Impacts of wood species and moisture content on emissions from residential wood heaters

ABSTRACT

Homeowners burn wood of a wide range of species and moisture content (MC) in residential cordwood and pellet stoves. An effective emission certification test protocol must account for and accurately measure the impact of those variables in order to ensure a reasonable correlation between laboratory results and in-use emissions and to promote the design and manufacture of cleaner burning appliances. This study explored the effect of wood species and MC on emissions and efficiency in four cordwood and four pellet stoves. PM emissions were consistently lower with pellets manufactured from softwood than for hardwood species and were highly correlated with ash content. Higher MC oak fuel substantially increased PM emissions in a non-catalytic cordwood stove; however, a hybrid cordwood stove was able to meet federal emissions limits even with the higher MC fuel. The results of this study, in combination with previous research, suggest that certification tests that use softwood fuel likely report lower emissions than tests that use hardwood. Requiring hardwood and higher MC cordwood fuel in certification tests would enable the assessment of an appliance’s ability to operate well even when fuel conditions are not optimized.

Implications: The emission testing results reported in this paper call into question the adequacy of the fuel moisture content and fuel species specifications in testing protocols approved for certifying compliance with EPA’s New Source Performance Standards for cordwood and pellet stoves. We recommend changes in those specifications, including the prohibition of testing with Douglas fir and other low ash softwood species, requiring the use of cordwood test fuel with a higher moisture content, and requiring pellet stoves to be tested using hardwood pellets. Adoption of these measures would increase the replicability of tests. allow for the identification of stoves that are unlikely to perform well in the field when fuel conditions are not ideal, and, ultimately, result in the design of cleaner burning stoves.

Introduction

The US Environmental Protection Agency’s (EPA) New Source Performance Standards (NSPS) establish particulate matter (PM) emissions limits for new wood-fired residential wood heaters (RWH). The 2015 revisions to the NSPS for RWH establish two tiers of (PM) emission limits for wood stoves; Step 1 standards, which were in effect from May 2015 to May 2020, and more stringent Step 2 standards, which became effective in May 2020. A model line is certified as compliant with the NSPS if emissions from a prototype wood heating appliance, as measured by an EPA-approved testing laboratory using EPA-approved testing protocols, are consistent with the NSPS emissions limits. (US EPA 2015) Emission certification protocols include fueling, operating, and emission measurement specifications. The 2015 NSPS revisions did not include efficiency standards due to insufficient data, but require the manufacturer to submit test data on overall efficiency (heat output divided by fuel input) in certification reports. The overall efficiency is calculated from measurements. This article explores the effect of two fueling elements, wood species and moisture content (MC), on emissions and efficiency test results.

Wood species

In an effective certification testing protocol, fueling requirements must represent the species and MC of wood commonly burned by consumers. In addition, the specificity of those requirements must be sufficient to minimize inter-test variability and to produce results that are replicable and allow for comparison of appliances.

US EPA Method 28R (US EPA, 2017a and 2019), the Federal Reference Method (FRM) for certification testing of cordwood stoves, was designed to limit inter-test variability. M28R requires tests to be conducted using crib wood of a single species, Douglas fir. Crib wood is dimensional oven dried lumber of uniform size. The protocol specifies the configuration and spacing of the wood in the firebox. While the M28R protocol is designed to provide consistency in test results, the form, configuration, and species of the fuel in that test are not representative of in-use fueling practices. Cribs may burn cleaner than cordwood due to a more uniform airflow and the absence of bark or large knots (Houck, Pitzman, and Tiegs 2008). In addition, Douglas fir cordwood is not available in all parts of the country and as discussed below tends to produce relatively low emissions in comparison with other species.

With the 2015 revisions to the NSPS, EPA took an important step toward improving the representativeness of certification testing by allowing the use of cordwood instead of crib wood as a fuel. In 2018, EPA accepted two cordwood methods based on the ASTM E3053 test developed by ASTM International (West Conshohocken, PA) as broadly applicable Alternative Test Methods (ATMs). Those ATMs, designated as Alt-125 and Alt-127, were revoked on December 21, 2021 in response to concerns from state agencies regarding the test method’s stringency (US EPA 2021a).

The Integrated Duty Cycle (IDC) Test Method for Certification of Wood-Fired Stoves Using Cordwood, an operation and fueling test method developed by the Northeast States for Coordinated Air Use Management (Morin et al. 2022; NESCAUM 2021a), was approved as a broadly applicable ATM on March 31, 2021, with an updated approval letter issued on June 25, 2021 (US EPA 2021b). With the revocation of the ASTM E3053-based ATMs, the IDC is the only EPA approved cordwood stove test method.

Unlike crib wood methods, it is not feasible to limit cordwood test fuels to one species of wood, due to restrictions on the transport of untreated wood designed to reduce the spread of exotic and native forest pests from firewood to urban trees and natural forests. Such restrictions have been adopted by at least thirty US states (Andersen 2016). Therefore, cordwood tests must use locally available wood species.

