glass holder directly behind the filters was used to collect the gaseous fraction of the PAH. The sampling flow rate, which was actively controlled and logged at 10-s intervals, was nominally 5 L/min giving a typical collected volume of 700 L. All filters were pre-cleaned by baking at 400 °C, for a minimum of 12 hours, before exposure. PUF plugs were kept in sealed dark glass sample bottles before exposure and were returned to these bottles after exposure. Filters were stored in pre-baked aluminium foil before exposure and in glass petri dishes wrapped in pre-baked aluminium foil after exposure. All samples were immediately refrigerated after exposure and stored at -10°C until analysis.
Both PUF plugs and aerosol filters were extracted in 15 mL of acetonitrile by ultrasonication for 15 minutes in glass vials. PAH concentrations in the extract were determined using HPLC gradient elution, with a mobile phase consisting of a water–acetonitrile mixture pumped by a Dionex GP-40 gradient pump. PAHs were separated by injecting 25 µL of extract on a 25 mm x 4 mm Supelcosil LC-PAH column with 5µm packing. After separation, PAH absorption peaks were detected with an LDC uv/visible spectrophotometer at a wavelength of 254 nm. The peaks were quantified against a Supelco EPA 610 PAH mixture containing the 16 US EPA priority PAHs (PAH-16). This standard was serially diluted to produce a range of six standards which covered the PAH concentration range found in the samples. The sixteen PAH set comprises naphthalene, acenaphthylene, acenaphthene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benzo(a)anthracene, chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(a)pyrene, dibenzo(a,h)anthracene, benzo(g,h,i)perylene, and indeno(1,2,3-cd)pyrene.
Analyses for aerosol mass and particle-bound species including heavy metals (arsenic, cadmium, chromium, mercury and lead), phthalates and fluoride were carried out on particles collected on stretched PTFE filters. Filters used were 2 µm pore Pall-Gelman 47-mm diameter Teflo. It should be noted that with this pore size in PTFE filters all particulate material is collected. The principal collection was from a dedicated line on the aerosol primary manifold, operated at a nominal 1 L/min using an active mass flow controller and with the flow rate logged at 10-s intervals. A secondary (duplicate) collection from sample 22 onwards, was taken using similar filters operating in the aldehyde-ketone line at a lower flow rate (at a nominal flow rate of 240 mL/min).
Aerosol: Mass Loading
All of the PTFE filters were pre-weighed before collection, after desiccation for a minimum of 24 hours at a maximum relative humidity of 20%. Combined aerosol and filter mass were determined to a precision of 0.1 µg using a Mettler UMT-2 ultramicrobalance, which is also maintained at low humidity. Filters were then stored in plastic petri dishes before use and were returned to the same containers after exposure. On return to CSIRO-AR the filters were again desiccated at a maximum relative humidity of 20% for 24 hours and reweighed. In both cases this procedure involves repeated weighing until a stable weight is achieved, with a minimum of three determinations. Radioactive neutralisers placed in the balance and balance chamber were used to eliminate static charging artefacts during weighing. The dry aerosol mass was determined from the difference between the pre- and post- collection weights. Typically this procedure results in an uncertainty in aerosol mass of a few micrograms.
Aerosol: heavy metals by Proton Induced X-ray Emission (PIXE) analyses
Concentrations of heavy metals, specifically arsenic, cadmium, chromium, mercury and lead were determined for samples collected on PTFE filters using proton induced x-ray emission, an ion-beam analysis method. The Nuclear Science Applications Group at Australian Nuclear Science and Technology Organization (ANSTO), Lucas Heights conducted these analyses. Concentrations of some other species frequently used to characterise smoke emissions (S,Cl, K) are also included in this report.
Aerosol: phthalate, fluoride compounds
After analysis by PIXE (a non-destructive method), filters were returned to CSIRO-AR for determination of soluble aerosol species. These analyses were primarily for phthalates and fluoride although concentrations of some other species were also determined.
For elution of the soluble aerosol species filter samples were placed in clean polyethylene bags and extracted in 12 mL of Milli-Q water and 120 µL of chloroform, after the hydrophobic filter was wetted with 150 µL of methanol. The fluoride and phthalate concentrations in the eluent were determined by suppressed ion chromatography using a Dionex DX500 ion chromatograph. A 50 µL sample extract was injected onto an AS11 Dionex anion column 250 mm x 4 mm after passing through an AG11 guard column. Peaks were resolved using a gradient run consisting of eluants containing 5 mM NaOH, 100 mM NaOH and Millipore-water and then detected using a Dionex CD20 conductivity detector after the background conductivity was reduced in a ASRS-Ultra anion suppressor. Fluoride and phthalate peaks were quantified by using six standards, serially diluted from an Ultra Scientific standard, which covered the concentration range found in the extracts.
Samples to be analysed for organic isocyanates were collected in a dedicated sampling line from the aerosol primary manifold at a nominal flow rate of 1 L/min, controlled by an active mass flow controller and with the flow rate logged at 10-s intervals. The sampling methodology used followed standard procedures for determination of organic isocyanates in air as specified in 'Methods for the Determination of Hazardous Substances, Organic Isocyanates in Air', MDHS 25/2, Health and Safety Executive U.K, August (1994).
The procedure followed used the designated alternate sampling method employing impregnated filters. The sampling head comprised two impregnated 25-mm diameter glass filters cascaded in a polycarbonate housing. Filters were impregnated with 1-(2-methoxyphenyl)piperazine (1-2MP) absorbing solution. This method was chosen over a toluene impinger–bubbler method to avoid cross contamination with other sampling lines (VOC) and for safety reasons. The method, as used, will collect both gas-phase and particulate isocyanates.
Each filter unit was pre-sealed and exposed only on connection to the sample train. Immediately after exposure the collecting filters were removed from their holders and placed into a vial containing 1-2MP in approximately 2 mL of toluene to derivatise any unreacted organic isocyanates to urea derivatives. These were subsequently determined using HPLC with ultraviolet detection at 242 nm and electrochemical detection. The analytical detection limit is 0.1 µg NCO (isocyanate group) per sample. The Laboratory Services Unit of Workcover NSW prepared the impregnated filters for sample collection, and this laboratory also carried out the analyses by HPLC using method WCA110 (Standard method, Laboratory Services Unit, WorkCover New South Wales, Thornleigh, NSW).
Hydrofluoric acid (HF) was collected from the smoke by trapping the gas on 47 mm alkaline Millipore absorbent pads, mounted in polyethylene filter holders, in the principal aerosol mass-filter line. Air was drawn through the filters at a nominal flow rate of 1 L/min. The pads were impregnated by soaking in a solution containing 0.2 g K2CO3 and 63 g glycerol in 500 mL of Milli-Q water. This resulted in the pads absorbing about 1.5 mL of the solution or about 6 µmoles of K2CO3, which is far in excess of the HF collected on the filter.
After sampling was complete, filters were removed from the filter holder and sealed in a plastic bag and stored at 4 °C until analysis. Fluoride extraction from the filter was performed in the plastic bags by the addition of 10 mL of Milli-Q water. The fluoride concentration in the extract was determined by suppressed anion chromatography using a Dionex DX500 ion chromatograph and a 10 µL sample injection. Peaks were separated on an AS11 separation column after passing through an AG11 guard column; an ASRS-ultra 4 mm suppressor was used for background suppression. Peak detection was achieved using a CD20 conductivity detector and quantification was attained from an Ultra Scientific NaF standard, which was serial diluted to give a range of six standards that bounded the concentrations found in the extracts.
A dedicated sampling line from the primary aerosol manifold was used to collect aldehydes and ketones. This sampling line was operated at a nominal flow rate of 240 mL/min, with active flow control and the flow rate was logged at 10-s intervals. Total air volumes ranged from 8 L to about 160 L. For samples 1–21 a quartz pre-filter was used in this line but the filter type was changed to a stretched PTFE filter after sample 21 to provide a backup for the aerosol mass determination.
Aldehydes and ketones were collected from the smoke and derivitised on Supelco LpDNPH H10 cartridges located downstream of the aerosol filter. After sampling was complete, cartridges were sealed with plugs, packed in alfoil bags and stored at 4 °C until analysis.
