Environ. Sci. Technol. 1993, 27, 2412-2419
Efficiency of Capping Contaminated Sediments in Situ. 2. Mathematics of Diffusion-Adsorption in the Capping Layer Greg J. Thoma, Danny D. Reible,' Kalilat T. Valsaraj, and Louis J. Thlbodeaux
Department of Chemical Engineering, Louisiana State University, Baton Rouge, Louisiana 70803 Simple analytical models for hydrophobic organic pollutant release from a contaminated sediment capped (covered) with a clean material are presented. Three situations are considered: a base case of uncapped contaminated sediment, a case associated with high sediment contamination levels with essentially constant contaminated layer pore-water concentration, and a more general case in which the leaching of the pollutant from the contaminated sediment layer is considered. The model results are compared to experimental data from laboratory bench-scale simulators. Both models show generally good agreement with the experimentally observed fluxes, indicating that the fundamental transport mechanisms are well understood. Sensitivity analysis suggests that the key factors controlling the effectiveness of a cap for isolating the benthic and pelagic communities from exposure to hydrophobic organic chemical pollutants are the cap thickness and its organic carbon content. A cap effectiveness factor is defined and plotted as a function of time and the above parameters.
Introduction Past industrial discharges to surface waters included levels of hazardous or toxic substances no longer regarded as acceptable. Many of the compounds discharged were hydrophobic organic compounds (HOCs). Once discharged, the HOCs tended to partition to the organic fraction of the suspended particulate and colloidal matter, eventually being deposited on lake and estuary bottoms and becoming incorporated into the bed sediments. One-eighth to one-fourth of all Superfund National Priority List sites include contaminated sediments, indicating the importance of contaminated sediment remediation (1). Remediation of contaminated sediments by conventional dredging techniques is often problematic due to releases to the environment during removal and the typically large volume of sediment and associated water requiring treatment. One remediation option in low hydrodynamic energy environments is in situ capping with clean sediment. Capping isolates the contaminants from the pelagic and benthic ecosystems and potentially contains the contaminants until natural degradation processes occur. For example, slow dechlorination of polychlorinated biphenyls in anaerobic environments has been reported (2,3). The dechlorination half-life (from higher to lower chlorine number) is roughly 30-60 years based on information given in ref 3. Transport modeling of capped contaminated sediment can quantify the effectivenessof a cap as a chemical barrier, permitting rational engineering design criteria to be developed and implemented. The purpose of the models presented in this paper is to explore the fundamental transport processes involved in hydrophobic pollutant release from bed sediments and to provide a starting point for estimation of pollutant release rates which in turn 2412
Envlron. Scl. Technol., Vol. 27, No. 12, 1993
provide the basis for exposure and risk assessment and ultimately resource allocation. Benthic organisms are effective mixers (this mixing process is referred to as bioturbation) of the top several centimeters of bed sediments. Order of magnitude estimates for the rate of PCB release indicated bioturbation to be 100 000 times more effective than diffusion through the pore water (4). This is due to the reworking of heavily contaminated sediment particles. Thus, the addition of a clean sediment layer should decrease pollutant release by the elimination of direct bioturbation of contaminated particles. In addition, capping will reduce the pollutant flux by (1) increasing diffusive and advective transport lengths, (2) retarding pore-water processes through sorption, and (3) eliminating resuspension and direct desorption of pollutants to the water column. Capping has been shown to be technically feasible from an engineering perspective in field-scale tests, although cap integrity over long periods has not yet been demonstrated ( 5 , 6 ) . The Army Corps of Engineers Waterways Experiment Station monitored capped contaminated dredged material in Puget Sound for 18 months. No evidence of significant chemical migration into the cap was observed (7). However, another study using laboratory-scale reactors showed that a