Field-scale treatment of landfill gas with a passivemethane oxidizing biofilter

Andrew Philopoulos, Christian Felske, and Daryl McCartney

Abstract: Municipal solid waste landfills are a major contributor to global methane emissions, a potent greenhouse gas. Atreatment alternative was evaluated by installing three biogenic methane oxidizing biofilters into the landfill cover at theLeduc and District Regional Landfill (Alberta). Mature yard-waste compost was used as the biofilter medium. The results,collected over a period of 10 months, showed that two sites performed well as low surface emissions (20 8C) and mois-ture (>0.25 L L1) levels to support a high level of biogenic activity, the former despite cold winter temperatures (

and provides a suitable environment (e.g., neutral pH andmicrobiologically mature) to support a high level of bio-genic activity, into a landfill cover system. Two approacheshave been used in field-scale trials. In the biocover ap-proach, the methane oxidation layer was applied to replacethe intermediate and final landfill covers. In the biofilter ap-proach, the LFG was either trapped by a liner or accumu-lated in a collection system, and then passed (passively oractively) through the biofilter medium.

Rajbhandari et al. (2006) used a thin biocover (0.3 to0.4 m) as the temporary cover for 5 m waste lifts. A mixtureof compost and tree mulch (9:1, wet weight) was used forthe second waste lift, and was found to emit low methanefluxes (

cover at the Leduc and District Regional Landfill (Alberta).Two biofilters were constructed with a SAR of 10.8, withone constructed over a gas well, while a third site was builtwith a biofilter SAR of 4.8. It was anticipated that the use ofthe gas well would result in larger LFG influent fluxes, andtherefore the effect of this on the performance of the biofil-ter could be examined. The biofilter surface area was in-creased to lower the SAR from 10.8 to 4.8, which allowedfor an assessment of lowering the anticipated influentLFG fluxes on the performance of the biofilter. The per-formance objectives were to achieve low surface emissions(

walls of a wooden frame that contained the biofilter. Com-post was selected as the biofilter medium, and was placedin the wooden frame. The compost, of yard-waste origin,was taken from an open windrow operation at the LeducLandfill. The compost was turned twice per month for sixmonths and then was left to cure for one year. The compostwas passed through a 12.7 mm screen.

Analytical methodsThe yard-waste compost was analyzed following the

methods outlined by the Test Methods for the Evaluation ofCompost and Composting (TMECC). The bulk density(BD), MC, organic matter (OM), pH, electrical conductivity(EC), and maturity were tested using TMECC methods03.01-A, 03.09-A, 05.07-A, 04.11-A, 04.10-A, and 05.08-B,respectively (TMECC 2002). The total air space (TAS) re-fers to the total amount of air volume in the medium, andincludes both the inter and intra particle pore air volumes.The TAS and porosity were calculated as described by Balland Smith (2001). This included determining the particledensity of the compost using a boiling method (Das 2002).Based on the particle density, MC, and BD, the TAS andporosity could be determined.

The total carbon and nitrogen (CN) analysis was con-ducted with a Leco1 TrueSpec CN Carbon/Nitrogen Deter-minator (Leco Co., St. Joseph, Mich.). Before the analysis,samples (10 g) were air dried for 24 h at 36 8C, and werethen passed through a 1 mm screen.

The pilot biofilters were each equipped with monitoringinstruments at different depths (0.20, 0.55, 0.90, 1.25,and 1.60 m), and quadrants (AD). Figure 2 shows theplacement of the polyethylene gas collection tubing, thermo-couple sensors (type K), and time domain reflectometry(TDR) probes. The gas collection tubing (6.4 mm ID) wasconnected to a perforated PVC end cap (12.7 mm ID and100 mm length). The end cap was placed in the biofilterand was used as a filter to prevent the tubing from gettingclogged with soil particles when gas samples were collected.A Landtec GEM1 2000 (Colton, Calif.) gas analyzer wasused to measure the CH4, CO2, and O2 concentrations at thedifferent depths and quadrants shown in Fig. 2. The balanceof the sum of those gases was considered to be the nitrogenconcentration. Additional thermocouple sensors were placedin the shade at each site to measure ambient temperature, aswell at the surface (0.05 m depth) of each biofilter. Allsites contained temperature data loggers for continuoushourly measurements.

