Journal of Contaminant Hydrology 107 (2009) 101–107

Contents lists available at ScienceDirect

Journal of Contaminant Hydrology

j ourna l homepage: www.e lsev ie r.com/ locate / jconhyd

Gas phase dispersion in compost as a function of differentwater contents and air flow rates

Prabhakar Sharma⁎, Tjalfe G. PoulsenDepartment of Biotechnology, Chemistry, and Environmental Engineering, Section for Environmental Engineering, Aalborg University,Sohngaardsholmsvej 57, DK-9000, Aalborg, Denmark

a r t i c l e i n f o

⁎ Corresponding author. Tel.: +45 9940 9936; fax:E-mail address: ps@bio.aau.dk (P. Sharma).

0169-7722/$ – see front matter © 2009 Elsevier B.V.doi:10.1016/j.jconhyd.2009.04.005

a b s t r a c t

Article history:Received 29 January 2009Received in revised form 7 April 2009Accepted 11 April 2009Available online 18 April 2009

Gas phase dispersion in a natural porous medium (yard waste compost) was investigated as afunction of gas flow velocity and compost volumetric water content using oxygen and nitrogenas tracer gases. The compost was chosen because it has a very wide water content range andbecause it represents a wide range of porous media, including soils and biofilter media. Columnbreakthrough curves for oxygen and nitrogen were measured at relatively low pore gasvelocities, corresponding to those observed in for instance soil vapor extraction systems orbiofilters for air cleaning at biogas plants or composting facilities. Total gas mechanicaldispersion–molecular diffusion coefficients were fitted from the breakthrough curves using aone-dimensional numerical solution to the advection–dispersion equation and used todetermine gas dispersivities at different volumetric gas contents. The results showed that gasmechanical dispersion dominated over molecular diffusion with mechanical dispersion for allwater contents and pore gas velocities investigated. Importance of mechanical dispersionincreased with increasing pore gas velocity and compost water content. The results furthershowed that gas dispersivity was relatively constant at high values of compost gas-filledporosity but increased with decreasing gas-filled porosity at lower values of gas-filled porosity.Results finally showed that measurement uncertainty in gas dispersivity is generally highest atlow values of pore gas velocity.

© 2009 Elsevier B.V. All rights reserved.

Keywords:Gas dispersionWater contentPore gas velocityGas-filled porosityCompost

1. Introduction

Understanding gas phase transport in natural porousmedia is important when assessing movement of gaseousphase compounds, for instance in connectionwith removal ofvolatile organic contaminants by soil vapor extraction,subsurface migration of toxic gases from old landfill sites,intrusion of gaseous contaminants from soil into buildings,design of bio-covers for methane oxidation at landfills,estimation of methane gas emission from wetlands andfrom thawing permafrozen areas, aeration of the soil rootzone, cleaning of air containing odorous gases using biofilters,and migration of oxygen in compost piles during composting(El-Fadel et al., 1997; De Visscher et al., 1999; Liang et al.,2000; Poulsen et al., 2001).

+45 9635 0558.

All rights reserved.

Advection, molecular diffusion and mechanical dispersionare the three most important physical mechanisms governinggas flow in porous media. Advection is the movement of thegas phase caused by gas pressure gradients in the porousmedium, molecular diffusion is the migration of individualcompounds in the gas phase due to concentration gradients inthe medium and mechanical dispersion is spreading of gasphase compounds due to spatial variations in advectivevelocity and differences in the distances traveled caused bythe pore system tortuosity. While porous media gas phaseadvection and molecular diffusion, including parameterscontrolling these processes in natural porous media havebeen extensively described in both soil science and chemicalengineering literature (Delgado, 2006), much less is knownabout the parameters controlling mechanical dispersion inthe gas phase of natural porous media. Porous media physicalproperties controlling gas phase advection velocity andmolecular diffusion coefficients in natural porous materials