clean, sandy cap material was not 100% effective at preventing the breakthrough of contaminants from contaminated dredged material (8). One aim of the models presented here is to quantify the effectiveness of a cap as a chemical barrier. Experiments examining the effectiveness of capping were also conducted in our laboratory and reported in a previous paper (9). An uncapped base case and four capped simulations were performed. Each of the capped simulations used a different cap material over the same contaminated sediment. Simple models for the steadystate flux of a pollutant through a cap and the breakthrough time were presented. The assumptions in these models were severe. One method for estimation of the breakthrough time assumed an error function solution with asemi-infinite sediment cap domain. Sucha model cannot predict the transient chemical flux through a finite cap. Other calculations for the steady-state flux, breakthrough time, and time-to-steady-state assumed the contaminated sediment concentration of the pollutant remained constant at its initial value even though experimental evidence (decreasing flux at long times) suggested this was not true. This paper eliminates the above drawbacks, presenting and discussing models for a capped sysem which can predict transient chemical fluxes. Several parameters are necessary for model implementation. These include the porosity and bulk density of the sediments, the partition coefficient for the chemical(s) between the pore water and the sediment(s), and the molecular diffusivity chemical(s) in water. Because of the number and the uncertainty associated with the measurement or estimation of these parameters, we do not feel that development of complex numerical models 0013-936X/93/0927-2412$04.00/0
0 1993 Amerlcan Chemical Soclety
is presently warranted. The models presented here are simple analytical models based on pollutant pore-water concentrations. Local equilibrium and linear partitioning for the pollutant between sediment particles and the aqueous phase are assumed. Explicit mathematical treatment of the sediment particles is not included, and kinetic effects associated with desorption from sediment particles are not incorporated into these models. Model predictions are compared to our earlier experimental data (9). Key factors affecting the breakthrough and maximum chemical flux are also discussed. Development of a Conceptual Model We conceptualize the capped system as two finite layers; the cap is initially contaminant free, and the underlying sediment is uniformly contaminated. Note in the application of these models that the origin (Le., z = 0) and positive direction have been chosen differently for mathematical convenience in the analysis of each case. The focus of this paper is diffusive processes in a cap/sediment system, and hence advective transport, including advective transport during cap consolidation, is not considered. Thibodeaux ( 4 ) has estimated, for PCBs in New Bedford Harbor, the depth of chemical penetration into a typical cap during primary consolidation (when the chemical is advected upward) would be of the order of 1mm and should occur in a few months time (6). Due to the high sorptive capacity of most sediments in comparison to the water solubility of many hydrophobic organic chemicals, the concepts of the solubility limit and critical loading of contaminated sediment are important ( 4 ) . Linear partitioning of the HOC between the aqueous and sediment phases is assumed, thus WA = KdpA, where W A is the sediment loading (mg/kg); the partition coefficient, Kd, has units of L/kg and the water concentration, PA, has units of mg/L. Linear partitioning can only hold until the aqueous solubility, p i (mg/L), of the chemical is reached at the critical sediment loading, W: = Kdp;. Above this loading the pore-water concentration remains constant at the chemical solubility. Different boundary conditions apply at the bottom of the cap layer if WA > W: than if WA < W: as discussed below. Field conditions with W A > W: are common. Sediment-Sideand Water-SideResistancesto Mass Transfer. The benthic boundary layer resistance to mass transfer may be an important factor. For the uncapped situation with WA > w:, it is the only mass transfer resistance. To determine the conditions under which water-side resistances to mass transfer should be incorporated, we first consider an uncapped, uniformly contaminated sediment layer. In order to compare the resistances to mass transfer on each side of the sedimentwater interface, estimates of the mass transfer coefficients are required, the inverse of which estimates the resistance to mass transfer. Our purpose here is to estimate the water-side mass transfer coefficient in our laboratory simulations. It has been shown that the mass transfer coefficient for dissolution of a (solid) solute into fully developed laminar flow is given by (10)