The 2,4-dinitrophenylhydrazine (2,4-DNPH) derivatives of the aldehyde and ketone gases were extracted from the cartridges with a known volume of acetonitrile (approximately 4.6 mL). Extracts were stored in glass vials before they were transferred to vials for injection. Separation of the extracts was by HPLC, using a 250 mm x 4.6 mm Supelcosil LC-18 HPLC column with 5 µm packing. Aldehydes and ketones were eluted from the column using a mobile phase consisting of a gradient of acetonitrile and Milli-Q water pumped by a Dionex GP40 gradient pump. Peaks were detected by an LDC ultraviolet/visible spectrophotometer at a wavelength of 360 nm and then quantified by serial dilution of a Supelco Carb method 1004 2,4-DNPH mix 2. This standard was diluted by 50 times and then serially to give a range of standards to encompass the sample concentrations.
Continuous measurements of aerosol mass during each burn were made using a TSI Dustrack monitor with a PM10 size selective inlet and an Rupprecht & Patashnick TEOM® Tapered Element Oscillating Mass continuous monitor, operating without a size selective inlet. Measurements of the aerosol light scattering coefficient were made using a Radiance M903 nephelometer and a series of aerosol size distributions were made using a differential mobility analyser system comprising a TSI 3071A electrostatic classifier and a TSI 3010 condensation nucleus counter. A second (reference) nephelometer, also a Radiance M903, was operated in the sampling laboratory as an indicator of general air quality during sampling (to guard against unintentional contamination of the laboratory dilution air due to external sources).
Calibration of the nephelometer was carried out for span using FM200 fire-retardant gas before and after the test series and zeros were checked daily using filtered air. The TEOM calibration was verified using the standard procedure of gravimetric determination of a series of clean filter cartridges. Flow rates were verified using a bubble flow meter. The DustTrak was zeroed regularly during the study, usually at least twice per day, using a HEPA filter cartridge on the inlet line. The nominal span for the DustTrak was kept constant, however all mass emission rates were normalised by equating the integrated output for a sample to the observed gravimetric mass for that sample (from the filter collection). All data were also corrected for small zero drifts during the determination using the before and after sample zero determinations.
SO2 concentration was determined using a pulsed fluorescence SO2 analyser (model 9850, Monitor Labs, Eaglewood, CO, USA) following AS3580.4.1-1990. The instrument was calibrated at the beginning of the experiment using a multi-point calibration from 0 to 5 ppm by diluting a certified SO2 standard (b-standard, BOC, Melbourne) with zero grade air. The instrument span was checked at least twice each test day using a 5 ppm SO2 standard, and instrument zero was established at the start and end of each test using ambient air. All calibrations were performed via the complete SO2 sample line including the Teflon filter. Sample losses via the inlet line were also assessed and were found to be less than 1%.
Methane and total nonmethane VOCs were analysed with a total hydrocarbon analyser (model 55, Thermo Environmental Instruments, USA), following AS3580.11.1-1993. Instrument span was checked twice daily against a standard containing 3 ppmC CH4 and 2.7 ppmC propane in nitrogen (National Bureau of standards, USA). Instrument zero was established using zero grade air. The gas pressure in the sample loop was monitored throughout each test and all concentrations were adjusted to STP (standard temperature and pressure).
The NOx, SO2, CH4 and total NMVOC instruments were logged continuously by computer, and integrated over the time-course of each test burn.
Carbon monoxide was measured by Fourier Transform Infra-red Spectroscopy (FTIR). Infrared spectra were measured at 1 cm-1 resolution using a Bomem MB104 FTIR spectrometer fitted with a globar source, MCT detector and 22 m path, 8 L volume White cell (IR Analysis, Anaheim Ca.) housed in a thermostated enclosure. Cell pressure and temperature were monitored and logged along with each spectrum measured. Air was drawn from the dilution tunnel at approximately 2 L/min, filtered, dehumidified in a cooling coil in an ice bath, and further dried through a semi-permeable membrane drier (Nafion, Dupont, USA) and flushed through the White cell. Spectra were collected continuously and accumulated in one minute (20 scans) groups. The operation was fully automated under the control of a single Array Basic program running in the GRAMS (Galactic Industries, New Hampshire) spectrometer operating environment. The spectrometer details and methods are similar those described in detail by Esler et al. (2000). The sample flow through the FTIR White cell is small relative to cell volume and therefore the time-series of FTIR spectra is, in effect, filtered by an exponential smoothing function with a half time of four minutes. Prior to each test burn, the sample cell was flushed twice with zero grade air, evacuated to 50 hPa and 20 spectra were recorded and averaged to yield the reference spectrum for the subsequent time-series. FTIR spectra were analysed on-line by Classical Least Squares (CLS) using calculated calibration spectra as described in detail by Griffith (1996). However the concentration range typically observed during these tests was large and no single calibration spectrum was suitable for the full range. Therefore the concentrations calculated by CLS instead were used as initial estimates for the non-linear MALT procedure (NLM), which yielded the actual CO concentrations.
Nitrogen oxides were analysed by a chemiluminescent NOx analyser (model 9841A, Monitor Labs, Eaglewood, CO, USA) following Australian standard AS3580.5.1-1993. The instrument was calibrated at the beginning of the experiment using a multipoint calibration from 0 to 10 ppm by diluting a certified NO standard (BOC) with NO-free zero grade air (BOC, Melbourne). The instrument span was checked at least twice each test day using a 10.1 ppm NO b-standard, BOC, Melbourne), and instrument zero was established at the start and end of each test using ambient air. The efficiency of the thermal NO2 converter was determined by converting the NO to NO2 using a CrO3 converter (Scintrex, Concorde, Ontario, Canada) and measuring the concentration of NO recovered via the thermal converter. Conversion efficiency was better than 99%. Nitrous oxide, N2O, was measured using FTIR with the system already described for carbon monoxide.
Canister samples were analysed for VOCs using a Hewlett Packard 6890A gas chromatograph (GC) equipped with a mass selective detector (MSD) and a capillary column (Alltech AT-1, 60 m x 0.25 mm ID, 1.0 µm film thickness), temperature programmed from -50 to +240 °C. The MSD was operated in ion scan mode. VOC concentrations were quantitated on extracted ions and peaks identified on the basis of both comparison of retention times with those in a 42 component working standard, (as well as some other standards that are held) and by mass spectrum. A measured volume from the canister sample, the working standard or the blank was cryogenically concentrated for injection onto the capillary column using an Entech Model 7000 Preconcentrator and an Entech Model 7016BCA Canister Autosampler (Entech Instruments Inc., Simi Valley, CA, USA). Automated control of the analysis system was achieved with Entech Smartlab software running in conjunction with Hewlett Packard ChemStation software on a personal computer. The amount of sample used was in the range 100 to 1000 mL, with smaller volumes being used for the higher concentration samples. A 5 ppbv per component working standard was used to calibrate the GC system. The working standard was prepared by static dilution of a nominal 100 ppbv 42 component C1-C10 standard or a 39 component 100 ppbv air toxics standard (Scott Speciality Gases, San Bernadino, CA, USA, Micrograv mix certified to ±10%). The dilution was made using a high precision pressure transducer (Type RPT301, Druck Limited, Groby, Leicester, UK). The operating range of this instrument was 35-3500 HPa absolute and the manufacturer has certified it to ±0.01 HPa. As a check on this dilution process, working standards prepared have been analysed against another low concentration primary standard (National Institute of Science and Technology, Gaithersburg, USA. SRM 1804a, 19 toxic volatile organic compounds in nitrogen, nominal concentration 5 ppbv and certified to ±2%). Based on a comparison of the benzene and toluene analyses the agreement between the concentration of the working standard determined by the dilution measured by pressures and alternatively determined by comparison with the NIST standard was around 2%. During the course of each analysis sequence samples, standards and blanks are measured. The blanks consisted of the boil-off gas from the high purity liquid nitrogen cryogenic fluid in which the VOC concentration was less than 0.1 ppbC. High concentrations in some samples caused problems with peak identification due to overloading of the analytical capillary column. To overcome this, these samples were accurately diluted with high purity nitrogen before analysis.