102 P. Sharma, T.G. Poulsen / Journal of Contaminant Hydrology 107 (2009) 101–107

are therefore well known, whereas this is not the case for gasphase mechanical dispersion. Mechanical dispersion innatural porous media has been widely researched foraqueous-phase transport in saturated and to some degreealso in unsaturated porous media (Delgado, 2006). Theresults indicate that pore water velocity, micro-porosity, andsolute molecule size are the main parameters controllingaqueous-phase mechanical dispersion (Rose, 1973; Rao et al.,1980; Roberts et al., 1987; Brusseau, 1993). In case of gasphase mechanical dispersion, most investigations have beenconducted using artificial porous media such as glass beads,Raschig rings or similar (Carberry and Bretton, 1958; DeMariaand White, 1960; Edwards and Richardson, 1968; Rolstonet al., 1969; Scott et al., 1974; Ahn et al., 1986; Batterman et al.,1995; Yu et al., 1999; Ruiz et al., 1999; Popovicova andBrusseau, 1997; Delgado, 2006). Only a few studies of gasphase mechanical dispersion have been carried out usingnatural materials (Blackwell et al., 1959; Edwards andRichardson, 1968; Rolston et al., 1969; Yu et al., 1999;Costanza-Robinson and Brusseau, 2002; Gidda et al., 2006;Poulsen et al., 2008).

In general observations, the gas phase mechanical disper-sion in natural porous materials increases with gas flowvelocity (Rolston et al., 1969; Gidda et al., 2006; Poulsen et al.,2008), which has also been extensively documented foraqueous-phase mechanical dispersion. A few recent studiesusing relatively fine grained materials (clay and sand) havealso shown that mechanical gas dispersion coefficients tendsto increasewith increasingwater content (Costanza-Robinsonand Brusseau, 2002; Gidda et al., 2006; Poulsen et al., 2008).Gidda et al. (2006) concluded that increased heterogeneity inthe gas-filled pore network distribution due to increasedwater content caused the increase in gas dispersivity. All ofthese studies were carried out for relatively fine, homoge-neous materials ranging from clay to medium sand and it istherefore difficult to assess if the results are applicable tocoarser materials such as those of biofilters, bio-covers orcompost piles that are much more heterogeneous and havemuch larger particle sizes. The measurements were furthercarried out only for a limited number of water contents or gasflow velocities for the same medium. Gidda et al. (2006) andPoulsen et al. (2008) presented breakthrough data for twowater contents for a given material at several flow velocities,whereas Costanza-Robinson and Brusseau, (2002) presenteddata for several water contents but only one gas velocity.

Due to the very limited amount of data available, it is alsodifficult to assess the general shape of the relationshipbetween mechanical dispersion and porous media water con-tent and it is thereforenot possible to develop relationships fordirectly predicting the mechanical dispersion coefficient innatural porous materials from gas velocity and porous mediaphysical characteristics such as is the case for the moleculardiffusion coefficient (Moldrup et al., 2005).

At low gas velocities, gas migration caused by moleculardiffusion is more important than effects of mechanical disper-sion; however, as gas phase advective velocity increases,mechanical dispersion can become more important thanmolecular diffusion. A very limited number of studies on therelative importance of mechanical dispersion are available.Edwards and Richardson (1968) found for argon transport indry fine sand (particle diameter 0.016 cm) that mechanical

dispersion coefficients were larger than molecular diffusioncoefficients at pore gas velocities above 400 cm min−1,Costanza-Robinson and Brusseau (2002) found for methanetransport in sand at variable water contents that moleculardiffusion was most important below water contents ofapproximately 0.15 cm3 cm−3 at a pore gas velocity ofapproximately 10 cm min−1. These results indicate thatmolecular diffusion is most important at low water contentsand low advective gas velocities and that it is likely thecombination of water content and gas velocity that controlsthe relative importance ofmolecular diffusion versusmechan-ical dispersion. Due to the very limited quantity of data, it is,however, not possible to make any general conclusionswith respect to the relationship between the mechanicaldispersion–molecular diffusion ratio, gas velocity and watercontent.

The purpose of this paper is therefore to investigate therelationship between the gas phase mechanical dispersioncoefficient, gas velocity, and porous mediumwater content ina porous medium with properties relevant for natural soilsystems, bio-covers and biofilters. Gas phase dispersion wasinvestigated at relatively low gas flow velocities relevant formost natural soils, bio-covers and to some degree alsobiofilters.