where r = dU$dz is the velocity gradient at the solidfluid interface, D , is the chemical diffusivity in water (cm2/
s), and L is the length of the mass transfer region. Assuming maximum shear stress, TO, at the solid-fluid , the interface and a linear decrease of shear stress, r Z x to free surface ( z = H),the velocity profile above the sediment surface is
where p is the fluid viscosity. The volumetric flow rate past the surface, q , can be found from: (3)
with Was the width of the dissolving region. For a given flow rate, eqs 2 and 3 provide a relationship for the velocity gradient, J?, needed in eq 1,giving finally:
s)1’3
k, = l.l65( IPLW
(4)
We are not aware of a previous publication of this mass transfer coefficientrelationship although it follows directly from ref 10. Clearly, at very short times the water-side resistance will be controlling because no concentration gradient exists within the sediment. Thus, the sediment-side resistance will increase with time as the sediment is depleted of the chemical and the sediment-side diffusive path length increases. We will use a semi-infinite-domain errorfunction solution for the concentration profile (11),which should be valid during the initial stages of contaminant depletion, as the basis for estimating the sediment-side mass transfer coefficient. The flux N A= -De d(p~)/dxl,,,i, through the sediment-water interface (swi) is given by N A = p ~ [ ( D & ) / ( a t ) lwhereNA ~/~ is the chemical flux (pg/cm2 s), po is the initial pore-water concentration in the contaminated sediment (pglcrns),E is the bed porosity (cm3 water/cm3 total), t is time (s), De = D , ( ~ / T=) DWe4I3is an effective diffusivity which corrects the molecular diffusivity of the chemical in water for the reduced area, E , the tortuousity, 7 , of the sediment bed. For a saturated unconsolidated granular sediment, the tortuousity is approximately € 4 3 (12). Rj = t + pd(d is a retardation factor associated with accumulation on the immobile (sediment) phase, pb is the bulk density (g/cm3) (11). Taking the overall mass transfer resistance, llKo, to be the sum of the water-side and sediment resistances, given by l/kl and p o / N ~respectively, , yields l/Ko = l/kl + ~(at)/(D,€4/3Rj)]1/2. A simple calculationusingparameters taken from the experimental simulators described in our earlier paper (9) and the physicochemical properties of 2,4,6-trichlorophenol (Table I) shows that the time for the sediment-side resistances to become equal to the waterside resistance is a few hours. Note that more hydrophobic chemicals (large Rj) are more likely to be water-side controlled. The same calculation for another compound, pyrene, shows that the time required for the resistances to become equal is about 2 months (pyrene diffusivity = 6 X lo4 cm2/s; Kd = 6000 L/kg). As t 0 or Rj a,the water side controls; while as t a, the sediment side controls. It may at first seem counterintuitive that a chemical more highly retarded in the sediment would be more likely to be water-side controlled. Consider two chemicals with initially equal pore-water concentrations with different values of Rf.The
- - -
Envlron. Sci. Techno)., Vol. 27, No. 12, lSS3 2419
Table I. Parameters for Model Fit.
inoculation batch I1
inoculation batch I
balsam sand cap Tao River cap quartz sand cap predicted fit to data predicted fit to data predicted fit to data capprop D,and cap D,and cap D,and prop only KDZonly andIC &only prop only KD, only cap properties D, (cmz/day) €1
PI
KDI(cm3/g) cap depth (cm) contaminated sediment properties €2
PZ (g/cm3)
0.62 0.38 1.53 0.98 0.7
0.7 0.5 1.53 1.1 0.8
0.85 0.38 1.53 0.98 0.7
0.62 0.5 0.84 4.6 0.7
0.8 0.5 0.84 3.0 0.8
0.72 0.5 0.84 4.6 0.7
0.62 0.37 1.61 1.79 0.6
1.2 0.3 1.61 1.5 0.7
0.82 0.37 1.61 1.79 0.6
0.62 0.45 0.81 27.5 0.7
1.21 0.45 0.81 113 0.7
0.45 0.81 27.5 1.5
0.45 0.81 27.5 1.5
0.45 0.81 23.5 1.5
0.45 0.81 27.5 1.5
0.45 0.81 27.5 1.5
0.45 0.81 30 1.5
0.45 0.81 27.5 1.5
0.45 0.81 27.5 1.5
0.45 0.81 120 1.5
0.45 0.81 27.5 1.5
0.45 0.81 113 1.5
KDZ(cm3/g) contaminated depth (cm) 150 150 150 250 310 150 150 150 initial [TCPI (mg/L) a Boldface numbers indicate the parameters which were significant in fitting the observed data.