Species analysed in this study are summarised in Table 1. This includes the Commonwealth's Living Cities Air Toxics priority list (excluding nickel carbonyl) and additional pollutants, including those known as criteria pollutants.
|Species||Method||Phase medium||Aerosol||Gas medium|
|Arsenic and compounds||PIXE||aerosol||PTFE filter|
|Cadmium and compounds||PIXE||aerosol||PTFE filter|
|Chromium & compounds||PIXE||aerosol||PTFE filter|
|Condensed & semi-volatile carbon||thermal decomposition/NDIR||aerosol||Quartz filter|
|Fluoride HF||Ion chromatography||gas||impregnated filter|
|Fluoride compounds||Ion chromatography||aerosol||PTFE filter|
|Lead and compounds||PIXE||aerosol||PTFE filter|
|Mercury and compounds||PIXE||aerosol||PTFE filter|
|Methyl ethyl ketone||HPLC||gas||DNPH trap|
|Methyl Isobutyl Ketone.||HPLC||gas||DNPH trap|
|Methylenebis (phenylisocyanate)||HPLC||both||Quartz filter||impregnated filter|
|Nickel carbonyl||not analysed|
|Oxides of nitrogen||NOx||gas||continuous|
|Aerosol mass loading||gravimetric||aerosol||PTFE filter|
|Aerosol mass loading||TEOM||aerosol||continuous|
|Aerosol mass loading||particle spectrometry||aerosol||continuous|
|PCBs||GCMS||both||Quartz filter||PUF/XAD-2 resin trap|
|Phthalates||Ion chromatography||aerosol||PTFE filter|
|Polychlorinated dioxins and furans||GCMS||both||Quartz filter||PUF/XAD-2 resin trap|
|Sulfur dioxide||pulsed fluorescence||gas||continuous|
|Toluene-2,4-diisocyanate||HPLC||both||Quartz filter||impregnated filter|
|Total Volatile Organic Compounds||VOC||gas||canister|
|Vinyl chloride (monomer)||VOC||gas||canister|
The sampling program undertaken in the study is summarised in Table 2, which groups the various tests according to appliance type, fuel, fuel moisture category, appliance setting and the disposition of the fuel in the appliance. In all, 53 samples were collected from 45 individual burns.
|1||1||Fuel||C2 (3.7 g/kg)||redgum||seasoned||high||standard|
|2||2||C2 (3.7 g/kg)||redgum||seasoned||high||standard|
|37||30||C2 (3.7 g/kg)||redgum||seasoned||high||standard|
|38||31||C2 (3.7 g/kg)||redgum||seasoned||high||standard||repeated|
|19||17||C2 (3.7 g/kg)||jarrah||seasoned||high||standard|
|20||18||C2 (3.7 g/kg)||jarrah||seasoned||high||standard|
|21||19||C2 (3.7 g/kg)||jarrah||seasoned||high||standard|
|22||20||C2 (3.7 g/kg)||bluegum||seasoned||high||standard|
|23||21||C2 (3.7 g/kg)||bluegum||seasoned||high||standard|
|24||22||C2 (3.7 g/kg)||bluegum||seasoned||high||standard|
|44||37||C2 (3.7 g/kg)||pinus r.||seasoned||high||standard|
|45||38||C2 (3.7 g/kg)||pinus r.||seasoned||high||standard|
|49||42||C2 (3.7 g/kg)||pinus r.||seasoned||high||standard|
|41||34||C2 (3.7 g/kg)||manufacured||seasoned||high||standard|
|42||35||C2 (3.7 g/kg)||manufacured||seasoned||high||standard|
|43||36||C2 (3.7 g/kg)||manufacured||seasoned||high||standard|
|27||24||Moisture||C2 (3.7 g/kg)||redgum||very green||high||standard|
|39||32||C2 (3.7 g/kg)||redgum||very green||high||standard|
|30||26||C2 (3.7 g/kg)||redgum||green||high||standard|
|31||27||C2 (3.7 g/kg)||redgum||very green||low||standard|
|33||28||C2 (3.7 g/kg)||redgum||very green||low||standard|
|28||25||C2 (3.7 g/kg)||redgum||green||low||standard|
|46||39||C2 (3.7 g/kg)||pinus r.||very green||low||standard|
|50||43||C2 (3.7 g/kg)||pinus r.||very green||low||standard|
|52||45||C2 (3.7 g/kg)||pinus r.||wet surface||low||standard|
|25||23||Low flow||C2 (3.7 g/kg)||redgum||seasoned||low||standard|
|35||29||C2 (3.7 g/kg)||redgum||seasoned||low||standard|
|40||33||C2 (3.7 g/kg)||redgum||seasoned||low||standard|
|47||40||C2 (3.7 g/kg)||pinus r.||seasoned||low||standard|
|48||41||C2 (3.7 g/kg)||pinus r.||seasoned||low||standard|
|51||44||C2 (3.7 g/kg)||pinus r.||seasoned||low||standard|
|3||3||Stove type||C1 (0.9 g/kg)||redgum||seasoned||low||standard|
|4||4||C1 (0.9 g/kg)||redgum||seasoned||low||standard|
|5||5||C1 (0.9 g/kg)||redgum||seasoned||low||standard|
|16||15||Constantly||C2 (3.7 g/kg)||redgum||seasoned||low*||overloaded|
|17||16||low -||C2 (3.7 g/kg)||redgum||seasoned||low*||overloaded|
Note: * indicates low flow rate from fuel insertion
For four burns, separate samples were taken in the early and late phases of the burn and four room samples were taken to establish typical background levels in the dilution air.
Emission compositions from four different appliances were studied. These were two (new) modern AS4013:1999–compliant heaters, a widely-used heater manufactured in 1985 and in well-used condition and finally, a new open fireplace insert. The first three devices were operated as controlled-combustion heaters and the fireplace insert was operated as an uncontrolled-combustion heater. The three controlled-combustion heaters were free-standing models and the fireplace insert was shrouded with insulating bricks to simulate the thermal surrounds of a domestic installation in a chimney cavity. The three new heaters were all 'burnt-in' using a series of progressively hotter burns to minimise any chance of contamination by uncured finishes or material left over from manufacture. In Table 2, appliances are identified as C1, a small size heater with a rated aerosol emission factor of 0.9 g/kg (to standard AS4013) and C2 a medium size heater with a rated aerosol emission factor of 3.7 g/kg. Emission factors given here for the two compliant heaters are the nominal values from the 'type' compliance tests, the actual heaters used were production models obtained commercially and neither had been tested previously.
Fuels used included redgum (Eucalyptus camaldulensis), jarrah (Eucalyptus marginata), bluegum (Eucalyptus globulus), softwood (Pinus radiata) and a manufactured fuel obtained commercially in Melbourne. Information from the manufacturer of this fuel indicates a typical composition of mixed softwood and hardwood chips held together by high-pressure extrusion with no binders. The density is approximately 1070 kg/m³. Together, these fuels give a good representation of the major fuel types in use in Australia as well as including two minor fuels (softwood and manufactured fuel).
The survey of fuels used in Australia as part of this study (Appendix 2) found that in south-eastern Australia, redgum is the predominant fuel-wood type (67% of fuel sales in Melbourne, 83% in Adelaide and 14% in Canberra) in Western Australia jarrah (97% of fuel sales in Perth) and in Tasmania various lower density eucalypts (93% of fuel in Hobart) that should be well represented by the bluegum. Softwood is a minor fuel in all regions with the highest fractional use in the ACT (14% of fuel sales). The survey also indicated negligible current use of manufactured fuels although these may be potentially attractive from a resource utilisation perspective in coming years. Table 3, reproduced from Appendix 3, gives the summary of the fraction of fuel types used across southeastern Australia based on fuel-merchant-supplied data for the previous year (2000 burning season).