Yard waste compost with relatively high inorganic mattercontent was used as test medium. This compost has a wideparticle size distribution, contains aggregates and thereforerepresents a wide range of porous media where gas transportis important. The physical properties of the compost alsoresemble those of many natural porous media includingaggregated media as well as constructed media such asbiofilters and bio-covers. The compost is also capable ofachieving a wide range of water contents covering the rangesfound in most natural porous media including most soils.Atmospheric air and nitrogen were used in transport experi-ments as tracer gases as they are inexpensive, easy and safe touse and relatively easy to detect. Results are used to evaluatethe relation between gas phase mechanical dispersion coef-ficient, pore gas velocity, and compost water content. Also therelative importance of mechanical dispersion in comparisonwith molecular diffusion in the gas phase was evaluated as afunction of gas phase advective velocity and compost watercontent.

2. Theory

Transport of a gaseous tracer in porous media is tradi-tionally described using the advection–dispersion differentialequation for one-dimensional transport of a non-conservativetracer that may also partition into the solid and liquid phasesof the medium under assumption of local equilibrium as:

ACAt

=Dtot

RA2C

Ax2+

uRACAx

− rρb

eð1Þ

where C is the concentration of the tracer in the gas phase(M L−3), t is time (T), Dtot is the overall dispersion coefficient(L2 T−1), R is the retardation factor, u is the pore gas velocity(L T−1), r is the tracer consumption or production rate inthe porous medium (M M−1 T−1), ρb is the dry bulk density(M L−3) and ε is the gas-filled porosity of the porous medium

Table 1Experimental conditions and properties of compost for oxygen uptakeexperiments.

Bulk density,ρb (g cm−3)

Water content,ω a (g H2O (g dm)−1)

Total porosity,ϕ (cm3 cm−3)

Gas-filled porosity,ε b (cm3 cm−3)

0.87 0.037 0.60 0.570.87 0.248 0.60 0.380.87 0.308 0.60 0.330.87 0.354 0.60 0.290.87 0.451 0.60 0.210.81 0.529 0.63 0.200.78 0.569 0.64 0.190.78 0.608 0.64 0.160.78 0.647 0.64 0.13

a Gravimetric water content of the compost was calculated based on totaldry weight of compost.

b Calculated by subtracting the volumetric water content with totalporosity.

103P. Sharma, T.G. Poulsen / Journal of Contaminant Hydrology 107 (2009) 101–107

(L3 L−3). The overall dispersion coefficient is the sum of thecontributions frommolecular diffusion (Ddiff) andmechanicaldispersion (Dmech) coefficients.

Dtot = Ddiff + Dmech: ð2Þ

The molecular diffusion coefficient in sieved and repackedporous media can be accurately estimated using the relation-ship of Moldrup et al. (2005).

Ddiff = D0e3−/

/ð3Þ

where D0 is the gas diffusion coefficient in free air (L2 T−1),and ϕ is the total porosity of the medium (L3 L−3). Themechanical dispersion coefficient can be estimated as:

Dmech = α u ð4Þ

where α is the coefficient of dispersivity (L).

3. Materials and methods

3.1. Experimental approach

Yard waste compost was collected from a municipalcomposting facility at Aalborg municipality of northernDenmark. The compost had a moisture content of 0.47 gH2O (g dry compost)−1 and an organic matter content of0.187 g organic matter (g dry matter)−1. The specific particledensity of the compost was calculated to 2.16 g cm−3

assuming a particle density for organic and inorganic matterof 1.2 and 2.65 g cm−3 respectively. The compost wasscreened to a maximum particle size of 5 mm prior to use.