Moisture Point1 (ESI Environmental Sensors Inc., Victo-ria, B.C.) TDR probes were used to measure volumetricMC. Site 2 was equipped with a data logger for daily mois-ture measurements, while manual measurements were re-quired at sites 1 and 3. The placement of the TDR probes,shown in Fig. 2, was for sites 1 and 2 only. Each TDRprobe is composed of several segments, in which volumetricMC is measured over the respective segment length. LongerTDR probes (5 segments, 1.20 m total sensor length)were placed in the center of the filter bed at the 0.20and 1.25 m depths. Shorter TDR probes (4 segments,0.60 m total sensor length) were placed between thequadrants AB and CD at the 0.55 and 0.90 m depths.For site 3, shorter TDR probes were placed vertically in-between each quadrant (AB, BC, CD, and AD), such thatmeasurements were recorded in 0.15 m increments to adepth of 0.60 m. The TDR system was calibrated for de-termining the actual volumetric MC of the compost. Thiswas conducted by filling a column with compost at fivedifferent moisture levels (0.090.36 g g1, wet basis). Ateach respective moisture level, the volumetric MC was de-termined with a TDR probe. In addition, the column wouldbe compacted (by dropping the column from 0.2 m) typi-cally three times at each gravimetric moisture level, withrespective volumetric MC measurements with the TDRprobe at each different BD. The TDR results were thencompared with calculated volumetric MC values, deter-mined based on the measured gravimetric MC and bulkdensities used in the experiment. After repeating the testat different moisture levels and bulk densities, the follow-ing linear relationship was observed to predict the volumet-ric MC of compost based on the TDR measurement:

3 MCv 1:13MCTDR 0:078 R2 0:95

where MCv (L L1) is the predicted volumetric MC, andMCTDR (L L1) is the averaged result from the TDR probe.

Surface emissions (CH4 and CO2) were measured by us-ing static flux chambers. Frames (0.38 m2) were buried atleast 0.05 m into the surface of the four quadrants of eachbiofilter medium. The chambers could then be placed ontop of the frames and were sealed with water or an anti-freeze mixture during colder periods. The combined volumeof the frame (not buried) and chamber is 0.13 m3. Eachchamber contained a fan to mix the accumulating gas in or-der that a representative gas sample could be collected. Avalve on the chamber exterior was equipped with a needleto collect gas samples. Most chambers also contained a ther-mocouple sensor to measure temperature. Gas samples werecollected in 7 mL Vacutainers1 serum tubes (Becton Dick-inson, Franklin Lakes, N.J.), equipped with rubber septumlids, every 10 min in duplicate for 60 min. The CH4 (andsimilarly the CO2) flux could then be determined

4 JCH4 C V

t A

where JCH4 is the effluent flux (g CH4 m2 d1), DC/Dt is theslope of gas concentration versus time curve (m3 CH4 m3air d1), is the density of the gas determined from theideal gas law (g m3), V is the combined above ground vo-lume (m3) of the chamber and frame, and A is the surface

Table 1. Pilot biofilter design properties.

Properties Site 1Site 2 (gaswell)

Site 3 (largebiofilter)

Location on thelandfill body

Slope Top Top

Surface area ratio 10.8 10.8 4.8Surface area of MSW

covered (m2)100 100 100

Biofilter surfacearea (m2)

9.3 9.3 20.9

Biofilter depth (m) 1.5 1.5 1.5Gas well (yes/no) No Yes No

534 J. Environ. Eng. Sci. Vol. 7, 2008

# 2008 NRC Canada

area (m2) of the filter covered by the frame. The slope wasdetermined using linear regression, and analyzed for signifi-cance (usually a p-value < 5%) using analysis of variance(ANOVA). The flux was equal to zero when no significantrelationship was observed. The atmospheric pressure valueswere taken from an Environment Canadas weather station(Environment Canada 2006), located at the Edmonton Inter-national Airport, which is approximately 30 km northwestfrom the landfill site.