Green redgum was sourced from Echuca in northern Victoria, and comprised two batches, one freshly cut in early 2001, and the other cut earlier, in mid 2000. Dry redgum was sourced locally in Gippsland. Jarrah was sourced from Western Australia; this consisted of a single log. This log was approximately 30-cm diameter x 4.1 m long and was estimated by the mill to have been dead for 10 years. The water content appears relatively high, although seasoning of timber is much slower for intact large logs than when the same timber is cut into firewood lengths and split. Bluegum was sourced through a Gippsland plantation company and the softwood was also obtained locally in Gippsland as a green log. Part of the green softwood batch was dried specifically for the study and the same dried batch was used for the re-wetted fuel burn.
|jarrah mill ends||20|
Water content for all the fuels was determined by HRL using procedures followed for fuel preparation for AS4013 testing. For dry redgum the water content ranged from 12.5% to 15.5%, for jarrah the moisture content was 28.8%, bluegum 21%, green redgum 24.9%, very green redgum 33.8%, manufactured fuel 9.6%, dried softwood 11.3%–14.0% and green softwood, 54.3%. For one test the dried softwood was re-wet by immersion in a tub of water for 45 minutes then air-dried for 2 hours, resulting in a moisture content of 44.4%. In all cases the percentage water quoted is taken relative to the wet fuel weight.
Two flow settings were utilised during the study. In all cases 'high' refers to the maximum setting for the appliance and 'low' refers to the minimum setting. Flow conditions for the overloaded fuel chamber burns were also the minimum setting but this was applied from the time of insertion of the fuel into the chamber. For normal low flow settings the AS4013 procedure includes an initial fuel burn-off of fuel to allow sustainable combustion in the heater to be established. This procedure was followed for establishing sustained combustion for the low flow setting samples but continuous sampling and discrete sample collections were commenced at the time of insertion of the fuel on to the prepared coal bed in the appliance. For high flow settings there was no initial burn-off. Burn-off fuel fractions are shown in Table 5. Fuel, appliance and operating parameters are given in Tables 4–6. A key is given in Table 7 for terms used in Tables 4–6.
Fuel amount and disposition in the heaters generally complied with the AS4013 requirements. For the overloaded cases, simulating an overnight burn, the heater combustion chamber was very tightly packed, giving approximately double the standard fuel loading. Fuel loads are indicated in Table 5. Dilutions shown in Table 6 (4th column) are values determined for the standard AS4013 aerosol sampling rate of 5.5 L/min. The actual dilution for each sample line is the ratio of the measured, integrated flow through the dilution tunnel and the measured integrated flow through that sample line.
|Burn||Sample||Date||Start||End||Stove||Flow||Fuel||Fuel Lot||Moisture (%)|
|23||26||20-Feb-01||12:34:41||13:25:25||C2||b low split||redgum||7e||12.8|
|25||29||21-Feb-01||17:40:39||21:16:30||C2||b low split||redgum||3b||24.9|
|27||32||22-Feb-01||13:26:30||16:02:30||C2||b low split||redgum||6c||33.8|
|28||34||22-Feb-01||20:32:00||22:30:20||C2||b low split||redgum||6b||33.8|
at 5.5 l/min
|C1||AS4013-compliant heater, nominal emission factor 0.9 g/kg|
|C2||AS4013-compliant heater, nominal emission factor 3.7 g/kg|
|NC||non-compliant heater, used, manufactured 1985.|
|OFP||open fireplace insert|
|overnight||low burn with overloaded fuel chamber, sampling from fuel loading|
|room||sample taken from room dilution air with heater running|
|b low split||sample taken from the second part of a split low burn|
|burn off||fraction of fuel load burnt on high flow setting to establish combustion|
A general picture of the variation in aerosol mass emission factor can be gauged from Fig. 4. This figure gives the mass emission factors for samples sorted first on the basis of fuel type, then heater flow and finally by fuel moisture. In this sorted sequence cases 1–4 are room samples, 5–41 eucalypt, 42–44 manufactured fuel and 45–53 softwood. The general picture shows higher emission factors for softwood, lower heater flows and increases with increasing fuel moisture. All emission factors in this report are based on the dry fuel charge for the burn.
The sequence number here does not correspond to actual sample numbers).
Room samples are labelled 1–4, eucalypt 5–41 (stippled), manufactured fuel 42–44 (diagonal stripe) and softwood 45–53 (checker).
Some simple sorting of the emission data are possible as shown in Fig. 5 which groups according to fuel type, appliance flow setting and two broad moisture categories. This figure clearly shows that use of softwood (pine) in these appliances had a very strong effect on aerosol mass emissions irrespective of whether the fuel was dried, re-wet or fresh unseasoned (green). Burning dry eucalypt on a low flow setting and green eucalypt also contribute to higher emissions, but significantly less than use of softwood (for the present test conditions). The emission factor for manufactured fuel, falls between those for hardwood and softwood as might be expected from the composition of the manufactured fuel.
Fuel type for individual samples is given in Table 4.
Fuel moisture content is indicated as a percentage of the wet fuel mass.
Expression of the emissions in terms of usable energy delivered (g/MJ), as shown in Fig. 6, tends to equalise the relative emissions from low burns but does not change the relativities already described.
A second grouping of aerosol mass emission data is shown in Figs. 7 and 8. This grouping allows a broad comparison of the different appliances for high and low settings for a common fuel type, dry redgum, with a narrow range of fuel moisture contents (12.5% to15.5% water, AS/NZS4014 1999, requires 12–16%). Figure 7 gives the mass emission data as an emission factor (g/kg of dry fuel).
Figure 8 shows the same emission data expressed relative to the useful energy output (g/MJ). The lower efficiency of the non-compliant heater is reflected in this presentation with a slightly improved emission factor relative to the compliant heater C2 at the high burn setting. The other relativities remain essentially unchanged, although the effect of the lower thermal efficiency of the open fireplace insert is also evident.
The question of the best unit for assessing aerosol mass emissions is discussed by Houck and Tiegs (1998) and Gurnsey et al. (2000). These authors consider four options, mass per hour, mass per mass of fuel, mass per fuel energy and mass per usable energy output. For a number of reasons Gurnsey et al. prefer the status quo, that is, mass of emissions per mass of fuel. Other options are also available, for example the use of different species such as CO, as in some European national standards. Given that the Australian standard is written in terms of aerosol mass per mass of fuel, and that this form is perhaps the most useful for emission inventories, emission concentrations for the rest of this report are expressed in that form.
One aspect of Fig. 7 that requires comment is the evidently poor performance of heater C2 on low flow burning dry redgum. For high flow rates burning dry redgum the mean emission factor was 1.8 ± 1.3 g/kg (± 1 standard deviation) and on low flow 12.8 ± 6 g/kg. This heater appears to perform less well than the type compliance test would suggest (3.7 g/kg averaged over high, medium and low flow rates). Individual determinations for the low flow runs ranged from 2 to 21 g/kg, which is a large range, but not particularly surprising. The present study was conducted with specific design parameters that make it different to an AS4013 compliance test and the results should not be interpreted as an audit of the specific heaters tested. For a full AS4013 test, repeated measurements are made at three flow settings and in general the operating requirements, such pre-burns, are more rigorously controlled and there are limits on acceptable variance in efficiency. More importantly, for the low flow cases with dry redgum for the present study, the flow rate setting was turned to the lowest setting after 10% of the fuel mass was burnt-off on the high setting (to establish combustion), compared with 20% for AS4013. The 10% burn-off in this case was specifically adopted to maximise the collection of the early smoldering emissions.
A criticism of the US methodology raised by Houck and Tiegs (1998) relates to the sensitivity of compliance tests to tuning of operating parameters by skilled operators. For the present study, relaxation of some of the AS4013 requirements and inclusion of data from all test burns better shows the level of variance that is intrinsic to such testing. As shown above, this can be quite large. As a further example, Cianciarelli and Morcos (2000) in a study with two heaters report mass emission factors for a conventional heater ranging from 1.7–5.2 g/kg for maple (3 samples) and 12.7–23.8 g/kg for spruce (3 samples).
Some useful insights into the factors contributing to aerosol mass emission factors can be obtained from examination of bivariate relationships between the emission factor and various appliance operating or fuel parameters. For example the relationship between mass emission factor and fuel moisture is given in Fig. 9, for all heater types and flow settings. This shows the expected increase in mass emissions with increasing fuel moisture for hardwood (at water content > 30%), although the increased emissions at moisture contents less than 15% clearly have other contributing factors. The AS4014 standard fuel moisture range, which is called up by AS4013, is 12–16%.