Compost samples of approximately 30 kg were wetted towater contents of 0.037, 0.25, 0.28, 0.35, 0.45, 0.53, 0.57, 0.61and 0.65 g H2O (g dry compost)−1, mixed well and placed inair-tight containers to allow redistribution of the water for atleast 24 h. The compost was then packed into 103-cm longand 14-cm inner diameter clear acrylic columns in 10 equalportions to achieve a homogeneous dry bulk density. Due tothe difference in water content between samples it was notpossible to achieve exactly the same dry bulk density forall columns and bulk densities ranging between 0.78 and0.87 g cm−3 were used (Table 1). The columnswere equippedwith air-tight polyethylene lids at both ends. The inlet end ofeach column was connected to an air/nitrogen supply via athree-way valve and a precision flow meter (model F150,Porter Instruments, Inc., Hatfield, PA) to control gas flow rates.The outlet lid was equipped with an oxygen sensor (KE-12oxygen electrode, GS Yuasa Power Supply Ltd., Japan) fordetermination of effluent oxygen concentrations. Readingsfrom the oxygen sensor were recorded by a data logger (CR-1000, Campbell Scientific, Logan, UT). More details of theexperimental set-up can be found in Poulsen et al. (2008) andSharma et al. (in press). All columns were prepared induplicate. An overview of the column sample properties isgiven in Table 1.

Columns were initially saturated with atmospheric air andthen flushed with N2 at a selected flow rate for a time periodranging from 30 min to 2 h after which they were flushedwith atmospheric air for the same amount of time. Experi-

ments were repeated for each column at flow rates of 0.2, 0.5,1.0, 1.5, and 2.0 L min−1.

3.2. Data analysis

The advection–dispersion equation (Eq. (1)) was fitted tothe measured oxygen and nitrogen breakthrough curves toobtain values for total dispersion coefficient, Dtot, andretardation factor, R at all water contents and gas transportvelocities (a total of 180 breakthrough curves). Eq. (1) wassolved using an explicit finite difference scheme corrected fornumerical dispersion. Values for r were determined from theoxygen uptake experiments by subtracting the outlet steadystate oxygen concentration from inlet concentration (Sharmaet al., in press). The r values depend on the oxygen con-sumption by compost medium as a function of compost watercontent and pore gas velocity through the column. Values foru, ρb and εwere those corresponding to the different columns(Table 1). Best fit values for Dtot and R were obtained byminimizing the sum of squared errors between effluentoxygen concentrations predicted by the model and the con-centrations measured by the oxygen electrode for all break-through curves. Values of molecular diffusion coefficient,Ddiff, for oxygenwere estimated for each column using Eq. (3)and a D0-value for oxygen of 0.219 cm2 s−1. Mechanicaldispersion coefficients, Dmech, were subsequently calculatedusing Eq. (2).

4. Results and discussion

Fig. 1 shows an example of breakthrough curves (relativeoxygen concentration as a function of time) for nitrogendisplacing air at a flow rate of 1 L min−1 at selected watercontents. The corresponding model predictions of oxygenconcentrations are also shown in Fig. 1. Model predictionsgenerally fitted the measured breakthrough data well. Thiswas also the case for the remaining data, indicating that thequantity of inactive gas-filled pores in the compost waslimited and that the assumption of local equilibrium gastransport was adequate.

Resulting values of Dtot as a function of pore gas velocityfor the 9 water contents considered are shown in Fig. 2a andratios of Dtot and its maximum value Dtot,max as a function of

Fig.1. Example of column oxygen breakthrough curves for the gas flow rate of1 L min−1 and for selected water contents for elution of air with nitrogen.

104 P. Sharma, T.G. Poulsen / Journal of Contaminant Hydrology 107 (2009) 101–107

pore gas velocity and volumetric water content are shown inFig. 2b. For all water contents Dtot increases with pore gasvelocity as expected and the slope of the relationship appearsto be highest for the highest water contents (Fig. 2a). Fig. 2bshows that although the relationship between Dtot/Dtot,max

and u is not a straight line in all cases as would be expectedbased on Eqs. (2) and (4), there is no consistent tendency for aspecific curve shape and the data generally confirms that alinear relationship can be assumed with good approximation.The main reason for the deviations from linearity is that

Fig. 2. (a) Total gas dispersion–diffusion coefficients (Dtot) and (b) normal-ized total gas dispersion–diffusion coefficients (Dtot/Dtot,max); as a function opore air velocity for different air-filled porosities.