mass of chemical present in the system will be larger for the chemical with the larger Rf,thus the time required for depletion of the sediment and build up of sediment-side resistance will be longer. Thus, given an initially uniform contaminant distribution in the sediment, the influence of water-side resistance will be greater for more hydrophobic chemicals. This effect has been observed in our lab. In the case of a capped sediment, the initial resistance is entirely in the cap. The steady-state resistance to mass transfer from the cap can be estimated as a/(D,c44 = p o / N ~where a is the effective cap thickness (cm). Assuming that the contaminated sediment pore-water concentration remains constant provides the minimum estimate of the sediment-side mass transfer resistance. The water-side resistance is estimated from eq 4 above. For 2,4,6-trichlorophenol with the same conditions used in the calculations above, 89% of the resistance at steady state can be attributed to the cap layer. The actual overall sediment-side resistance will be larger if depletion in the contaminated sediment zone occurs during the transient approach to steady state. In a field situation, the cap thickness will be much greater (50 cm or more), and the water-side mass transfer coefficient should also be greater. Thus, in most field applications we do not expect that water-side resistances will be important after capping. However, analysis of laboratory data and possibly fieldscale demonstrations with thin or no caps should include water-side resistances.
Development of Mathematical Models These models are based on contaminant mass balances in the pore water of both the cap and contaminated sediment layers. Equilibrium between the pore water and sediment is assumed to be rapidly established. The following equation is a differential mass balance for the diffusive transport for a nonreactive (conserved) sorbing species in a porous medium (13) (5) 2414
University Lake cap predicted fit to data D, and KDZonly
Envlron. Scl. Technol., Vol.
27, No. 12, I993
.
+
*
250
250
250
Contam. layer
v p =o
k - - - - - - L d ' Figure 1. Conceptual diagram of a capped contaminated sediment. The rate of diffusive transport from the sediment (I) is equal to the rate of flushing of the overlying water (neglecting evaporative losses) (ii) and the rate of transport through the benthic boundary layer (111).
Invariance of the sediment and chemical properties has been tacitly assumed in the derivation of eq 5. It should be noted that the factor Rf is not applicable in steadystate analyses, nor should it be applied in Fick's first law since it arises from the accumulation term in the transient analysis. Equation 5 applies to the capping layer under all conditions considered and to the contaminated sediment layer when W A < coi. We now consider three cases: a base case with no cap and one- and two-layer models of a capped system. Figure 1depicts a capped system. The base case model incorporates water-side effects while both capped models neglect water-side resistances consistent with the preceding discussion. The one-layer model considers only the dynamics of the capped layer; the twolayer model considers the dynamics in both the contaminated and cap layers. Base Case: No Cap. To assess the effectiveness of a cap as a chemical barrier, the release rate in the absence of a cap must be estimated. We apply eq 5 to an uncapped
sediment subject to the following initial and boundary conditions: PA(Z,O)= PA0
to approach steady state. Appropriate initial and boundary conditions are
(6a)
where z = 1 at the sediment-water interface and z = 0 at the bottom of the contaminated layer. These conditions imply an initially uniformly contaminated sediment and no downward contaminant flux at the bottom of the polluted layer. We now consider the boundary condition for z = 1. The rate of diffusive transport from the sediment is equal to the rate of flushing of the overlying water (neglecting evaporative losses) and the rate of transport through the benthic boundary layer. Thus (7)
where q is the volumetric flow rate of water past the sediment-water interface (cm3/s), and A = WL is the exposed sediment-water interfacial area (cm2) as shown in Figure 1. Solving the right equality for p i , then substituting into the left equality, and rearranging yields
with /3 = qkl/(D,(q + klA)). Equation 8 incorporates both the benthic boundary layer (water-side) resistance and a statement that the mass of chemical leaving the sediment equals the mass carried away by the overlying water flow; it assumes quasi-steady-state conditions in the overlying water (accumulation is neglected). Application of Fick's first law to the solution of eq 5 subject to the boundary conditions (eqs 6a,b and 8) yields the contaminant flux through the sediment-water interface (14)