Emissions from softwood also show an increase with fuel moisture, but were always relatively high, (and exceed the current 4 g/kg AS4013 criterion). The increase in emissions for hardwood is associated with low power (slow) burns as shown in Fig. 10. Green and re-wet softwood follow this same general pattern whilst the main exception is from dry softwood where the burns were of short duration and high heat (power) output.
One measure frequently used for combustion quality is the combustion efficiency (C[CO2]/ΣC, where C[CO2] is the carbon emitted as CO2 and ΣC is the carbon from all gaseous emissions). The dependence of mass emission factor on combustion efficiency is given in Fig. 11, showing that the pine burns are all characterised by low combustion efficiency. Fuel moisture is one factor that, intuitively, might be expected to strongly influence combustion efficiency. In practice the influence is not strong, as shown in Fig. 12. Whilst some overall decrease in average combustion efficiency occurs with increasing fuel moisture levels the variance in emissions for any given fuel moisture level is much stronger, indicating that other factors, including fuel size, loading, density and airflow are actually more important. This is discussed, for example by Shelton (1983), also by Gurnsey et al. (2000).
The multifactorial nature of the control of wood heater and fireplace emissions is well known and summarised, for example, by Shelton (1983). Various aspects are also discussed in Ballard-Tremeer (1997), USEPA (1998a), Houck et al., (2000) and others. Whilst the present study was not designed as a process study for determination of the controlling factors, from the perspective of aerosol mass emission factor, the bivariate relationships discussed so far allow a number of tentative conclusions to drawn. Softwood performed very badly in the tested heater, grossly exceeding the AS4013 emission target of 4 g/kg. This differs somewhat from earlier work (summarised in Technical Report No. 4: Review of Literature on Residential Firewood Use, Wood-Smoke and Air Toxics), where pine and eucalypt were tested in an Australian-design heater at the University of Tasmania. In that study little difference was found between pine and eucalypt at the high and medium flow settings whilst on low setting the emissions from pine were two to three times those from eucalypt. For the present study the fuel moisture was at either end of a wide range, with one group around 11–14% (largely overlapping the 12–16% AS4013 moisture range), the other 44–54%, measured on a wet-fuel basis. In assessing the poor emission performance for pine in this study it should also be noted that the appliances tested have all been designed or optimised for burning hardwood.
In general, apart from pine, burns with higher average power (shorter and hotter burns) produced lower mass emissions. Hardwood fuel moisture, by itself, is not the major determinant for variation in mass emission factor for the fuels and conditions of these tests (note particularly the distribution of observed emission factors in Fig. 9 for hardwood with moisture levels less than 15%). For both hardwood and pine, the overall combustion quality, as reflected in the combustion efficiency, can explain the general pattern of aerosol (particulate) mass emissions across the range of fuel and flow conditions. That is, burns with high combustion efficiency tend to have low mass emissions and burns with low combustion efficiency tend to have high mass emissions. It should be noted, however, that all the pine burns fall into a relatively narrow range of low combustion efficiencies and that both the emission factor and combustion efficiency data contain noise.
Emission factors for the various aerosol and gas species can be examined in the same manner; some examples only are given here. The first of these is PAH (gas) emissions, which are shown as a function of average power output in Fig. 13 and fuel moisture in Fig. 14. Whilst these bivariate plots show no obvious systematic trends they do show clearly that the highest PAH (gas) emissions are related to the burning of softwood. One trend, which is evident in Fig. 15, is to increasing PAH (gas) emissions with increasing initial flue temperature. This measure is expected to act as a surrogate for the combustion chamber temperature during the initial phase of the burn and indicates production of PAHs during combustion. A similar relationship is evident for aerosol-phase PAH emissions, as shown in Fig.16. Elevated PAH concentrations in both cases show a fuel dependence (softwood or manufactured fuel containing softwood) and heater temperature dependence.
Launhardt (1999) used a mass balance approach to determine whether production or destruction of dioxins and PAH dominated during the combustion process. For his test conditions, combustion of untreated wood in domestic furnaces was found to be a sink for PCDD/F, but a source for PAH. PAH concentrations were found to be greater for old technology, high fuel water and low heat loads. Production of PAHs during high-temperature combustion of wood with and without additives, in oxygen-limited conditions was investigated by Khalfi et al. (2000). These authors found that PAH production was a maximum in the range 900–954 °C and that production for low fuel densities was significantly more than for high density (solid) fuel. The presence of additives (e.g. phenol resins) resulted in a slight decrease in PAH emissions. For their experimental conditions Khalfi et al. (2000) found a correlation between PAH and CO emissions. For the present study there is only limited correlation between CO and PAH emission factors, as shown in Fig. 17 for the combined aerosol and gas PAH emissions. For the hardwood samples only, correlation of CO and PAH (PAH-16 gas and aerosol) gave an r2 = 0.15 for 37 samples. Removing two obvious outliers, still only gave r2 = 0.34. As shown in Figs. 13–17 fuel type resulted in far more significant variance in PAH emissions than all the parameters for the hardwood.
Emission factors for PAH concentrations have been compiled in a number of inventories including US EPA AP-42 review documents (USEPA, 1996, 2001). Houck et al. (2001) have reviewed several aspects of the US EPA AP-42 draft, including emission factors for PAH. They point out the lack of recent references, none since 1986 relevant to non-catalytic high technology heaters and none since 1990 for catalytic heaters. PAH concentrations have also recently been determined by Cianciarelli and Morcos (2000), in a new Canadian wood heater emissions study. The Canadian study used a certified (non-catalyst) heater and a conventional heater with two fuels, maple (hardwood) and spruce (softwood).
Unlike the present study where the partitioning of PAHs was about equal for the lower molecular weight species and predominantly aerosol phase for the heavier species (C >16, MW > 200), the Canadian data indicate predominantly gas-phase. This may be a function of different sampling procedures. The Canadian study also returned somewhat lower PAH concentrations than those obtained in the present study, presumably reflecting different burning conditions.
Concentrations of PCDD/F observed in this study, are shown in Fig. 18. Note that in this, and other figures, some emission factors are plotted as negative values. This arises from the use of blank-corrected concentrations, whereby all concentrations are corrected for the mean loading found in a series of unexposed filters. They should be interpreted as less than the practical detection limit. The highest emission values were associated with eucalypt or eucalypt-containing fuel, particularly where the initial flue temperature was around 200 °C. For this study the dominant congener group is the PCDFs: on average the concentration of PCDF was close to two times that of PCDD. The mechanisms giving rise to PCDD/F have been subject to discussion for some time. Production of PCDD/F, de novo, in fires was proposed by Bumb et al. (1980) who speculated that such production was probably common to all combustion of organic material. These authors also report previous tests showing 99.95% destruction of PCDD at 800 °C, concluding that destruction was relatively incomplete at lower temperatures. More recent work on PCDD/F from wood combustion includes a study reported by Vikelsee et al. (1994) who carried out emission tests for four stoves, three fuels (spruce, birch and beech) and two operating conditions. They found that the dominant congener group were the PCDFs. All of the variable factors in their study (stove and fuel type and operating conditions) were found to be important contributors to variation in PCDD/F emissions and the introduction of other combustion parameters (CO, VOC etc) did not improve their prediction model based on these factors. As a bivariate relationship, heat output and PCDD/F concentration showed highly significant positive correlation. Anderson and Marklund (1998) in their process study of organic compounds from biofuel combustion found PCDD/F concentration variations that could not be explained by the other combustion parameters (CO2, NO etc). PCDD/F formation was observed, indicative of secondary, de novo, production by observations at multiple locations. These authors point out the need to minimise residence time in the critical temperature region of 300–450 °C to minimise PCDD/F production. They also note that it is predominantly PCDFs that are catalysed on the fly ash surface. Use of fuel containing 15% bark significantly increased PCDD/F production (ten fold). For the studies reported by Launhardt (1999), combustion of untreated wood in domestic furnaces was found to be a sink for PCDD/F, unlike PAH where production was observed. Cianciarelli and Morcos (2000) also examined differences in PCDD/F concentration between conventional and certified (non-catalyst) heaters (higher for certified) and differences between fuels (maple and spruce, higher for maple). Partitioning towards the gas-phase was observed, suggesting decoupling between the production of PCDD/F and aerosol mass. Highest PCDD/F emissions were observed for a certified heater burning hardwood (maple) and lowest for a conventional (old design) burning softwood (spruce) with the difference being a factor of 4 to 5.