Fig. 3. (a) Contribution of diffusion (Ddiff) to total dispersion (Dtot), Ddiff/Dtot,as a function of pore air velocity (u) for different gas-filled porosity (ε);(b) Ddiff/Dtot as a function of gas-filled porosity for different pore airvelocities; and (c) relative effect of gas-filled porosity and porewater velocityon mechanical diffusion and total dispersion (Dmech/Dtot). Circled data pointwas not included in the best fit line at (Dmech/Dtot)=90%.

f

some of the pore gas velocities used are very small (2 to 97 cmmin−1) and the dispersion coefficient is therefore verysensitive to even small changes in flow paths through thecolumns at the different velocities used during the course ofthe experiments. This problem will likely be more pro-nounced in more heterogeneous materials.

The ratio, Ddiff/Dtot, (calculated using Eq. (3)) as a functionof pore gas velocity is shown in Fig. 3a for fixed values of gas-filled porosity. Indicated also are the levels where this ratioequals 0.5, 1, 2, 5, and 10%. As expected, the relative impor-tance of molecular diffusion increases with decreasing poregas velocity because of the decrease in Dmech with pore gas

Fig. 4. Average value of dispersivities calculated using Eq. (1) from totaldispersivity and predicted from model, as a function of relative gas satura-tion. The line denotes the predicted dispersivities from model, solid pointsdenote the observed data, and error bars denote the ±standard deviation forcompost and empty points denote the observed values of different materialin different literature. Insert figure shows a magnified view of literature data.

105P. Sharma, T.G. Poulsen / Journal of Contaminant Hydrology 107 (2009) 101–107

velocity. The data also show that the importance of moleculardiffusion increases with increasing gas-filled porosity forconstant pore gas velocity. As Ddiff increases with increasinggas-filled porosity according to Eq. (3), this means that Dmech

decreases with increasing gas-filled porosity for constant poregas velocity. As discussed by Gidda et al. (2006) and Poulsenet al. (2008), this is likely because the pore network tortuositydecreases with increasing gas-filled porosity. In general, it isseen that mechanical dispersion dominates over moleculardiffusion throughout the entire u–ε range investigated for theyard waste compost. Fig. 3b shows Ddiff/Dtot, calculated bylinear interpolation of the data in Fig. 3a, for selected pore gasvelocities as a function of gas-filled porosity. As indicated, thelevels corresponding to Ddiff/Dtot=0.5, 1, 2, 5, and 10%. Ddiff/Dtot increases very rapidly with ε at lower values of ε butbecomes much less dependent on ε at higher values of ε. Atlow values of ε, Ddiff is very small and a relatively small changein Ddiff therefore has a large impact on the ratio Ddiff/Dtot.Fig. 3b also shows that the shapes of the Ddiff/Dtot–ε rela-tionship are very similar for different values of u. This is likelyindicates that the shape is mainly controlled by the pore sizedistribution of the compost. Fig. 3c shows the relationshipbetween pore gas velocity and gas-filled porosity correspond-ing to fixed values of Dmech/Dtot=90, 95, 98, 99, and 99.5%.The best fit lines for the relationships are also shown in Fig. 3c.The relationships can with some approximation be regardedas linear. This means that for the compost, the pore gasvelocity corresponding to a given combination of gas-filledporosity and Dmech/Dtot ratio can be expressed as:

uDmech=Dtot = ADmech=Dtote + BDmech=Dtot ð5Þ

where uDmech/Dtot is the value of u corresponding to a givencombination of ε and Dmech/Dtot and ADmech/Dtot and BDmech/

Dtot are the slope and intercept of the u–ε–Dmech/Dtot rela-tionship, respectively. The slope, A increases with increasingvalues of Dmech/Dtot and will approach infinity as Dmech/Dtot

approaches 100%. All u–ε–Dmech/Dtot relationships shown inFig. 3c passes through or very close to the point (ε=0.12 cm3

cm−3, u=0 cm min−1). This means that for the compostEq. (5) can be rewritten as:

uDmech=Dtot = ADmech=Dtot e − 0:12ð Þ: ð6Þ

The physical significance of the value ε=0.12 cm3 cm−3 isthe value of gas-filled porosity below which the pore systembecomes discontinuous such that gas flow is no longerpossible. This value is of course only valid for the yard wastecompost packed to a dry bulk density of approximately 0.83 gcm−3, as used in this study.