Monitoring events were conducted once per month tomanually measure gas composition, moisture profiles (sites1 and 2), and surface emissions. Gas samples collectedfrom the flux chambers were transported to the Alberta Re-search Council (Edmonton, Alberta) and were analyzed,usually within 24 h, with a Varian CP-49001 (Palo Alto,Calif.) micro gas chromatography (GC) instrument,equipped with a thermal conductivity detector. A Molsieve5A column (90 8C, 200 kPa, 60 s run time) was used to an-alyze oxygen, nitrogen, and methane concentrations, while aPora PLOT Q column (65 8C, 200 kPa, 60 s run time) wasused to measure carbon dioxide concentrations (lower detec-tion limit for all gases was 10 ppm). The GC was calibratedby purchasing standard gas mixtures from Praxair Inc. (Ed-monton, Alberta). High purity helium (0.99999 L L1 He)was used as the carrier gas. CP-Maitre Elite software wasused to operate the GC.

Standard error (SE) was used to estimate the variability ofthe observed results. Standard error was determined by thefollowing equation:

5 SE SSqrt N 1

where S is the standard deviation and N represents the num-ber of samples.

Methane removal ratesA methane removal rate determination procedure was de-

veloped to assist in the assessment of the performance of thepilot biofilters. Since the incoming LFG flux into the pilotbiofilters was unknown, this proved to be challenging. Con-sideration was given to an approach being developed basedon the isotopic composition of LFG and the preferential up-take of 12C by methane oxidizing bacteria (Chanton et al.1999; De Visscher et al. 2004). However, budget constraintsprevented applying the isotopic method for the current in-vestigation.

To determine a rough assessment of the methane removalrates, an approach used by Zeiss (2002) was incorporated.The influent LFG flux was assumed to equal the effluentLFG flux (i.e., CH4 and CO2), since theoretically every unitof volume of CH4 that is oxidized produces an equal volumeof CO2 (see eq. [1]). The following equation was used tocalculate the CH4 influent flux:

6Jin;CH4 Cin;CH4 Jout;CH4 Jout;CO2

where Jin, CH4 is the influent methane flux (L CH4 m2 d1),Jout, CH4 and Jout, CO2 are the respective effluent methane andcarbon dioxide fluxes (L m2 d1), and Cin, CH4 is the con-centration of methane (L L1) in the tire shreds (1.60 mdepth). In practice, the effluent CO2 will be affected by theamounts of carbon assimilated into biomass and producedfrom the compost medium. The latter was calculated to be39 g CO2 m2 d1 based on the values provided in Tables 1and 2 (at 25 8C assuming the top 0.20 m of the biofiltermedium was aerobic). However, the amount of CO2 pro-duced aerobically from the compost will be less than thematurity value (1.02 mg C-CO2 g1 OM d1) shown in Ta-ble 2, as a result of biogenic competition for oxygen withthe methanotrophs.

Using eq. [6], the methane removal rate was determinedby the following equation:

7 CH4 removed Jin;CH4 Jout;CH4

Jin;CH4 100

where CH4 removed (%) is the percentage of methane re-moved in the biofilter. The use of eq. [7] sometimes yieldednegative results, since the influent flux calculated witheq. [6] resulted in lower values than the observed effluentmethane flux. In these instances, the methane removal ratewas assigned a value of zero.

Results

Several physical, chemical, and biological properties ofthe yard-waste compost are shown in Table 2. The compostwas a porous medium, that provided the methanotrophicbacteria the ideal environmental conditions for growth; themedium contained a low salt content, had a neutral pH, andwas biologically mature. The latter is of importance as it isundesirable to have a compost medium that is actively de-

Fig. 2. Instrumentation placement (not to scale).

Philopoulos et al. 535

# 2008 NRC Canada

composing. This would result in competing microbiologicalactivity with the methanotrophic bacteria.

Figures 3a and 3b show the average CH4 and CO2 surfaceemissions, measured from eight monitoring events con-ducted from the fall of 2005 to the spring of 2006. Sites 1and 2 were operational in August (2005), while site 3 wasoperational in November (2005). Generally low CH4 surface

emissions (15 gCH4 m2 d1) event was observed forsite 1 on March 21st.