where an are the roots of LY tan (a)= 01. This model is valid for WA < w:. For the uncapped situation with W A > ,(o the flux will be constant and liquid-side controlled with the flux given by NA = k,(p; - p i ) . The models presented here cannot describe the transition from OA > W: to WA < ui as the chemical is depleted from the system. The model predictions are compared to our earlier experimental data in Figure 2 (9). Equation 4 predicts a mass transfer coefficient between 2.0 and 2.8 cmlday depending on the overlying water depth (0.5-0.3 cm). The best fit (minimum sum of squared residuals) of the model to the data is for kl = 2.55 cm/day. The fit is reasonable even without including water-side resistances which, for a moderately soluble compound like TCP, should be expected as explained earlier. One-Layer Model. Here we consider a capped case with WA > W: in the contaminated sediment layer. Equation 5 applies only in the capping layer; the contaminated sediment layer has constant pore-water concentration. We assume that the sediment load of HOC will not be significantly depleted during the time required
=0
(loa)
PAW)=0
(10c)
PA(Z,O)
with z = a at the sediment-water interface and z = 0 at the bottom of the cap layer. Equation 1Oc states that the water column concentration is zero and there is no waterside mass transfer resistance. For laboratory analysis or field conditions where accumulation in the water column is important, eq 1Oc should be replaced with eq 8. A solution to this problem is also available (14). Application of Fick's first law to the solution to eq 5 subject to the conditions in eqs loa-c yields a relation for the chemical flux at the sediment-water interface (14)
This model may also be considered as a short-time approximation to a contaminated sediment layer with OA < W; since depletion of the contaminated sediment layer will not significantly affect the system dynamics until longer times. It represents an upper bound to the flux from a capped system. This model was used by Wang et al. (9)for analysis of breakthrough time and time to reach steady state. Two-Layer Model. Here we consider transient diffusive transport in both the cap and contaminated sediment layers. Equation 5 applies in both layers subject to the following initial and boundary conditions: PAl(Z,O)
aPAl
&-o
=0
(12a)
'PA2
= -&=o
with z = -a at the sediment-water interface, z = 0 at the contaminated sediment-cap interface, and z = 1 at the bottom of the contaminated sediment layer. The subscripts 1and 2 refer to the cap and contaminated layers, respectively. Equations 12a and 12b state that the cap is initially free of contamination and that the underlying sediment is uniformly contaminated. Equation 12cstates that there is negligible water-side mass transfer resistance or accumulation. Equations 12d and 12e state the continuity of concentration and flux at the cap-sediment interface. The no-flux boundary condition (120 at z = 1 describes the laboratory system we have used to collect data. Solving this system of equations by the method of Laplace transformation, and applying Fick's first law at the sediment-water interface, yields the following exEnviron. Scl. Technol., Vol. 27, No. 12, 1993 2415
2500
0 "
' I
5
0
'
' 1 " " "
10
" I " '
1 " " 1 " ' "
20 25 TIME (days) 15
30
35
Figure 2. Comparison of predicted to observed TCP flux in an experimental uncapped system.
pression for HOC flux into the water column: m
n=i
sin(&,)
+ (aa + p l ) sin(ua,) cos(pla,)l
(13)
where anare the roots of a cos(aa) cos (pla) - sin(aa) sin (dd;CL = [ ( D e ~ R f ~ ) / ( o e z R ~a ) = l ~ Dsd(De2~). '~; This solution has apparently not been previously reported in the literature.
Results and Discussion In this section we will compare the results of the twolayer model with our laboratory-scale experimental results and present a sensitivity analysis for key parameters affecting the effectiveness of a cap as a chemical barrier. Figure 3A-D presents a comparison of the model results to experimental data from laboratory-scale simulations of capped contaminated sediment (9). The solid line is the predicted flux based on measured or estimated parameters. The dashed line has been fit to the data using a pattern search for the minimum sum of squared residuals with the cap properties and initial concentration as adjustable parameters. The dotted line was similarly fit to the data but using only the water diffusivity and contaminated sediment partition coefficient as adjustable parameters. The parameters used in the model curves shown in Figure 3 are presented in Table I. In the experiments, the sediment used as the contaminated sediment layer in all four runs was a local sediment (University Lake). However, it was inoculated on two separate occasions, once for the Balsam and Tao cap runs (Figure 3A,B) and later for the sand and University Lake cap runs (Figure 3C,D). Note from Table I that the fitted contaminated layer partition coefficient is different for each inoculation batch. This parameter was the most important in fitting the data, and the difference suggests some change in sediment treatment or inoculation conditions. Literature correla2416
Envlron. Scl. Technol., Vol. 27, No. 12, 1993