Values of oxygen dispersivity (α) in the yard wastecompost were calculated from the measured values of u,Dtot, ϕ, and ε using Eqs. (2)–(4). Dispersivity as a function ofrelative gas saturation (ε/ϕ) is shown in Fig. 4. Values of αare averaged across all pore gas velocities analyzed for theindividual gas contents, and the standard deviations corre-spond to the individual α-values calculated for all individualcombinations of u, Dtot, ϕ, and ε at which dispersion coef-ficients were measured. Also shown are α-values measuredby Costanza-Robinson and Brusseau (2002), Gidda et al.(2006), and Poulsen et al. (2008) for fine sand, silt loam, and

filter sand, respectively. Fig. 4 shows that, for the yard wastecompost, gas dispersivity increases with decreasing relativegas saturation and the decrease being most prominent forlower values of relative gas saturation. Increasing α withdecreasing ε/ϕ is also seen for the silt loam (Gidda et al.,2006) and the filter sand (Poulsen et al., 2008) whereas thefine sand data (Costanza-Robinson and Brusseau, 2002) donot exhibit a clear trend. Considering all four data setstogether it appears that α is relatively constant at high valuesof ε/ϕ and increases with decreasing ε/ϕ at lower values ofε/ϕ. An explanation for this behavior is that at high values ofε/ϕ, gas-filled pore system tortousity is relatively low andless sensitive to changes in ε. At low ε/ϕ, gas-filled poresystem tortousity is relatively high and much more sensitiveto changes in ε. This means that even a small decrease in εwill greatly increase gas-filled pore system tortousity andthus the gas dispersion.

It is proposed here that the relationship between α andε/ϕ is modeled using a power function relationship. Thistype of relationship has been widely applied to model gastransport parameters in porous media as a function of mediaproperties (Ball et al., 1988; Moldrup et al., 2005; Poulsenet al., 2007; Poulsen and Blendstrup, 2008). Here it is pro-posed that α is modeled as:

α = C1 + C2e/

� �C3ð7Þ

where C1, C2 and C3 are constants. For the dispersivity datameasured in this study (in cm) for the yard waste compost,the values of C1, C2 and C3 were fitted to 1.4 cm, 0.36 cm,and −1.7, respectively. Eq. (7) with these parameters isplotted in Fig. 4 together with the measured data.

The values of α, determined at givenwater content for thefive different pore gas velocities considered in this study weregenerally observed to be independent of pore gas velocity,which is also the case for the liquid dispersion. This is seen in

Fig. 5. The relative dispersivities (ratio of actual dispersivity to thedispersivity at 1 L min−1 gas flow rates) as a function of pore gas velocities.

106 P. Sharma, T.G. Poulsen / Journal of Contaminant Hydrology 107 (2009) 101–107

Fig. 5 that shows the ratios of gas dispersivity at any givengas flow rate to the gas dispersivity at a gas flow rate of 1.0 Lmin−1 for the nine gas-filled porosities considered in thisstudy. The figure also shows that the uncertainty in gasdispersivity values is larger for low values of pore gas velocity.This is likely due to differences in gas flow channeling andinhomogeneous packing and indicates that gas dispersivity isvery sensitive to porousmedia physical conditions at low poregas velocities.

The average value of retardation factor, R, across all experi-ments was 1.16, indicating relatively low retardation asexpected since oxygen is a relatively sparingly soluble com-pound with a very weak sorption affinity. No relationship wasobserved between R and ω, indicating that dissolution intothe water phase was likely not responsible for the observedretardation of oxygen transport through the yard wastecompost. A relatively weak and not statistically significantrelationship between R and gas flow velocity (u) was ob-served which indicates a weak tendency for higher apparentretardation at higher gas flow rates. A reason could be a slighttendency for increased gas flow channeling at higher flowrates and the creation of zones with lower gas flow whichcould cause the retardation.

5. Conclusions

A total of 180 breakthrough curves for gaseous oxygentransport through compost at pore gas velocities rangingbetween 2 and 97 cm min−1 and water contents rangingbetween 0.04 and 0.65 g H2O (g dry compost)−1 were mea-sured and gas dispersion coefficients and retardation factorsfor oxygen transport were fitted from the breakthrough datausing the classical advection–dispersion equation. The equa-tion generally fitted the measured data well indicating localequilibrium and a low quantity of inactive pore space.