One monitoring event showed a negative methane flux.This was observed for site 1 (1.17 g CH4 m2 d1) on Feb-ruary 28th. Barlaz et al. (2004) observed negative fluxes in a

Fig. 4. Relationship between changing atmospheric pressure andsurface emissions. (a) CH4 emissions. (b) CO2 emissions.

Fig. 3. CH4 (Top) and CO2 (Bottom) emissions (g m2 d1). Note:SE bars are shown; some SE may be too small to be visible.

Table 2. Yard-waste compost properties.

Property CompostStandarderror

Moisture content (g g1, wet basis) 0.31 0.001Bulk density (g L1, wet basis) 772.5 6.6Porosity (L L1) 0.69 Total air space (L L1) 0.49 Organic matter (g g1, dry basis) 0.18 0.003Total carbon (mg g1, dry basis) 81.01 0.90Total nitrogen (mg g1, dry basis) 8.50 0.14Carbon to nitrogen ratio 9.53 pH 7.49 0.02Conductivity at 25 8C (dS m1) 2.75 0.03Maturity (mg C-CO2 g1 OM d1) 1.02 0.04

Note: OM, organic matter.

536 J. Environ. Eng. Sci. Vol. 7, 2008

# 2008 NRC Canada

biocover for 59 of 107 flux chamber measurements, using ayard-waste compost as the medium. They attributed this tothe biocover uptake of atmospheric methane, which was mostlikely the reason for the negative flux observed at site 1. LowCO2 surface emissions (52.51 g CO2 m2 d1) were ob-served at site 1 on February 28th, which likely indicatedlow influent LFG flows. This meant that the methanotro-phic bacteria at the surface of the biofilter most likely hadthe capacity for uptake of atmospheric methane.

Figure 5 shows the methane removal rates, as determinedby eqs. [6] and [7], for all sites. The average removal forsites 1, 2, and 3 were 76, 68, and 35%, respectively. Usingeq. [6], the average influent methane fluxes were calculated(converting to a gravimetric basis using the ideal gas law) tobe 37.4, 53.5, and 1.2 g CH4 m2 d1 for sites 1, 2, and 3,respectively.

Figures 6a6f show the average gas composition profilesfor several selected days, for sites 1 and 2. Overall, the re-sults show the variability in LFG and atmospheric air move-ment through the biofilter. On October 25th, Figs. 6a and6b, high CH4 concentrations were measured throughoutboth biofilters. Low oxygen concentrations (

Fig. 6. Gas composition profiles. (a) Site 1, 25 Oct. 05. (b) Site 2 (gas well), 25 Oct. 05. (c) Site 1, 28 Nov. 05. (d) Site 2 (gas well), 28Nov. 05. (e) Site 1, 1 May 06. (f) Site 2 (gas well), 1 May 06. (g) Site 3 (large biofilter), 28 Nov. 06. Note: SE bars are shown; some SEmay be too small to be visible.

538 J. Environ. Eng. Sci. Vol. 7, 2008

# 2008 NRC Canada

showed warmer temperatures in the deeper layers of thebiofilter medium, with a maximum value observed atthe 1.25 m depth. As mentioned in the introduction, oneobjective was to maintain temperatures above 20 8C. Ta-ble 3 also shows the percentage of days in which the aver-age temperature was above 20 8C, for sites 1 and 2. Site 2was clearly achieving the objective more frequently thansite 1. The minimum, maximum, and range values describethe temperature variations observed, and are also shown inTable 3. The range decreases with increasing depth at bothsites. Site 3 (results not shown) showed colder tempera-tures than sites 1 and 2, and never reached the 20 8C ob-jective.

Figure 7 shows the change in temperature for the 0.20and 1.60 m depths for each biofilter over time, as well asthe average ambient temperature measured at all three sites.The results show the temperature at the 0.20 m depth, forsites 1 and 2, following the general ambient trend. Sites 1(9.4 8C) and 2 (11.1 8C) showed minimum temperatures onMarch 25th. At site 3, colder temperatures were observed atthe 0.20 m depth, which remained constant before increas-ing at the end of March. Consistent with the results shown

in Table 3, the temperatures at the 1.60 m depth for allsites did not fluctuate as much as the those at the 0.2 mdepth.