tions, based on mostly pesticides and polyaromatic hydrocarbons, suggest that the partition coefficient should fall between 2 and 70 L/kg (15). Again, the model predictions are qualitatively correct, predicting both the form of the flux versus time relation and the characteristic time to reach the maximum flux. Sensitivity Analysis. The results presented in this section are based on the parameters presented in Table I1 for a hypothetical contaminated sediment site; we do not purport that these simple model predictions actually describe the situation centuries hence. The purpose of the simulations is three-fold: (1) to demonstrate that capping can be effective for long-term chemical isolation if the cap integrity is maintained, (2) to point out the potential trade-off of acute for chronic exposure, and (3) to highlight the importance of the various parameters in designing a cap. Figure 4 compares the model flux as a function of time for each of the models above. The predicted flux from the two-layer model is completely bounded by the one-layer model prediction, and the time to the maximum flux is approximated by the approach to steady state in the onelayer model. All real systems should have leaching rates bounded by the one-layer model since an upper bound flux is determined by a constant contaminated sedimentcap interface concentration. Figure 5A,B shows the TCP flux from the hypothetical site as a function of the effective cap thickness and cap partition coefficient,respectively. As expected, increasing either parameter decreases the maximum flux and increases the time required to reach the maximum flux. The reason that the curves intersect a t long times is that the degradation is not incorporated into the model and, since the initial TCP mass is the same in each case,the integrated area as t m under each curve must be the same. That is, without degradation processes, the effect of a cap is to reduce the maximum flux but increase the total time of contaminant release. The range of organic carbon in the cap in Figure 5B is from 1.3 to 6.3 % . This is a reasonable range for natural sediments which might be used. It may be possible to amend natural sediments with fly ash or organic cations (by ion exchange with natural cations) to increase the organic carbon content of the cap, thus increasing its effectiveness as a chemical barrier (16,17). As indicated above, the observed best-fit model parameters are in reasonable agreement with predicted values. The degree of isolation of HOCs in bed sediments afforded by a cap changes with time as shown by the model results presented above. An appropriate comparison for emissions from a capped system is the base case above. Normalization of cap effectiveness is awkward since at short times the flux from a capped system will be zero, and at long times the flux from an uncapped system approaches zero (Le., all of the HOCs leaches from the sediment). We define a normalized effectiveness factor + N,) where Nu,and N , are for the cap as (Nu,- Nc)/(Nuc the (hypothetical) uncapped and capped fluxes, respectively. Thus at short times when the capped flux is zero, the effectiveness factor is 1; and at long times when the contaminant would have completely leached from an uncapped sediment, while the delayed emissions from a capped sediment would not yet have ceased, the effectiveness factor is -1. If the fluxes are equal, the effectiveness factor is zero. Figure 6A,B presents a cap effectiveness factor as a function of time, cap thickness,
-
0
5
10
15
20
25
0
5
10
Time(days) 0
5
10
15
30
35
40
15 20 25 30 """""""""""""I
35
40
15
20
25
Time (days) 20
500
25
0
5
t"""
10 " " '
450
400 h
4E
\
0
5 -3
350 300 250 200 150 100 50
0 Flgure 3. Comparison of predicted to observed TCP flux for experimental capped systems: (A) balsam sand cap; (B) Tao River cap; (C) quartz sand cap; (D) Unlverslty Lake cap.
Table 11. Properties of TCP and Sediment Used in Sensitivity Analysis TCP Properties molecular weight diffusivity in water (Wilke-Chang method) (cm2/s) aqueous solubility (mg/L) octanol-water partition coeff., log KO, organic carbon partition coeff., log K, Sediment Properties contaminated sediment bulk density (g/cms) porosity organic carbon fraction capping sediment bulk density (g/cm3) porosity organic carbon fraction effective cap depth (cm) Contamination Parameters TCP concentration (mg/kg) pore-water concentration (mg/L) contaminated depth (cm) contaminated area (ha) critical loading (mg/kg) water velocitv (cm/s)
197.46 7.2 X lo4 (18)
800
I\
-
Two Layer
..... One Layer
No Cap
800 (19) 3.72 (20) 3.23 (9) 1.3 0.4 0.03 1.5 0.4 0.04 35
5000 98 15 10 40 760
7.5
and cap partition coefficient. For the range of parameters used in the simulation, the effectiveness approaches -0.9 in approximately 500 years. In other words, if no action is taken the TCP will have nearly completely leached by this time while the capped system will continue to release TCP to the aquatic ecosystem. This analysis assumes that there will be no biodegradation during the project lifetime. In fact many organics will degrade in the decades time scale of containment, and a significant benefit of capping will be realized. For the uncapped and two-layer models,
0
500
1000
1500
2000
Time (years) Figure 4. Comparison of model predictions at a hypothetical contamination site.