Linear relationships between total dispersion coefficient,Dtot, (the sum of mechanical dispersion, Dmech, and moleculardiffusion, Ddiff) and pore gas velocity, u were generallyobserved with some deviations from the linear trend likelycaused by temporal changes in flow paths associated with theuse of different pore gas velocities. It was also observed thatDmech dominated over Ddiff at all water contents and u-valuesinvestigated. Dmech accounted for between 56 and 99.9% of

Dtot. In general, Dmech and its relative importance increasedwith increasing u and decreasing compost volumetric gas con-tent, ε. Decreasing ε likely increases pore system tortousity,resulting in increased dispersion. Although the ratio Dmech/Dtot is strongly dependent on both u and ε, the shape of theDmech/Dtot–ε relationships at different u are generally verysimilar. This is likely because the shape of the relation-ships is controlled by the pore size distribution and waterretention properties of the compost rather than pore gasvelocity. Relationships between u and ε corresponding tofixed values of Dmech/Dtot are approximately linear andintercept each other at a point close to u=0 cm min−1 andε=0.12 cm3 cm−3. This point also corresponds to the criticalvalue of ε (εcrit) where the gas-filled pore network becomesdiscontinuous and gas flow no longer is possible.

The gas dispersivity, α, was observed to be relativelyconstant at high values of relative gas saturation (ε divided bytotal porosity, ϕ) and increases with decreasing ε/ϕ at lowervalues of ε/ϕ. An explanation for this behavior is that at highε/ϕ, gas-filled pore system tortousity is low and less sensitiveto changes in ε. At low ε/ϕ, gas-filled pore system tortousityis relatively high and more sensitive to changes in ε. Thismeans that even a small decrease in ε will greatly increasegas-filled pore system tortousity and, thus, the gas disper-sion coefficients. It was further observed that the shape of theα–ε/ϕ relationship followed a power function which is alsothe case for other porous media characteristic parameterssuch as gas permeability or molecular diffusion coefficients,however, in order to develop a predictive model for gasdispersivity, a larger quantity of data corresponding to severaldifferent types of porous media is required.

Values of α exhibited larger variability based on break-through measurements taken at low pore gas velocities butwere less uncertain at higher gas velocities. This indicates thatthe flow field is more sensitive to changes in flow paths due tovariations in channeling at low gas velocities and makingmeasurements of dispersivity under these conditions becomemore difficult. Observations also indicate that temporalchanges in flow velocity can change the flow field channeling.

References

Ahn, B.J., Zoulalian, A., Smith, J.M., 1986. Axial dispersion in packed beds withlarge wall effect. AIChE J. 32, 170–174.

Ball, B.C., O'Sullivan, M.F., Hunter, R., 1988. Gas diffusion, fluid flow andderived pore continuity indices in relation to vehicle traffic and tillage.J. Soil Sci. 39, 327–339.

Batterman, S., Kulshrestha, A., Cheng, H., 1995. Hydrocarbon vapor transportin low moisture soils. Environ. Sci. Technol. 29, 171–180.

Blackwell, R.J., Rayne, J.R., Terry, W.M., 1959. Factors influencing the efficiencyof miscible displacement. Petrol. Trans. AIME 216, 1–8.

Brusseau, M.L., 1993. The presence of solute size, pore water velocity, andintraparticle porosity on solute dispersion and transport in soil. WaterResour. Res. 29, 1071–1080.

Carberry, J.J., Bretton, R.H., 1958. Axial dispersion of mass in flow throughfixed beds. AIChE J. 4, 367–375.

Costanza-Robinson, M.S., Brusseau, M.L., 2002. Gas phase advection anddispersion in unsaturated porous media. Water Resour. Res. 38.doi:10.1029/2001WR000895.

Delgado, J.M.P.Q., 2006. A critical review of dispersion in packed beds. HeatMass Transf. 42, 279–310.

DeMaria, F., White, R.R., 1960. Transient response study of a gas flowingthrough irrigated packing. AIChE J. 6, 473–481.