Correlation between the PCDD/F and aerosol chlorine emissions is weak (Fig. 19) and there is also a weak, but positive, correlation between PCDD/F and output power (r2 = 0.12) (Fig. 20).
Most of the largest PCDD/F concentrations were found in burns where the mean combustion efficiency (C[CO2]/ΣC) was greater than 0.8 (Fig. 21), and there is a broad tendency for higher PCDD/F concentrations to occur at lower CO concentrations and CO/CO2 ratios. Probably the clearest relationship is between PCDD/F and aerosol mass emissions (Fig. 22). Large PCDD/F concentrations were observed with the lower mass emissions and vice versa. A similar dependence can be seen in the work of Cianciarelli and Morcos (2000). The present observations, taken together, support previous observations that PCDD/F production is a complex process involving a broad range of operating factors including fuel, heater design and operating conditions. The lack of correlation with aerosol mass emissions is evidence that distillation - condensation from the fuel is not the main determinant for PCDD/F concentration, but there appears to be insufficient evidence to draw conclusions on whether the process is de novo production or formation from precursors. One important observation is that testing based on aerosol mass is not a good indicator for PCDD/F and, indeed, operating the heater to minimise aerosol mass emissions (for the tested eucalypts) appears to lead to increased PCDD/F concentration. It is also likely that the designs which reduce particle mass emissions by extending residence time in the combustion zone to burn the evolved gases work against the need for minimal residence in the temperature region of 300–450 °C, which as suggested by Anderson and Marklund (1998) minimises PCDD/F production.
Emission factors for PCDD/F from residential wood heating have been developed for a number of recent inventories e.g. USEPA (1998a), Eduljee and Dyke (1996), Douben (1997), Environment Australia (1998) and are typically around 1–2 ng/kg dry fuel. For the US case a value of 2 ng WHO-TEQ/kg was taken to represent both controlled combustion heaters and open fireplaces although in other jurisdictions values of 13–28.5 I-TEQ/kg fuel have been adopted for open fireplaces. The lack of US test data and narrow uncertainty range of the emission factor have been criticized by the panel reviewing the draft US inventory data (USEPA 1998b).
Much of the emission data for PCDD/F for residential wood heating appear to be based on a common set of relatively few determinations. This includes studies in Switzerland by Schatowitz et al. (1993), Denmark, by Vikelsee et al. (1994), the Netherlands, by Bremmer et al. (1994) and Germany, Bröker et al. (1994). Results from a domestic stove with the door closed, reported by Schatowitz et al. (1993) are approximately equivalent to 0.8 ng I-TEQ/kg dry fuel for beech wood (value depends on conversions used from TEQ/m³). The same stove and fuel type with the doors open gave 1.25 ng I-TEQ/kg dry fuel. Bremmer et al. (1994) for clean wood in a cast iron stove with refractory lining, for three flow settings, found 1–3 ng I-TEQ/kg and for an open fireplace 13–28.5 I-TEQ/kg. Bröker et al. (1994) reported emissions for a wood stove on maximum and minimum setting, averaging 0.53–0.94 ng I-TEQ/kg from three determinations and from an open fireplace 0.2–1.06 ng I-TEQ/kg fuel. Vikelsee et al. (1994) tested four stoves, three fuels (spruce, birch and beech) and two operating conditions (optimum and normal). A mean concentration of 1.9 ng TEQ/kg (Nordic) was reported, overall, for clean wood combustion. Individual concentrations from the 24 determinations varied by more than two orders of magnitude (1.5 ng/m³ to 184 ng/m³) and emissions from spruce were three times greater those from beech. Concentrations of PCDD/F were also determined in the recent Canadian study for selected conventional and certified (non-catalytic) heaters, reported by Cianciarelli and Morcos (2000). Average emission factors ranged from 0.2–0.24 ng TEQ/kg wood for a conventional (old design) heater burning softwood (spruce) to 0.85–1 ng TEQ/kg wood for the certified heater burning hardwood (maple).
Values found in the present study for a range of Australian heaters and fuels, give an overall mean 6.5 ng WHO98-TEQ/kg dry fuel weight, and best estimate representative value of 4.1ng WHO98-TEQ/kg dry fuel weight (for a discussion on derivation of the best estimate emission factor, see section 4.4.2). Whilst these values are greater than some of the values cited above for controlled combustion heaters, they are quite consistent with the broader range of emissions. The small range in reported PCDD/F emissions from woodstoves is rather surprising given the possible variations in fuel type and condition, heater design and operating parameters that are known to influence PCDD/F emissions and are likely to be encountered in such independent tests. Large variance has been shown clearly with the round robin testing reported by Skreiberg et al. (1997) for other, easier to measure, emission parameters. For the present study a wide range of operating parameters were included, many of which are not typical of standardised compliance testing, and the fuel used is different to all of the previously-cited studies.
Several inventories give emission factors for domestic wood burning with pentachlorophenol (PCP) contaminated wood. Wood with preservative was not tested in the present study. Of the 1007 respondents to the survey of wood heater users across southern Australia conducted in 2000 as part of this study (see Appendix 2), 2.1% stated that they mainly burnt wood scrap in the previous winter. There was no further breakdown into the type of scrap and no information on possible contaminants.
Detectable levels of PCBs were found for only five of the congeners analyzed; 52, 101, 138, 153, 157, and for two of these congeners, 138 and 153, they were detected in only one sample each. Congener 157 was found in 19 (of 45) smoke samples, congener 52 in 17 samples and congener 101 in 7 samples. Non-zero PCB concentrations were mostly associated with eucalypt burns, were more likely to be observed for burns with lower mass emissions and generally appear to be more likely observed for burns where PCDD/F levels exceeded about a few ng TEQ/kg (Fig. 23). This suggests some possible mechanistic association, although the actual mechanism cannot be determined from the present study. Production de novo of PCBs in a laboratory-scale combustor is suggested by the results of Andersson and Marklund (1998), particularly when bark was included in the fuel, although the concentrations they observed are smaller than indicated here. These authors measured di-, tri-, penta- and hexachlorinated congeners but not tetrachlorinated congeners thus would not have measured congener 52, but should have seen congener 157, the two most abundant congeners observed in the present study. Very little information on PCBs in smoke from wood burning could be found in the literature.
The bivariate relationship for aerosol lead and fuel moisture is given in Fig. 24, and a similar relationship for mercury and fuel moisture in Fig. 25. Neither shows any evidence of functional relationships although there is perhaps some relationship with fuel batches (indicated by different moisture groupings). The clearest feature is the slightly elevated lead content in emissions from the manufactured fuel.
Formaldehyde concentrations are shown in Fig. 26 as a function of the initial temperature. Whilst higher concentrations are associated with higher temperatures there is also a fairly clear delineation based on fuel type, with the elevated concentrations being those produced from softwood fuel. The more elevated aerosol black carbon concentrations, as shown in Fig. 27, also tend to be associated with softwood and manufactured fuel. Filter samples from these fuels also appeared noticeably blacker than those from other burns. Aerosol material is emitted very rapidly during vigorous combustion near the start of a burn with hotter burns leading to a very short intense period of emission, likely to be oxygen limited and leading to higher concentrations of un-oxidised carbon, as shown in Fig. 27.
Nitrous oxide was included in the study's analysis suite as one of the species determined using FTIR. A significant difficulty with this arose however, because of the coincidence of the carbon monoxide and nitrous oxide absorbance spectra, and the very high carbon monoxide concentrations observed during all tests. This resulted in a minimum detectable N2O concentration of approximately 300 ppb, which is too large to reliably determine N2O emission factors. N2O is known, however, to be a very minor constituent of wood-smoke. This has been shown for example in both laboratory studies, including those of Lobert et al. (1990), Skreiberg et al. (1997), and Winter et al. (1999), and field measurements of smoke plumes (Hurst et al., 1994). These studies report that less than 1% of fuel nitrogen is emitted as N2O, compared for example with our observations from this study that 10–50% of fuel nitrogen was emitted as NO.