Figures 8a8c show the calculated volumetric MC pro-files (eq. [3]), based on the TDR measurements, for all threesites on several selected days. Sites 1 and 2 (Figs. 8a and8b) were generally meeting the 0.25 L L1 objective. In par-ticular, site 2 showed a higher MC at all depths. The mois-ture objective was met on November 25th at site 3 (Fig. 8c),however, the most recent data (March 21st) showed drierconditions.

Discussion

Site 3, which was designed with a smaller biofilter SAR(4.79), showed smaller LFG concentrations (Fig. 6f) andCH4 and CO2 surface emissions (Figs. 3a and 3b) than sites 1and 2. However, the CO2 surface emissions showed an in-creasing trend, with an effluent flux of 42.6 g CO2 m2 d1observed on May 31st. Temperatures were never observedto rise over 20 8C, a further indication of lower LFG influ-ent flows and methane oxidation. There were two causes

Fig. 7. Temperature versus time for selected depths.

Table 3. Sites 1 and 2 average daily temperature statistics.

Site 1 (8C) Site 2 (8C)

Depth (m) Min Max Avg Range % Days >20 8C Min Max Avg Range % Days >20 8CAmbient 24.3 25.2 0.8 49.5 4.0% 24.5 22.7 2.4 47.2 1.4%0.20 9.4 46.6 27.6 37.2 75.3% 11.1 43.1 26.5 32.0 78.7%0.55 11.7 35.7 25.6 24.0 82.8% 16.0 39.2 27.6 23.2 87.2%0.90 11.1 31.8 23.1 20.8 62.0% 20.9 38.2 29.8 17.3 100%1.25 11.7 30.6 22.4 18.9 61.6% 23.5 38.6 31.3 15.1 100%1.60 (Tire

Shreds)10.4 29.0 20.6 18.6 53.8% 29.0 39.5 34.5 10.5 100%

Philopoulos et al. 539

# 2008 NRC Canada

identified that could possibly explain the inactivity in the bi-ofilter medium. The first was that the location may not havebeen a hot spot for landfill gas generation. No surface emis-sion measurements were conducted prior to the constructionof the site. Any gas generated below the site may havemoved laterally, as the site was built on top of the landfill(20 m) beside a slope. The use of a gas well, similar tosite 2, may have resulted in larger LFG influent flows. Thesecond potential cause was that as a result of the large aver-age concentrations of oxygen (0.14 L L1) observed in thetire shreds (1.60 m), there was the possibility that methaneoxidation was occurring in the waste below the tire shreds.This is similar to what Humer and Lechner (2001) observedin the winter months during the operation of a biocover.They found that the majority of the methane had been oxi-

dized before the interface between the distribution layer(coarse gravel) and a sewage sludge medium. However, it isunclear if they were suggesting that methane was being oxi-dized in the gas distribution layer or in the waste below (orboth).

Sites 1 and 2 showed low (

on that day. Therefore, dropping atmospheric pressure andlow temperatures at the 0.20 m depth were identified asthe respective contributing factors for the low methane re-moval rates observed on October 25th and March 21stmonitoring events.

The large drop in atmospheric pressure was attributed forcausing the high surface emissions on October 25th. Othershave also observed this relationship. Poulsen et al. (2003)showed that the rate of atmospheric pressure change wasthe controlling factor in short-term (24 h) surface emissions.Long-term (1 year) surface emissions were showed to bemost dependent on the rate of atmospheric pressure change,soil moisture content, and gas permeability. Gebert andGroengroeft (2006) showed a similar relationship; as therate of atmospheric pressure change decreased, the rate ofLFG pressure increased in the biofilter supply pipe (whichwas connected to a collection system). They suggested therate of advective LFG flow through a collection system willbe dependent on the degree the landfill cover seals the land-fill body. The more permeable a landfill cover soil is, themore paths there are for LFG to migrate out of the landfillbody. The lower the permeability of the landfill cover soil,the less paths there are for LFG to migrate out of the landfillbody, and the more likely it will move to a gas collectionsystem, or through a porous biofilter integrated into thelandfill cover as in the current study.