which are valid below the sediment critical loading, the effect of a first-order loss process (presumably biodegradation) is easily incorporated by multiplication with the term e-kt where k is a first-order decay constant. The maximum value in the trace for the 40-cm-thick cap shown in Figure 5A would be 7 mg/m2 yr instead of 470 mg/m2 yr with k = 0.00693 yr-l which corresponds to a chemical half-life of 100 yr in the sediment bed. Environmental half-lives are not well quantified for many chemicals, and while degradation is an important benefit of capping, the application of this correction factor should be used cautiously. For metals or other elemental contaminants (e.g., As) of concern, biodegradation may not be a Envlron. Sci. Technol., Vol. 27, No. 12, 1993 2417
loo0 900
800
700
600
500 400 300 200 100 0 0
2Ooo
lo00
3000
0
4000-
lo00
TIME (years)
2Ooo
4000
3000
TIME (years)
Figure 5. (A) Effect of cap thickness on the flux of TCP. (8) Effect of cap partition coefficient on the flux of TCP. 100
90 80 h
E 70
3
8
60
50 50
E
+
40
Q
30
20 10
0 0
100
200
300
400
500
600
" 0
100
200
300
400
500
Flgure 6, (A) Effectiveness factor of a cap as a function of the cap thlckness and tlme. (B) Effectiveness factor of cap partition coefficient and tlme.
significant factor in reduction of the toxicity. For these contaminants and highly refractory organics, high-level acute exposure is exchanged for long-term chronic exposure by capping. This highlights the importance of long-range planning in the application of capping as a remediation technology.
Conclusions and Recommendations Simple analytical models can predict the flux behavior of a hydrophobic tracer compound in laboratory microcosms. The basis for a criterion indicating the importance of water-side mass transfer resistances was developed. In field applications this effect is likely to be relatively unimportant except for estimation of the base-case flux for highly hydrophobic chemicals, where it is possible that benthic boundary layer resistances may be a most important factor. In laboratory analysis of capped systems, the water-side resistances may be important since the actual cap thicknesses used normally do not approach those in the field and the flow rates of water are also typically much lower than in the field. Development of laboratoryscale tests for the effectiveness of proposed cap materials should account for water-side effects as water-side resistance in the lab will artificially enhance the apparent effectiveness of a cap material when compared to field placement where water-side resistance is probably negligible. 2418 Environ. Scl. Technol., Vol. 27, No. 12, 1993
600
Time (yr)
Time (yr)
a cap as a function of the
The agreement between the models and observed behavior of a hydrophobic tracer lays a firm foundation upon which to construct models which will more closely match conditions likely to be found in the field. As stated in the Introduction, the models presented here are primarily intended as tools for exploring the diffusive transport mechanism operative in bed sediments and thus are not intended to be directly applicable in the field. Many field conditions exist which are not incorporated in the models. These conditions include sediment heterogeneity (both physical properties and contaminant distribution), biodegradation, erosion and deposition, nonuniform initial contamination (an existing pollutant profile), and diffusion into deeper sediment. We are currently studying the advective transport of pollutants through sediment (which might be induced by tidal variations, water table height variations, salinity, or thermal gradients) and plan to incorporate advective transport in future models.