The recently revoked ASTM E3053 cordwood stove testing protocol does not specify acceptable test fuel species, but instead allows testers to burn any cordwood that has a specific gravity in the range of 0.48 to 0.73 on a dry basis. That specific gravity range encompasses a wide range of wood species, introducing the potential for significant fuel-related test-to-test variability. The IDC cordwood stove protocol, as currently approved, restricts test fuel species to maple (big leaf, red or silver) or birch (black, white, yellow, or paper). Oak (red or white) and beech are being considered for inclusion in future versions of the protocol.

Updates to the NSPS promulgated in 2015 specify that the PM emissions limits in the regulation apply to pellet stoves, as well as cordwood stoves. Wood pellets are manufactured under heat and pressure from ground sawdust and woodchip byproducts. The FRM for testing pellet stoves, ASTM E2779-10, does not limit the wood species that compose the pellets used for testing. However, the NSPS requires that certification tests use pellets that have been graded under a licensing agreement with an EPA-approved third-party organization and that meet EPA’s specifications for density, dimension, fines, chlorides, and ash, and that do not contain demolition or construction waste or other prohibited materials. Laboratories or manufacturers are allowed to specify or supply the fuel used in the testing [40 CFR § 60.534(e)].

Previous research on the impact of fuel species in wood-burning stoves has documented inter-species differences in PM emissions, although the magnitude and nature of those differences vary from study to study. Factors that may affect study results include stove design, control technology, operating conditions such as air setting, measurement methods and parameters, and whether or not start-up emissions were measured.

In studies that used stove types and fuel species common in the United States, oak cordwood fuel consistently produced higher PM emissions than Douglas fir cordwood. Hays et al. (2003) reported emissions factors (EF), in grams of fine particles (PM_2.5) per kilograms of dry fuel burned (g/kg), that were 1.5 times higher for white oak than Douglas fir with a dry-basis MC of 24.4%- 29.2%, and 2.1 times higher when the MC was 12.2%. Emissions rates (ER) in grams PM_2.5 per hour (g/h) were 1.5 times higher in the oak than the Douglas fir for the higher moisture woods and 2.5 times higher for the dryer wood.

Kinsey, Kariher, and Dong (2009) reported that the PM_2.5 EFs (g/kg) and ERs (g/h) measured in burns of wet Northern red oak (MC 28%) were both 1.7 times higher than those for wet Douglas fir (MC 22%), while the EF and ER in dryer wood (MC 11%) were 1.4 and 1.6 times higher in oak than in Douglas fir. The burns in this study were conducted in an EPA certified stove and included start-up, high and low fire phases.

Fine, Cass, and Simoneit (2004) measured PM_2.5 emissions from a wood stove operated with and without a catalyst. Measurements started immediately prior to ignition and continued until “particle-sizing instrumentation showed few additional particles being emitted, typically occurring 10 to 20 min after the fire began smoldering with no visible flames.” The white oak EF was 3.9 times higher than that for Douglas fir when the stove was operated without the catalytic controls and 1.8 times higher when the catalyst was deployed.

Fine et al. also conducted burns without the catalytic controls using three additional fuels that are common in the United States: red maple, sugar maple, and loblolly pine. The reported EF for red maple, 0.88 ± 0.16 g/kg, was lower than those of the other fuels, which were 1.1 ± 0.2 for Douglas fir, 1.4 ± 0.2 for sugar maple, 2.0 ± 0.3 for loblolly pine, and 3.4 ± 0.5 g/kg for white oak.

In 2017, EPA funded research to better characterize the impact of the species and form (crib versus cordwood) of fuel on PM emissions (Hearthlab Solutions 2017). The study used a pre-1988 NSPS stove and a hot-to-hot test similar to the M28 protocol which did not include measurements during startup or cool down periods. Red maple, red oak and white birch cord and crib wood fuel were tested at low and high air setting. Douglas fir was included in the test only as crib wood because cordwood of that species was not available in Vermont, the state where testing was conducted. The specific gravities of those fuel species are in the range allowed in the ASTM procedure.

The EPA study did not show consistent differences between cordwood and crib wood tests of the same fuel species. Some interspecies differences in cordwood results were noted, but due to the limited number and increased variability of cordwood burns, it is not clear that those differences would be reproducible. Inclusion of start-up emissions may also alter those results. As shown in Figure 1, the crib wood EF and ER for Douglas fir were substantially lower than those for the other crib wood species. The EF decreased linearly with increased burn rate. The ER, which is affected by both emissions and runtime, was highest at moderate burn rates. Curves shown are linear fits for EF and polynomial fits for ER.

Figure 1. Crib wood species total PM (a) emission factors (g/kg) and (b) emission rates (g/h) (Hearthlab Solutions 2017).