De Visscher, A., Thomas, D., Boeckx, P., Van Cleemput, O., 1999. Methaneoxidation in simulated landfill cover soil environments. Environ. Sci.Technol. 33, 1854–1859.

107P. Sharma, T.G. Poulsen / Journal of Contaminant Hydrology 107 (2009) 101–107

Edwards, M.F., Richardson, J.F., 1968. Gas dispersion in packed beds. Chem.Eng. Sci. 23, 109–123.

El-Fadel, M., Findikakis, A.N., Leckie, J.O., 1997. Environmental impacts of solidwaste landfilling. J. Environ. Manag. 50, 1–25.

Gidda, T., Cann, D., Stiver, W.H., Zytner, R.G., 2006. Airflow dispersion inunsaturated soil. J. Contam. Hydrol. 82, 118–132.

Liang, Y.K., Quan, X., Chen, J.W., Chung, J.S., Sung, J.Y., Chen, S., Xue, D., Zhao,Y.Z., 2000. Long-term results of ammonia removal and transformationby biofiltration. J. Hazard. Mater. 80, 259–269.

Moldrup, P., Olesen, T., Gamst, J., Yoshikawa, S., Komatsu, T., Rolston, D.E.,2005. Predictive–dispersive models for gas and solute diffusion coeffi-cients in variably saturated porous media coupled to pore-size distribu-tion, 1. Gas diffusivity in repacked soil. Soil Sci. 170, 843–853.

Popovicova, J., Brusseau, M.L., 1997. Dispersion and transport of gas-phasecontaminants in dry porous media, effect of heterogeneity and gasvelocity. J. Contam. Hydrol. 28, 157–169.

Poulsen, T.G., Blendstrup, H., 2008. Predicting air permeability in porousmedia with variable structure, bulk density and water content. VadoseZone J. 7, 1269–1275.

Poulsen, T.G., Christophersen, M., Moldrup, P., Kjeldsen, P., 2001. Modelinglateral gas transport in soil adjacent to old landfill. J. Environ. Eng.-ASCE127, 145–153.

Poulsen, T.G., Moldrup, P., Thorbjørn, A., Schjønning, P., 2007. Predicting airpermeability in undisturbed, subsurface sandy soils from air-filledporosity. J. Environ. Eng. 133, 995–1001.

Poulsen, T.G., Suwarnarat, W., Hostrup, M.K., Kalluri, P.N.V., 2008. Simple andrapid method for measuring gas dispersion in porous media, methodol-ogy and applications. Soil Sci. 173, 169–174.

Rao, P.S.C., Rolston, D.E., Jessup, R.E., Davidson, J.M., 1980. Solute transport inaggregated porous media, theoretical and experimental evaluation. SoilSci. Soc. Am. J. 44, 1139–1146.

Roberts, P.V., Goltz, M.N., Summers, R.S., Crittenden, J.C., Nkedi-Kizza, P., 1987.The influence of mass transfer on solute transport in column experi-ments with aggregated soil. J. Contam. Hydrol. 1, 375–393.

Rolston, D.E., Kirkham, D., Nielsen, D.R., 1969. Miscible displacement of gasesthrough soil columns. Soil Sci. Soc. Am. Proc. 33, 488–492.

Rose, D.A., 1973. Some aspects of hydrodynamic dispersion of solutes inporous materials. J. Soil Sci. 24, 285–295.

Ruiz, J., Bilbao, R., Murillo, M.B., 1999. Convective transport and removal ofvapors of two volatile compounds in sand columns under different airhumidity conditions. Environ. Sci. Technol. 33, 3774–3780.

Scott, D.S., Lee,W., Papa, J., 1974. Themeasurement of transport coefficients ingas–solid heterogeneous reactions. Chem. Eng. Sci. 29, 2155–2167.

Sharma, P., Poulsen, T.G., Kalluri, P.N.V., in press. Gaseous oxygen uptake inporous media at different moisture contents and air flow velocities. J. AirWaste Manage. Assoc. 59. doi:10.3155/1047-3289.59.6.1.

Yu, D., Jackson, K., Harmon, T.C., 1999. Dispersion and diffusion in porousmedia under supercritical conditions. Chem. Eng. Sci. 54, 357–367.