Acknowledgments We would like to thank Dennis Timberlake of the EPA Risk Reduction Engineering Laboratory in Cincinnati for support of our work on capping through Grant CR82053101-0. We have also received support for this capping project from the EPA Hazardous Waste Research Center
at Louisiana State University through Grant CR813888. Graduate student support was provided through the Louisiana State Board of Regents Fellowship program. Supplementary Material Available Details of the Laplace transformation solution to the twolayer case and the rationale for the biodegradation factor suggested for this model (9 pages) will appear following these pages in the microfilm edition of this volume of the journal. Photocopies of the supplementary material from this paper or microfiche (105 X 148 mm, 24X reduction, negatives) may be obtained from Microforms Office, American Chemical Society, 1155 16th St. N.W., Washington, DC 20036. Full bibliographic citation (journal, title of article, names of authors, inclusive pagination, volume number, and issue number) and prepayment, check or money order for $19.00 for photocopy ($21.00 foreign) or $10.00 for microfiche ($11.00 foreign),are required. Canadian residenb should add 7% GST. Literature Cited (1) Wall, T. Overview of Management Strategies for Contam(2) (3) (4)
(5)
(6)
inated Sediment; Workshop on Contaminated Sediments and the Gulf of Mexico, Aug 27-29, 1991. Brown, J. F.; Wagner, R. E.; Bedard, D. L.; Brennan, M. J.; Carnahan, J. C.; May, R. J.; Tofflemire, T. J. Northeast. Environ. Sci. 1984, 3, 167-179. Brown, J. F., Jr.; Bedard, D. L.; Brennan, M. J.; Carnahan, J. C.; Feng, H.; Wagner, R. E. Science 1987,236,709-712. Thibodeaux, L. J.; Bosworth, W. S. A Theoretical Evaluation of the Effectiveness of Capping PCB Contaminated New Bedford Harbor Bed Sediment, Final Report; Hazardous Waste Research Center, Louisiana State University: Baton Rouge, LA, 1990. O'Connor, J. M.; O'Connor, S. G. Evaluation of the 1980 Capping Operations a t the Experimental Mud Dump Site, New York Bight Apex; U.S. Army Waterways Experiment Station: Vicksburg, MS, 1983. Truitt, C. L. The Duwamish Waterway Capping Demonstration Project: Engineering Analysis and Results of Physical Monitoring; Technical Report D-86-2; U.S. Army
Engineer Waterways Experiment Station: Vicksburg, MS, 1986. (7) Brannon, J.; Poindexter-Rollins, M. E. Sci. Total Environ. 1990, 91, 115-126.
(8) Brannon, J. M.; Hoeppel, R. E.; Sturgis, T. C.; Smith, I.; Gunnison, D. Effectiveness of Capping in Isolating Contaminated Dredged Material from Biota; the Overlying Water; Technical Report D-85-10; U.S. Army Engineer Waterways Experiment Station: Vicksburg, MS, 1985. (9) Wang, X. Q.; Thibodeaux, L. J.; Valsaraj, K. T.; Reible, D. D. Environ. Sci. Technol. 1991, 25 (9), 1578-1584. (10) Kramers, H.; Kreyger, P. J. Chem. Eng. Sci. 1956,6,42-48. (11) Formica, S. J.; Baron, J. A.; Thibodeaux, L. J.; Valsaraj, K. T. Environ. Sci. Technol. 1988, 22 (12), 1435-1440. (12) Millington, R. J.; Quirk, J. M. Trans. Faraday SOC. 1961, 57, 1200-1207. (13) Reible, D. D.; Valsaraj, K. T.; Thibodeaux, L. J. In The Handbook of Environmental Chemistry; Hutzinger,O., Ed.; Springer-Verlag: Berlin, 1991; Chapter 2, Part F. (14) Carslaw, H. S.; Jaeger, J. C. Conduction of Heat in Solids, 2nd ed.: Oxford University Press: Oxford, England, 1959. (15) Lyman, W. J.; Reehl, W. F.; Rosenblatt, D. H. Handbook of ChemicalProperty Estimation Methods: Environmental Behavior of Organic Chemicals; American Chemical Society: Washington, DC, 1990. (16) Mott, H. V.; Weber, W. J., Jr. Environ. Sci. Technol. 1992, 26, 1234-1242. (17) Lee, J. F.; Crum, J. R.; Boyd, S. A. Environ. Sci. Technol. 1989,23, 1365-1372. (18) Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena; John Wiley and Sons: New York, 1960. (19) Montgomery,J. H.; Welkom, L. M. Groundwater Chemicals Desk Reference; Lewis Publishers, Inc.: Chelsa, MI, 1990; VOl. 1. (20) Shellenberg, K.; Leuenberger, C.; Schwarzenbach, R. P. Environ. Sci. Technol. 1984, 18, 652-657.
Received for review December 14,1992. Revised manuscript received June 11, 1993. Accepted June 15,1993." Abstract published in Advance ACS Abstracts,August 15,1993.
Environ. Scl. Technol., Vol. 27, No. 12, 1883 2419