Moisture content (MC)

The MC of fuel can also affect the efficiency and emissions of residential cordwood heaters. The MC of freshly cut cordwood ranges from about 30% to more than 100% and can vary considerably among species, between trees of the same species, and even in wood cut from the different parts of the same tree (Glass and Zelinka 2010). Dry wood may produce hotter, more complete burns, resulting in higher efficiency and lower emissions (Price-Allison et al. 2019) and changes in the composition of the PM emissions (Price-Allison et al. 2021) compared to wetter wood. Shen et al. (2013) reported a strong association between MC and EF of PM, organic carbon, and PAH with MC ranging from 5% to 27%. However, cordwood that is too dry (less than about 15%) can burn too fast, resulting in uncontrolled burn conditions that can have poor combustion and higher emissions from insufficient airflow (Wilton, Smith, and Webley 2006).

It is generally recommended that harvested wood be split, stacked, covered, and allowed to dry until an internal MC of 20% is achieved before burning. Seasoning firewood can take six months to two years, depending on the species. Many factors impact drying time, including local climate, the time of year that the wood is harvested, and the species (University of Maryland Extension 2018). Very dense wood like oak, a common firewood, requires longer seasoning times.

Firewood producers sometimes sell wood as “seasoned” only a few weeks or months after it is split. A University of Tennessee study surveyed bags of firewood purchased from a supermarket in Knoxville in October 2007 (University of Tennessee Agricultural Extension Service 2010). The MC of each piece of firewood in the bags was measured using the oven-dry method. The wood tested was all oak and the pieces were cut and split to convenient sizes. Despite being labeled as “Seasoned Firewood,” the average MC measured was 66%, which is only slightly less than the green MC of oak. Because many buyers do not know how to tell if wood is properly seasoned, it is likely that a significant amount of the cordwood used in residential wood burning devices has a much higher MC than the wood used in certification tests.

Previous studies have reported higher emissions from burning freshly cut wood than in burns of the same lot of wood after seasoning. In the 2003 study by Hays et al. cited above, the PM EFs (g/kg) and ERs (g/h) were 94% and 68%, respectively, higher in white oak with MCs of 28.4–29.2% than in the same fuel with a MC of 12.2%. For Douglas fir, the difference was even greater, the EF and ER for the wetter wood (MC of 24.2%) were 174% and 173% higher, respectively, than in the wood with a 12.2% MC. Kinsey, Kariher, and Dong (2009) similarly reported that the EF (g/kg) and ER (g/h) in burns of wet Northern red oak (MC 28%) were 73% and 67% higher than in that wood with a MC of 11%, while wet Douglas fir (MC 22%) produced an EF and ER that were 57% and 42% higher than those of 11% MC Douglas fir, respectively.

EPA certification test protocols for cordwood stoves allow the use of wood with a MC between 19% and 25%. A recent review by NESCAUM of certification test report data determined that those tests generally use wood with a MC in the lower quartile of the allowable range. In the above cited studies, the wetter wood tested was above or within that range, while the lower MC wood was considerably lower than the range that would be allowed in certification testing.

Current study

NESCAUM conducted a series of experiments to better characterize the impact of fuel species and MC on PM emissions and heating efficiencies of residential wood stoves and to inform an evaluation of the adequacy of the species and MC specifications in current certification methods. Many of the studies that previously documented associations between fuel species or MC and appliance performance and emissions were not conducted on EPA certified stoves or did not use EPA-approved certification testing methods. This study employed EPA approved operational protocols to test cordwood and pellet stoves certified as compliant with the Step 1 or Step 2 NSPS emissions limits using fuels of varying species and MCs.

Methods

PM was measured using a 200 to 230 cubic foot per minute dilution tunnel, with PM collected on Emfab™ (Pallflex® TX40 filters, Pall Life Sciences, Ann Arbor, MI) Teflon-coated glass fiber filters (Allen et al., 2017) using integrated gravimetric PM measurement methods without any size cut, consistent with EPA Method 5 G (US EPA 2017b) and ASTM E2515 (ASTM 2017). Filter weights were determined with an Ohaus Plus analytical balance (100 × 0.0001 g). Currently, EPA only allows certification tests to use ASTM E2515, a proprietary method similar to Method 5 G. The dilution tunnel allows the flue gas to cool and PM to condense prior to measurement, and reduces the sample dewpoint to below the filter temperature. Stack carbon monoxide (CO) was measured according to EPA Method 10 procedures (US EPA 2017c) and stack carbon dioxide (CO₂) according to EPA Method 3 (US EPA 2017d) using a Horiba VIA510 NDIR analyzer. Burn rates were measured with a Fairbanks PLF-HR34 stove scale according to Method 5 G.

Since it is difficult to properly match a burn phase to a manual filter pull sample, for PM data stratified by burn phase, a Thermo Scientific (Franklin, MA) Model 1405 Tapered Element Oscillating Microbalance (TEOM™) was used without any size cut or sample conditioning and operated at 30°C. The TEOM provides highly time-resolved (one-minute average) PM emissions data throughout the test period. The TEOM method as used here is described in detail in a companion paper in this issue (Allen et al. 2022). The Standard Operating Protocols for the operation of the TEOM in a dilution tunnel used in this study are publicly available (NESCAUM 2020, 2021b).