The available data have been summarised to the extent of segregation of emission factor by species, into three broad classes: hardwood, softwood and manufactured fuel. This summary is presented in Tables 8 and 9. Values presented are averages taken across all the fuel moistures and airflow settings used in the study. Uncertainties indicated are one standard error of the mean. For most, if not all of the emissions from residential wood combustion, the concentrations derived depend in a complicated manner on both the intrinsic composition of the fuel and also, in many cases critically, on how the combustion occurs. Emission factors derived from a study such as this strictly refer to the particular heaters, fuels, burn conditions etc used in the study. The accuracy of any laboratory test studies as predictors of emission levels that might be encountered in the community, including this study, must always be questioned, although such data will inevitably invite this use. The present study was not designed to exactly replicate the real-world burning pattern although a wide range of operating conditions was considered. Whilst the conditions tested represent many possible aspects of how fuel wood is burnt in the community, very wet fuel, dry fuel etc, the relative contributions of each of these factors is not necessarily proportionally represented in the test series. Emission factors presented in Tables 8 and 9 were derived primarily from an AS4013:1999-compliant heater with nominal, or type, emission factor of 3.7 g/kg (under AS4013 test conditions). None of the heaters tested gave very large mass emissions for AS4013-type operation, the well-used non-compliant heater manufactured in 1985 also returned surprisingly modest emission factors, as did the new fireplace insert. These values can be compared with a mean emission factor of 3.3 g/kg from 322 tested heaters in the Australian test data base (see Technical Report No. 4: Review of Literature on Residential Firewood Use, Wood-Smoke and Air Toxics).
|Species||Phase||Emission factors as g/kg dry fuel mass|
|Tetrachloroethylene||gas||< 7.2E-3||< 7.2E-3||< 7.2E-3||< 7.2E-3|
|Trichloroethylene||gas||< 5.8 E-3||< 5.8 E-3||< 5.8 E-3||< 5.8 E-3|
|Vinyl chloride (monomer)||gas||< 2.7 E-3||< 2.7 E-3||< 2.7 E-3||< 2.7 E-3|
|N2O||gas||< 5.3E-02||< 3.2E-02||< 1.3E-01||< 5.4E-02|
|Fluoride HF||gas||< 3.7E-4||< 2.0 E-4||< 5.4E-4||2.8E-04||1.7E-04|
|HCl||gas||6.5E-04||3.5E-04||< 3.4 E-4||< 1.1E-3||1.1E-03||3.9E-04|
|Methyl ethyl ketone||gas||1.3E-03||3.8E-04||< 1.3E-5||< 4.4E-5||< 3.7E-4|
|Methyl isobutyl ketone||gas||2.0E-02||3.0E-03||7.0E-03||5.1E-03||8.2E-02||2.8E-02||2.3E-02||2.0E-03|
Averages are taken across all fuel moisture contents and airflows.
The best estimate values were derived using weighted averages from the survey of wood burning appliance users, conducted in late 2000, see text (or Appendix 2).
To obtain an overall best estimate of emissions that could be considered typical from this test series, weighted averages were constructed, based on the operating parameters derived from the survey of wood burning appliance users conducted in late 2000 as part of this study (see Appendix 2). Factors considered for categorizing the emissions included heater type (three classes – compliant, non-compliant plus cooking stove, open fireplace + potbelly), fuel type (two classes – hardwood, pine), fuel moisture content (two classes, water > 30%, water < 30%) and loading (two classes – normal (AS4013), overloaded/overnight). In the absence of knowledge on flow rate settings used in the community, the flow settings used for testing (high and low) were treated equally. Best judgment of the investigators has been used where necessary in setting the categories and their boundaries. Possible outstanding limitations in the approach are the small number of used, older, non-compliant heaters (one) and also the ability of one fireplace insert to represent all fireplaces, particularly older, less efficient built-in masonry designs.
Sorting the test burns based on the selected categories resulted in only eight independent sets of results, from a possible 24. Emissions were averaged over the eight sets and each set was weighted using proportions of each of the eight sets derived from the survey. The actual weightings that resulted from the combination of the eight classes are shown in Table 10.
|Species||Phase||Emission factors as g/kg dry fuel mass|
|Toluene-2,4-diisocyanate||g & a||< 1E-4||< 1E-4||< 2E-4||< 1E-4|
|Methylenebis (phenylisocyanate)||g & a||< 1E-4||< 1E-4||< 2E-4||< 1E-4|
|Arsenic||aerosol||< 4E-5||7.7E-04||2.1E-04||< 3E-4||< 3E-5|
|Cadmium||aerosol||< 1E-3||< 9E-3||< 1E-2||< 1E-3|
|Chromium||aerosol||< 2E-5||< 8E-5||< 2E-4||< 2E-5|
|Lead||aerosol||< 1E-4||2.3E-03||1.3E-03||< 9E-4||< 1E-4|
|Mercury||aerosol||< 8E-5||< 5E-4||< 1E-3||< 7E-5|
|Aerosol mass loading||aerosol||4.5||0.9||4.2||0.7||15.8||4.4||4.3||0.5|
|PAH Aerosol + gas Sum (PAH16)||g & a||6.1E-02||1.5E-02||5.5E-01||1.4E-01||1.4E+00||4.9E-01||8.4E-02||2.2E-02|
|PAH 7||g & a||4.6E-03||1.7E-03||5.7E-02||1.5E-02||1.2E-01||3.9E-02||6.2E-03||2.0E-03|
|Naphthalene||g & a||5.5E-03||1.1E-03||2.4E-02||9.3E-03||2.7E-01||2.9E-01||1.4E-02||1.5E-02|
|Acenaphthylene||g & a||1.6E-02||4.1E-03||1.5E-01||4.5E-02||2.9E-01||7.8E-02||2.0E-02||3.5E-03|
|Acenaphthene||g & a||6.4E-03||1.5E-03||3.2E-02||4.8E-03||6.0E-02||1.5E-02||5.0E-03||9.8E-04|
|Fluorene||g & a||8.3E-03||1.9E-03||7.1E-02||2.0E-02||1.8E-01||5.8E-02||1.2E-02||2.7E-03|
|Phenanthrene||g & a||8.4E-03||2.0E-03||6.8E-02||1.9E-02||1.6E-01||3.5E-02||9.6E-03||1.9E-03|
|Anthracene||g & a||1.8E-03||4.4E-04||1.9E-02||5.6E-03||4.3E-02||9.3E-03||2.4E-03||4.9E-04|
|Fluoranthene||g & a||5.8E-03||1.8E-03||7.2E-02||2.0E-02||1.7E-01||5.7E-02||8.9E-03||2.1E-03|
|Pyrene||g & a||4.1E-03||1.3E-03||4.3E-02||1.1E-02||8.3E-02||3.3E-02||5.0E-03||1.3E-03|
|Benzo(a)anthracene||g & a||7.5E-04||2.0E-04||1.2E-02||2.5E-03||4.2E-02||1.4E-02||1.5E-03||4.8E-04|
|Chrysene||g & a||8.0E-04||2.3E-04||7.6E-03||1.8E-03||1.4E-02||5.9E-03||1.0E-03||2.3E-04|
|Benzo(b)fluoranthene||g & a||6.8E-04||2.4E-04||7.6E-03||2.3E-03||1.4E-02||6.8E-03||8.4E-04||2.7E-04|
|Benzo(k)fluoranthene||g & a||2.9E-04||1.1E-04||3.4E-03||9.1E-04||6.2E-03||3.3E-03||3.6E-04||1.4E-04|
|Benzo(a)pyrene||g & a||8.2E-04||3.1E-04||1.0E-02||3.0E-03||2.0E-02||1.0E-02||1.1E-03||4.3E-04|
|DiBenzo(a,h)anthracene||g & a||9.0E-04||3.5E-04||1.1E-02||3.6E-03||2.0E-02||1.2E-02||1.1E-03||4.9E-04|
|Benzo(g,h,i)perylene||g & a||5.2E-04||2.3E-04||9.1E-03||2.9E-03||1.1E-02||6.6E-03||6.3E-04||2.9E-04|
|Indeno(1,2,3-cd)pyrene||g & a||5.9E-04||2.5E-04||5.1E-03||1.5E-03||5.2E-04||2.2E-04||3.6E-04||1.2E-04|
|PCDD/F (as WHO98-TEQ)||g & a||7.5E-09||1.6E-09||1.9E-08||1.3E-08||< 1.0 E-9||4.1E-09||8.8E-10|
Note: In the 'Phase' column, entries listed as 'g & a' refer to combined gas and aerosol emissions.