The Leduc and District Regional Landfill cannot be con-sidered a well sealed landfill. The pilot biofilters are locatedin an unlined portion of the Leduc and District RegionalLandfill. In addition, sites 1 and 3 are backfilled with an in-termediate clay cover (no assurance of optimum moisturecontent, compaction, or particle size distribution), whilesite 1 was backfilled with a non-clay soil. The permeabilityof the cover soil surrounding the pilot biofilters may havevaried throughout the experiment. For example, there wasmore precipitation observed in September and October(32.8 and 23.8 mm, respectively) than in November(7.0 mm) of 2005. This may have been a factor in explain-ing why despite a drop in atmospheric pressure of 0.7 hPa h1during the November 28th monitoring event at site 1,which was the same drop observed at site 2 during the Oc-tober 25th monitoring event, no surface emissions were ob-served (i.e., with a more permeable cover soil, surfaceemissions through the biofilter are less dependent onchanging atmospheric pressure). It is important to high-light, the addition of a low permeable cover system (i.e.,compacted clay cover, with subsoil and topsoil) surround-

ing the pilot biofilters would likely result in an increase inthe influent LFG fluxes into the biofilters.

Generally, oxygen was only found at the 0.20 m depthfor sites 1 and 2, and on average was less than 5% (vol. ba-sis). This indicated that the methane oxidation horizon (thearea where the methanotrophic bacteria were most active)was in the top 0.200.55 m as indicated by the consumptionof oxygen. Humer and Lechner (2001) found O2 penetratingto 0.500.90 m depths in a MSW compost used as a bio-cover medium. During the first monitoring event, wheredropping atmospheric pressure was attributed for the highsurface emissions observed, O2 diffusion may have beenlimited by the advective LFG flow. Gebert and Grongroft(2006) found this to be a limiting factor in the removal ca-pabilities of their biofilter. Changes in TAS will also impactO2 penetration. This may explain why on November 28ththe largest O2 concentration (0.12 L L1) was observed atsite 2 (Fig. 6d), as the MC (0.23 L L1) was lower (Fig. 8b)than observed on other monitoring events.

The biofilter temperature depends on several factors, themajor ones being ambient temperature, heat released frommethane oxidation, the quantity of influent LFG, and thethermal characteristics of the medium. The objective tomaintain average temperatures above 20 8C was achieved atsites 1 and 2 at all depths. However, site 2 met the objectivemore frequently (Table 3). Site 2 showed larger LFG influ-ent flows and concentrations at the 1.60 m depth (tireshreds), most likely explaining the warmer conditions ob-served. The use of the gas well at site 2 may be the reasonfor the larger LFG influent flows and concentrations ob-served. However, it is also possible the site was generatingmore LFG (i.e., hot spot) from the waste below it thansite 1.

The MC profiles (Figs. 8a8b) show that sites 1 and 2were generally meeting the moisture objective of 0.25 L L1.However, site 2 showed larger moisture levels throughoutthe biofilter medium. Site 2 had a berm built around it,consisting of stockpiled cover soil (not by design), whichmay have prevented some wind desiccation. The higherLFG influent flows and concentrations in the tire shredsobserved at site 2 may have resulted in more condensation(as LFG is typically saturated with moisture). In addition,the higher average methane influent fluxes at site 2 couldhave resulted in more oxidation and water production.

ConclusionThe approach to integrate a biofilter into the landfill

Table 4. Comparison of results for sites 1 and 2 for several selected days.

25 Oct 21 Mar 1 May

Results Site 1 Site 2 Site 1 Site 2 Site 1 Site 2CH4 emissions (g CH4 m2 d1) 148.2 199.6 57.7 14.7 1.4 11.6CO2 emissions (g CO2 m2 d1) 562.3 782.5 102.4 28.4 93.6 121.4Methane removal (%) 24% 29% 0% 0% 96% 94%D Atm. pres. (hPa h1) 1.1 0.7 0 0 0.2 0.2Temp. (8C) at 0.20 m 36.4 31.7 12.3 12.5 36.6 35.1Moisture (L L1) at the 0.20 m N/A 0.37 0.29 0.55a 0.24b 0.49

aObserved on 20 Mar.bObserved on 24 Apr.

Philopoulos et al. 541

# 2008 NRC Canada

cover showed promising results, as low surface emissions(