Cordwood stoves

Four cordwood stoves were tested to evaluate the effect of fuel species and MC on PM emissions and overall thermal efficiency. Table 1 summarizes the characteristics of those stoves.

Table 1. Characteristics of cordwood stoves used for species and moisture tests.

Stove #	Construction Type	Firebox Size	Emission Controls	EPA Certification
Stove 6	Steel	2.2 ft^3 (0.062 m^3) Medium	Non-catalytic tube	Step 1
Stove 7	High mass*	1.9 ft^3 (0.054 m^3) Medium	Hybrid – catalytic and non-catalytic	Step 2
Stove 15	Steel	0.8 ft^3 (0.023 m^3) Small	Non-catalytic tube	Step 2
Stove 23	Cast iron	1.8 ft^3 (0.051 m^3) Medium	Non-catalytic, optional catalyst	Step 2 (with and without catalyst)

*High mass stoves are designed to trap heat and release it slowly into the building’s living space. The thermal storage is usually provided by a large mass of a dense material.

Tables 2a and 2b list the tests performed on the cordwood stoves. Tests to evaluate the effect of wood species on PM emissions were performed on Stoves 6, 7, and 15 and the impact of MC was evaluated in Stoves 6, 7, and 23. The Stove 23 tests were conducted with and without the optional catalyst (identified in Tables 2a and 2b as 23-cat. and 23-No cat.).

Table 2a. Cordwood stove species tests.

Stove	Protocol	Species
6	ASTM E3053	Red Maple Red Oak
		Beech
6	IDC	Red Maple
		Red Oak
7	IDC	Red Maple Red Oak Spruce
		Poplar
15	IDC	Red Maple Red Oak
		Beech

Like the NSPS emissions limits, PM emissions measured with all EPA-approved wood stove testing methods are not size fractionated. Overall efficiency of cordwood stoves were calculated from CO and CO₂ emission measurements according to calculation procedures specified in the IDC protocol. Overall efficiencies reported in this paper are based on the higher heating value (HHV), which includes fuel moisture in the fuel weight.

The MC of the fuel was measured using the procedures specified in the IDC cordwood stove protocol with a Delmhorst G-30 Moisture Meter. That instrument calculates MC from the electrical resistance of wood measured between two electrode pins. The IDC protocol specifies that the electrical resistance meter must be calibrated to measure fuel moisture to within 2% MC.

The IDC MC measurement method is based on recommendations in Smith et al. (2014), which demonstrated that an error of less 2% can be obtained in resistance meter MC measurements for red oak cordwood pieces with MCs between 17% and 25%, as compared to measurements using the conventional oven drying method. This method is designed to take into account the variability in MC within, as well as between, wood pieces and to avoid possible loss of volatile compounds. In the current study, the MC of each fuel piece was calculated as the average of three moisture meter readings, one from each of three sides of the piece, measured parallel to the wood grain. The MCs reported in Table 2b are the average of the MCs measured for each wood piece consumed in the run, weighted by the mass of the piece. Wood with a MC greater than 29% was not tested since it can be difficult to successfully complete a standard test burn using high MC wood without substantial operational interventions (stirring, opening door) to maintain a fire at lower stove air settings and lower burn rates.

Table 2b. Tests evaluating impact of MC (%, dry basis) on cordwood stove performance.

Stove	Protocol	Species	Dry Mean	Dry SD	Wet Mean	Wet SD
6	IDC	Red Oak	21.4	0.5	28.1	0.2
7	IDC	Red Oak	21.2	0.4	28.7	0.3
23-No cat	IDC	Red Maple	21.1	0.3	27.0	0.7
23-cat	IDC	Red Maple	21.2	0.1	27.2	0.6

To further examine the reproducibility of moisture measurements, we calculated the 95% confidence interval of the unweighted mean MC of the pieces burned in each of the study runs. As shown in Figure 2, the confidence interval was less than ±1% MC for all but one of the runs and the wet and dry fuel had distinctly different MCs.

Figure 2. Distribution of MCs in fuel pieces used in moisture experiments (error bars represent 95% confidence intervals).

A complete test was performed on each species and MC tested. Complete ASTM E3053 tests consist of two runs, which were conducted on consecutive days. Both runs begin with a start-up/high air setting phase. After 90% of the high burn load is consumed, a medium air setting burn is conducted in one run and a low air setting burn in the other. IDC cordwood tests consist of three replicate runs, each with four segments designed to represent a range of typical fueling and operating conditions. The IDC method is described in detail in Morin et al. (2022) and is summarized in Table 3. In both methods, stoves are placed on a calibrated platform scale that is tared prior to the start of each test run to measure fuel consumption during that run.

Table 3. Summary of IDC cordwood protocol.