|Set||Fuel moisture||Load||Fuel||Heater||Weight||Samples included|
|1||< 30%||standard||eucalypt||compliant||0.216||1-5, 19-26, 28-29, 35-38, 40-41|
|2||< 30%||standard||eucalypt||open fireplace||0.173||6, 7, 8|
|3||< 30%||standard||eucalypt||non-compliant||0.346||10, 11, 12, 13|
|4||< 30%||overload||eucalypt||non-compliant||0.127||14, 15|
|5||< 30%||overload||eucalypt||compliant||0.079||16, 17|
|6||> 30%||standard||eucalypt||compliant||0.035||27, 31-34, 39|
|7||< 30%||standard||manufactured||compliant||0.000||42, 43, 44|
|8||< 30%||standard||pine||compliant||0.020||45, 47, 48, 49, 51|
|9||> 30%||standard||pine||compliant||0.003||46, 50, 52|
Best estimate emission factors, for this study using the weightings, are given in Tables 8 and 9. Uncertainties indicated for the weighted mean are one (weighted) standard error of the mean.
Selected emission factors including some recent reports and frequently quoted sources, e.g. Larson and Koening (1993) are presented in Tables 11–14, for species covered by the present study.
|VOC unspeciated||7–27 , 8.5 ||102 ||37 , 2.6 |
|Benzene||0.6–4.0 ||0.87 , 0.31 |
|Toluene||0.15–1.0 ||0.33 , 0.18 |
|Vinyl chloride (monomer)|
|Xylenes||9.0E-2 , 8.15E-2 |
|Formaldehyde||0.1–0.7 , 0.65 |
|Methyl ethyl ketone||1.3 E-1 |
|Methyl isobutyl ketone|
|CO||80–370 ||113 , 64.1 ± 40.7 ||1.0E-2 |
|NOX||0.2–0.9 ||1.16 ||1.3 |
|SO2||0.16–0.24 ||0.18 ||1.8E-1 |
|CH4||14–25 ||13 |
|Acetic acid||1.8–2.4 , 2.25 |
Note: For data sources  see Table 15.
|Chromium||2.0E-5–3.0 E-3 ||< 4.5E-7 |
|Aerosol mass loading||7-30 , 70 ||17.3 , 11.8 ± 11.6 ||14 , 10.4 (1.7–23.8) |
|Black carbon||0.3–5 |
|Organic carbon||2–22 ||7.1E-3  POM|
|PAH Aerosol + gas Sum (PAH16)||3.2E-1 , 2.23E-1 |
|PAH 7||2.0E-2 , 2.9E-2 , 2.0E-2 |
|Naphthalene||0.24–1.6 ||1.3 E-1 |
|Acenaphthylene||9.5E-2 , 5.58E-3 |
|Acenaphthene||4.5E-3 , 5.9E-4 |
|Fluorene||4.0E-5–1.7E-2 ||1.1E-2 , 2.9E-3 |
|Phenanthrene||2.0E-5–3.4E-2 ||3.5E-2 , 8.9E-3 |
|Anthracene||5.0E-5–2.1E-5 ||6.3E-3 , 1.6 E-3 |
|Fluoranthene||7.0E-4–4.2E-2 ||8.9E-3 , 3.2E-3 |
|Pyrene||8.0E-4–3.1E-2 ||1.1E-2 , 2.4E-3 |
|Benzo(a)anthracene||4.0E-4–2.0E-3 ||8.9E-3 , 8.2E-4 |
|Chrysene||5.0E-4–1.0E-2 ||5.4E-3 , 7.9E-4 |
|Benzo(b)fluoranthene||6E-4–5E-3 ||2.7E-3 , 9.3 E-4 |
|Benzo(k)fluoranthene||8.9E-4 , 2.2E-4 |
|Benzo(a)pyrene||3E-4–5E-3 ||1.8 E-3 , 6.9 E-4 |
|DiBenzo(a,h)anthracene||2E-5–2E-3 ||6.7E-5 |
|Benzo(g,h,i)perylene||3E-5–1.1E-2 ||1.8E-3 , 3.7E-4 |
|Indeno(1,2,3-cd)pyrene||2E-4–1.3E-2 ||4.1E-4 |
|PCDD/F (as I-TEQ)||13–28.5E-9 , 0.2–1E-9  1.3E-9 ||3E-10, 2.2E-10 , 1–3E-9 ,0.2–1E-9 , 0.8E-9 |
Note: For data sources  see Table 15.
|VOC unspeciated||13 , 0.51 ||12 |
|Benzene||0.17 ||6.5E-1 |
|Toluene||5.8E-2 ||2.3E-1 |
|Vinyl chloride (monomer)|
|Xylenes||9.5E-3 ||8.3E-2 |
|Methyl ethyl ketone||2.8E-2 |
|Methyl isobutyl ketone|
|CO||58 ||47 |
|SO2||1.8E-1 ||1.8E-1 |
Note: For data sources  see Table 15.
|Cadmium||8.9E-6 ||2.1E-5 |
|Chromium||< 4.5E-8 ||< 4.5E-9 |
|Aerosol mass loading||8.8 , 0.64 (0.4–0.84) , 6.0||9.1 |
|PAH Aerosol + gas Sum (PAH16)||1.8E-1 , 1.3E-1 ||1.8E-1 , 1.6E-1 |
|PAH 7||2.0E-2 , 3.0E-3 , 1.0E-2 ||2.2E-2 , 5.0E-3 |
|Naphthalene||6.4E-2 ||8.3E-2 |
|Acenaphthylene||1.4E-2 , 1.3E-3 ||3.0E-2 |
|Acenaphthene||4.5E-3, 9E-5 ||2.7E-3 |
|Fluorene||6.3E-3 , 5.3E-4 ||6.3E-3 |
|Phenanthrene||5.3E-2 , 3.6E-3 ||2.1 E-2 |
|Anthracene||4.0E-3 , 4.8E-4 ||3.6E-3 |
|Fluoranthene||3.6E-3 , 1.5E-3 ||5.4E-3 |
|Pyrene||3.6E-3 , 1.1E-3 ||4.5E-3 |
|Benzo(a)anthracene||< 4.5E-4 , 3.3E-4 ||1.1E-2 |
|Chrysene||4.5E-3, 3.4E-4 ||4.5E-3 |
|Benzo(b)fluoranthene||1.8E-3 , 4.2E-4 ||1.8 E-3 |
|Benzo(k)fluoranthene||< 4.5E-4 , 1.1E-4 ||8.9E-4 |
|Benzo(a)pyrene||2.7E-3 , 2.2E-4 ||1.8E-3 |
|DiBenzo(a,h)anthracene||1.8E-3 , 2.5E-5 ||8.9E-4 |
|Benzo(g,h,i)perylene||8.9E-3 , 1.4E-4 ||8.9E-4 |
|Indeno(1,2,3-cd)pyrene||8.9E-3 , 1.6E-4 ||1.8E-3 |
|PCDD/F (as I-TEQ)||5.5E-10, 9.5E-10 |
Note: For data sources  see Table 15.
| Larsen & Koening (1993) / EPA (1993)|
| Environment Australia (1999), values for smokehouse continuous smoke zone|
| USEPA (1996)|
| Houck, Crouch, Huntley (2001)|
| Cianciarelli and Morcos (2000)|
| Bremmer et al. (1994)|
| Broker et al. (1992)|
| Schatowitz et al. (1993)|
| Houck and Tiegs (1998), best judgement representative value for non-catalyst wood heaters|