Segment	Fueling and Operating Parameters
Start-up	Cold Start, kindling density – 1 lb/ft^3 (16 kg/m^3) max, starter fuel-3 lb/ft^3 (48 kg/m^3) Air fully open. Burn 75% of load, by weight.
Reload 1 High-fire	Load density – 7 lb/ft^3 (112 kg/m^3), small cordwood pieces, random loading pattern. High air setting. When 50% is burned, switch to lowest air setting. Burn 90% of load.
Reload 2 Maintenance	Load density – 5 lb/ft^3 (80 kg/m^3), random loading pattern. Lowest air setting. Burn 90% of load.
Reload 3 Overnight	Load density – 12 lb/ft^3 (192 kg/m^3), mix of small and large cordwood pieces, loaded in one direction to maximum volume. Lowest air setting. Burn 90% of load.

Pellet stoves

The study evaluated the impact of the composition (wood species and ash and MC) of seven commercially available pellets on emissions in four pellet stoves using the FRM protocol for certifying pellet stove NSPS compliance, ASTM E2779-10. The pellet stoves are equipped with an automatic ignition system which eliminates the need to start the fire manually. Pellets are loaded into a storage hopper and then are automatically fed into the stove’s burn chamber by the stove’s auger. The hopper can either be located at the top or bottom of the stove. Study Stoves 10 and 14 are bottom-fed appliances and Stoves 11 and 13 are top-fed. All study stoves are currently certified as Step 2 compliant. All of the pellets tested meet EPA certification testing fuel requirements.

The species and MC of the test pellets and the number of ASTM E2779-10 tests conducted with each pellet on each stove are listed in Table 4. The E2779-10 is a hot-to-hot test that integrates emissions measured in a 60-min high burn followed by a 120-min medium burn and then a 180-min low burn. Three runs were conducted with three pellet types in two of the stoves to evaluate the replicability of results.

Table 4. Pellet composition and tests.

Pellet#	Pellet Type	MC%	Ash %	Number of Tests per Stove
				Stove 10	Stove 11	Stove 13	Stove 14
1	Eastern White Pine	3.3–4.2	0.21	1	1	3	1
2	Douglas Fir	5.6–6.2	0.30	1	1	1	1
3	Douglas Fir	2.3–3.1	0.16	3	1	1	1
4	Hardwood	5.0–5.9	0.50	3	1	1	1
5	Hardwood	4.8–5.8	0.48	1	1	3	1
6	Hard-soft Mix	4.2–5.0	0.52	1	1	1	1
9	Douglas Fir	6.8	0.23	3	0	3	0

Pellet MC was measured using ASTM E871, which determines moisture by measuring the weight loss in a sample when heated according to the specified temperature, time, and atmosphere conditions. A replicate of this test was completed to document the range of pellet MCs in the fuel. Ash content was measured as the percentage of residue remaining after dry oxidation of a wood sample using ASTM D1102. The ash content of the hardwood pellets, 0.48% to 0.50%, was higher than in softwood pellets, 0.16% to 0.30%. However, the pellets labeled as a hardwood-softwood blend had the highest ash content (0.52%).

Results

Cordwood stoves – MC

The effect of MC was initially evaluated by burning wet and dry (properly seasoned) oak fuel in Stoves 6 and 7. As shown in Table 5 and Figure 3, the Stove 6 wet oak runs produced significantly higher PM emissions and had a significantly lower efficiency than the seasoned oak burns. The 95% confidence intervals for the mean of the PM metrics for the wet and seasoned oak tests, calculated parametrically, did not intersect. In all phases of the Stove 6 runs after the start-up phase, PM ERs, in g/h, and PM EFs, in g/dry kg fuel, were higher in the wet oak burns than the dry oak burns, as shown in Figure 4, an indication that MC-related emission differences observed in that stove occur across the range of operating and fueling conditions represented by the IDC phases.

Table 5. Effect of fuel MC on emissions and efficiency – mean [range].

Stove	Fuel	Emission Rate (g/h)	Emission Factor (g/kg)	Overall Efficiency (%), HHV basis
6	Seasoned Oak (21%)	13.4 [10.4–17.1]	6.9 [5.2–8.9]	61.6 [61.0–62.6]
	Wet Oak (28%)	34.1 [26.7–41.2]	18.0 [13.7–22.4]	57.1 [56.8–57.5]
7	Seasoned Oak (21%)	2.2 [1.4–3.3]	1.4 [0.9–2.1]	68.8 [67.8–69.9]
	Wet Oak (29%)	3.4 [2.6–4.7]	2.4 [1.9–3.1]	67.9 [67.8–69.5]
23 No Cat.	Seasoned Maple (21%)	7.6 [4.9–11.4]	8.2 [2.7–13.6]	60.8 [55.2–66.1]
	Wet Maple (27%)	4.9 [2.9–6.4]	3.1 [1.6–4.3]	63.2 [60.4–66.2]
23 Cat.	Seasoned Maple (21%)	4.0 [2.7–6.1]	2.5 [1.7–3.5]	n/a
	Wet Maple (27%)	5.5 [3.5–8.7]	4.4 [1.9–7.6]	n/a

Figure 3. (a) Comparison of PM emission rates (g/h), (b) emission factors (g/kg) and (c) HHV efficiencies (%) in wet and dry fuel (average, minimum, and maximum).

*Stove 23 Cat efficiency available for only one run with each fuel type.

Figure 4. Comparison of Stove 6 (a and b) and Stove 7 (c and d) wet and properly seasoned (dry) wood emission rates and emission factors by IDC phase (average, minimum, and maximum).

In Stove 7, however, MC did not have a substantial impact on emissions metrics. Figure 4 shows that Stove 7 performed consistently well with both wet and seasoned fuel in all phases after start-up. While Stoves 6 and 7 both have medium-sized fireboxes, Stove 6 is a Step 1-certified stove with non-catalytic controls and Stove 7 is a much lower emitting hybrid (catalytic and non-catalytic controlled) Step 2-certified appliance. The Stove 7 IDC results met the NSPS Step 2 PM emission limit for cordwood tests of 2.5 g/h, even with the wet oak fuel. These results suggest that good stove design and control technology can enable stoves to perform well, even when the operator burns fuel that has not been appropriately seasoned.

Stove 23, which has been certified as Step 2 compliant with and without optional catalytic controls, was used to further investigate the role of technology in MC-related emissions. A total of 12 IDC runs were performed on that stove, three each with wet and dry (properly seasoned) maple fuel and with and without the catalyst. The results of those tests are shown in Table 5 and Figure 3. Note that HHV efficiency is not reported for the catalytic runs due to a lack of sufficient CO and CO₂ data needed to calculate those efficiencies in several of those runs.

When the stove was operated without the catalyst, the PM ER and EF in the dry maple runs were more variable and, on average, higher than in the wet maple runs, although there was overlap in emissions of individual runs in the two MC groups. Deployment of the catalyst reduced the dry maple fuel emissions but was less effective in reducing emissions in the wet fueled runs. As a result, the average EFs and ERs in the Stove 23 catalytic tests were lower in the dry maple than in the wet maple runs, although there was considerable overlap in individual runs in the two fuel groups.

Cordwood stoves – fuel species

As discussed above, fuel species impacts were evaluated in the study using both ASTM E3053 (the recently revoked EPA cordwood ATM) and the IDC protocol that is now the only EPA cordwood ATM. PM emissions rates measured for Stove 6 high, medium, and low burn rate ASTM runs are shown in Table 6. Although the low and medium burn rate ERs were lower in the maple-fueled tests than in those that used the other fuels, the lack of replicate runs for those phases precludes a determination of the significance of that difference. Further, the variability in the ERs measured in the replicate high-burn rate of the maple and oak runs was too large to allow for interspecies comparisons. Therefore, no conclusions about species impacts could be drawn from the ASTM cordwood test data. This is consistent with the earlier crib and cordwood work (Hearthlab Solutions 2017) when Douglas fir is excluded.

Table 6. Stove 6 ASTM E3053 PM emission rates (g/h) by fuel species.

	High Burn Phase	Medium Burn Phase	Low Burn Phase
Maple	2.4	4.8	4.6
	14.6	n/a	n/a
Oak	9.7	6.7	9.8
	4.0	n/a	n/a
Beech	3.3	5.5	7.3
	3.0	n/a	n/a

Figure 5 shows the ERs, EFs and HHV efficiencies for the entire IDC runs, by fuel, for the three stoves tested using that test with multiple fuel species. In all three stoves, the ERs for the red maple and oak IDC tests were similar. In Stove 7, the average ER for the poplar runs was lower and the average spruce ER was higher than those for oak and maple, but the results of individual runs in the fuel groups overlapped. Similarly, the average Stove 15 beech ER was higher than in the corresponding oak and maple runs. In Stove 7, the Step 2-certified hybrid stove, the ERs measured in the spruce runs were highly variable, while very little variability was seen in the runs that burned poplar. Note that efficiency data are available for only one of the Stove 7 poplar runs and were not available for the Stove 15 beech runs.

Figure 5. IDC runs with different fuel species for Stoves 6, 7, and 15: (a) PM emission rates, (b) PM emission factors, (c) HHV efficiencies (average, minimum, and maximum).

Pellet stoves

Six different pellets (Pellets 1–6) were tested in all four of the study’s pellet stoves and an additional pellet (Pellet 9) was tested in two of the stoves. Figure 6 shows the PM ER and EF measured for each pellet in each stove. In the graphs in that figure, the pellets are ordered by ash content.

Figure 6. (a) PM Emission rates and (b) emission factors by stove and pellet (where available, error bars indicate minimum and maximum).

As shown in Table 4, three replicate tests were performed in Stoves 10 and 13 to assess reproducibility. The bars for those pellet-stove combinations in Figure 6 represent the average of the runs and the error bars show the maximum and minimum values measured. All other pellet and stove combinations were tested once.

In all of the stoves, the ERs and EFs measured in the runs that burned hardwood (Pellets 5 and 6) were considerably higher than in those that burned Douglas fir (Pellets 2, 3, and 9). The average ER for the hardwood pellet runs was 2.2 times higher than that for the Douglas fir runs in Stove 10, 2.5 times higher in Stove 11, 3.0 times higher in Stove 13, and 2.7 times higher in Stove 14. This finding is important because pellet stoves can be certified using Douglas fir pellets and those tests may under-represent emissions in the field. Note, however, that pellet stoves tend to produce lower emissions than cordwood stoves, and the ERs measured in 33 of the 38 test runs were lower than the 2.0 g/h emission standard specified in the NSPS.

As expected for pellet stoves, after startup PM emissions were relatively stable during a run. The reproducibility of pellet stove PM emissions allows a clear assessment of the influence of species. Figure 7 shows PM ER data from the TEOM for three softwood (two Douglas fir and one eastern pine) and three hardwood ASTM E2779 pellet test runs on stove 11, with a clear separation of soft and hardwood ERs. One-minute TEOM PM data have been smoothed with a 15-min running average to damp the short-term cycling of emissions due to variations in auger fuel feed rate.

Figure 7. Stove 11, 15-minute running average TEOM PM rate with soft and hardwood pellets.

The reported MC in the pellets ranged from 2.3% to 3.1% in Pellet 3 to 6.8% in Pellet 9, but as expected with fuel this dry and the strong species influence, no relationship was observed between MC and PM emissions for pellets.

Discussion

Previous studies have documented significantly higher EFs in tests that burned oak and Douglas fir cordwood with MCs between 22% and 29% than in burns of the same wood with lower (11–12%) MCs. Since the current EPA approved cordwood certification testing protocols limit the MC of test fuel to 19–25%, the MC in the dry wood in those previous studies is considerably lower than the allowable range used in the current tests.

In the current study, the MC measured in the seasoned wood, 21%, was toward the lower end of the allowable certification testing range. The MCs of the wetter fuel in those tests, 27–29%, were somewhat above the higher end of the allowable range, but consistent with MCs in wood commonly available to stove owners and much lower than unseasoned green wood.

In Stove 7, the well-controlled Step 2 certified stove equipped with hybrid technology, burning higher MC oak fuel did not substantially increase PM emissions. However, in Stove 6, a Step 1 certified stove with noncatalytic controls, the wetter fuel produced considerably higher emissions than the drier wood and that difference was observed across a range of operating and fueling conditions. The average ER for the wet oak test runs in that stove was more than twice as high as that for the dryer wood. This suggests that for some stoves a test conducted with fuel with a MC at the lower end of the currently allowed range may under-represent the emissions that would occur in consumer use, since operators may not use wood that has been properly seasoned.

It is also interesting to note that while the PM emissions in the wet and dry maple runs in Stove 23 were not substantially different, the optional catalyst appeared to be more effective in reducing emissions in the dry fuel runs than in the wet fuel runs. This observation is based on limited data and is not conclusive, however it is an indication that the relationship between MC and control technology on PM emissions bears further study.

As discussed above, previous studies reported that Douglas fir fuel tends to produce lower PM emissions than other species, including oak. The current study did not test Douglas fir cordwood, due to the lack of availability of that species in the eastern United States, where the testing was conducted. However, the pellet stove tests demonstrated that pellets composed of Douglas fir consistently produced considerably lower PM emissions than hardwood pellets. Softwoods generally have lower ash content than hardwoods (Risse and Gaskin 2013). Figure 8 shows PM EF versus ash content for two Douglas fir, one eastern pine, two hardwood, and a hard/softwood mix pellet. These are the same six pellets shown in Figure 7. Ash is a robust predictor of PM emissions, with R² of 0.89.

Figure 8. PM EF vs. pellet ash content for three soft and three hardwood pellets.

This result is similar to the R² of 0.94 for wood pellet ash and PM ER reported by Sippula et al. (2007). Pellets burn very reproducibly relative to crib or cordwood and thus make relationships between fuel characteristics and PM emissions much easier to observe. It is likely that this relationship also exists for crib and cordwood; Figure 1 shows a substantial difference between Douglas fir and the other crib wood species. Hays et al. (2003), (Table 1) reported a cordwood white oak PM_2.5 EF 2.1 times higher than for Douglas fir at a MC of 12%, which is consistent with our hardwood/softwood pellet results at a MC below 10%.

PM emissions from larger and hotter-burning pellet systems used in Industrial, Commercial, and Institutional (ICI) boilers that operate at much higher temperatures than residential wood heaters (RWH) are typically dominated by ash. However, RWH operate at lower temperatures where carbonaceous aerosol is not completely burned off, and it is not clear from the literature if the increase in RWH PM emissions with higher ash content is driven by ash emissions directly, either fly ash or condensed ash species such as KCl, or also by increases in carbonaceous PM emissions.

There is substantial literature on the relationship between biomass ash content and PM emissions. Rabaçal and Costa (2015) cite several studies that show PM emissions are strongly correlated with the ash content and ash composition of biomass pellets, even across a range of operating conditions. Kortelainen et al. (2018) reports that inorganics accounted for at most 10% of the PM1 mass emitted from burning cordwood in a modern masonry heater, suggesting that ash emissions alone cannot explain the difference observed between Douglas fir crib wood and the other species as shown in Figure 1. Vicente et al. (2015) showed that higher K₂O in the fuel degraded overall combustion quality and increased carbonaceous PM emissions (organic carbon, soot) and CO in a pellet stove. Du, Lin, and Glarborg (2021) added varying amounts of KCl to cordwood in a modern stove and reported an increase in carbonaceous PM emissions with KCl, stating this could be due to an increased concentration of ash nucleus or particles and that ash condensation nuclei from KCl could promote condensation of semi-volatile organic compounds that otherwise might not condense. Further research is needed to determine the mechanisms that are involved in the association between ash and PM emissions for RWH appliances, and to what extent the increased emissions are due to inorganics (ash) or carbonaceous PM for both pellet and cordwood stoves.

ASTM E3053, which until December 2021 was approved as an EPA broadly applicable cordwood ATM, allowed the use of Douglas fir, a species which is generally available for use as a stove fuel only where it is locally grown. These data suggest that typical in-use PM emissions could be higher than the emissions measured in certification tests that used this species.

The study did not see substantial emissions differences in burns using three other hardwood fuels that are allowed in ASTM E3053 testing; maple, oak, and beech. Tests in the well-controlled Stove 7 using spruce fuel produced inconsistent emissions; however, spruce has a specific gravity outside of the range allowed by ASTM E3053. The IDC cordwood stove protocol, as currently approved, restricts test fuel species to maple (big leaf, red, silver) or birch (black, white, yellow, paper), but additional species are being considered for inclusion in future versions of the protocol. Any additional species should be limited to hardwoods to minimize test variability.

The study showed that PM emissions from burning pellets composed of softwood (Douglas fir or eastern white pine) were considerably lower than the hardwood pellets tested. In addition, in two of the stoves (10 and 14) PM ERs were lower in runs that burned Douglas fir pellets than those that burned a softwood mix or another softwood. Two of the four study pellet stoves (10 and 11) met the Step 2 limit with all of the pellets tested. Stove 13 exceeded that limit in two of the three runs that used one of the hardwood pellets and Stove 14 exceeded the limit when tested with the two hardwood pellets and the hard-soft mix. The ash content of the softwood pellets was considerably lower than those of the hardwood pellets; that property is important in defining acceptable pellet fuel to be used in certification tests in the future. The EPA is now considering changes to the current pellet stove testing protocol to better represent field operating conditions. Those changes may lead to further exceedances of the NSPS ER limit when hardwood pellets are burned.

Conclusion

The results of recent emission testing call into question the adequacy of the MC and fuel species specifications in the currently approved NSPS cordwood and pellet certification test methods. The study results demonstrated that emissions associated with burning softwood pellets were lower than those that burned hardwood pellets. In addition, in some of the stoves, Douglas fir pellets produced emissions lower than produced by other softwood. Previous research discussed in this paper has documented that Douglas fir also tend to produce lower emissions than those other fuels in cordwood stoves. Therefore, certification testing of cordwood stoves with Douglas fir likely underrepresents PM emissions in the field, especially in areas of the country where Douglas fir cordwood is not available to the consumer.

The EPA should also consider requiring certification tests to be performed using cordwood fuel with a higher MC in order to assess appliance performance in a fuel MC range more representative of in-use conditions. In some cordwood stoves, PM emissions were considerably higher when the MC was slightly above the range currently allowed in testing protocols than with fuels in the lower end of that range.

Certification test methods for pellet stoves should require the use of hardwood fuels and should specify a range of allowable ash and MC to decrease the variability in test results. Ash contents are higher in hardwood than in softwood; therefore, a minimum ash content specification would be an easily documented proxy for eliminating testing on softwood. Refinement of the species and moisture specifications in cordwood stove testing procedures would allow for the identification of stoves that are unlikely to perform well with fuels typically employed in the field. The relationships between MC, fuel species (e.g., ash content), stove technologies and PM emissions and efficiency should be studied further to promote the design of lower emitting appliances.

Acknowledgment

The authors thank Dr. Ellen Burkhard, NYSERDA Project Manager and Dr. Thomas Butcher, Brookhaven National Laboratory for helpful comments throughout the project. The opinions expressed in this report do not necessarily reflect those of NYSERDA or the State of New York. Mention of product manufacturer names or trademarks does not imply endorsement by NESCAUM or NYSERDA

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from NESCAUM upon reasonable request to NESCAUM’s Executive Director.

Additional information

Funding

This work was supported by the New York State Energy Research and Development Authority [Agreement #123059].

Notes on contributors

Barbara Morin

Barbara Morin is an Environmental Analyst at NESCAUM.

George Allen

George Allen is the Chief Scientist at NESCAUM.

Arthur Marin

Arthur Marin is the former Executive Director of NESCAUM.

Lisa Rector

Lisa Rector is a Policy and Program Director at NESCAUM.

Mahdi Ahmadi

Mahdi Ahmad, Austin, TX, is a NESCAUM consultant.