BARRIER SYSTEMS for ENVIRONMENTAL CONTAMINANT CONTAINMENT and TREATMENT - PART 2 ppsx

74 Barrier Systems for Environmental Contaminant Containment & Treatment2.2.1.2 Layers and Features In very rare cases, a cap comprises a single soil layer over waste material.Typically

Trang 1

Transport through Barriers

* With contributions by Calvin C Chien, DuPont, Wilmington, Delaware; Thomas O Early, Oak Ridge National Laboratory, Oak Ridge, Tennessee; Clifford K Ho, Sandia National Laboratories, Albuquerque, New Mexico; Richard C Landis, DuPont, Wilmington, Delaware; Alyssa Lanier, University of Wisconsin, Madison, Wisconsin; Michael A Malusis, GeoTrans, Inc., Westminster, Colorado; Mario Manassero, Politecnico I, Torino, Italy; Greg P Newman, Geo-Slope International Ltd., Calgary, Canada; Robert W Puls, U.S Environmental Protection Agency, Ada, Oklahoma; Terrence M Sullivan, Brookhaven National Laboratory, Upton, New York

4040_book.fm Page 71 Wednesday, September 14, 2005 12:43 PM

rized, and research needs are identiﬁed Chapter 1 dealt with system performancesystems improves, the design of containment systems is moving from a prescrip-

Trang 2

72 Barrier Systems for Environmental Contaminant Containment & Treatment

2.2 CAPS 2.2.1 F EATURES , E VENTS , AND P ROCESSES A FFECTING

P ERFORMANCE OF C APS

Covers and caps are engineered structures that must perform within a largerdynamic natural system and, as such, must be designed with consideration ofnatural system inﬂuences Understanding these physical processes and applyingappropriate numerical analyses to these processes can help the engineer to build

an appropriate overall system that will perform with the desired objective Theprimary processes acting on a cap are described in the subsections below

2.2.1.1 Hydrologic Cycle

The purpose of a cap is usually to minimize water infiltration into underlyingwaste, and sometimes to minimize gas transport to the atmosphere As shown inthe cap slope, cap soil properties, cap moisture conditions, and the duration andmagnitude of precipitation, ponding and water run off can occur Water that doesnot run off of the cap is either stored in depressions in the cap surface, or infiltratedinto the surface layer of the cap Water infiltrating into the surface layer of thecap is subject to evapo-transpiration Rates of evapo-transpiration depend onsurface vegetation, soil properties, surface temperatures, soil and air relativehumidities, and net solar radiation The remainder of the precipitation not trans-formed to run off or evapo-transpiration remains as storage in the cap, or, if thestorage capacity of the cap is exceeded, the water percolates through the cap.Contaminant vapors can migrate through caps by advection or diffusion.Advection rates depend on gas-phase permeabilities and pressure gradients acrossthe cap Variations in barometric pressures can increase contaminant vapor advec-tion to the atmosphere Vapor diffusion is driven by the gas-phase concentrationgradient existing across the cap Diffusion coefficients depend on soil porosityand water content, as well as contaminant molecular weight It is often assumedthat diffusion at the ground surface occurs across a stagnant surface boundaryconditions (Thibodeaux, 1981)

Water percolation and contaminant transport through the cap can also be

the migration of water or contaminant vapors through the system Natural eventssuch as earthquakes, tornadoes, ﬂoods, and melting snow can also be disruptive

can be signiﬁcant and should therefore be considered

Figure 2.1, water originates as precipitation that falls on the cap Depending on

Figure 2.2 Animal burrows or other passageways through the cap can accelerate

altered by human or biointrusion into the cap and other natural events, leading

to disparities between probable current and future percolation rates as shown in

and processes as discussed in Chapter 1, their potential impact and consequence

to the cap Although a great deal of uncertainty is associated with these eventslayer the depth of which depends on surface topography, vegetation, and wind

Trang 3

Modeling of Fluid Transport through Barriers 73

FIGURE 2.1 Features, events, and processes associated with a long-term cap.

FIGURE 2.2 Cumulative probability distribution of water percolation reaching the mill tailings for present and future conditions (From Ho, C.K et al., 2001 Sandia National Laboratory Report SAND2001-3032; October.)

Waste Percolation/leaching

Trang 4

2.2.1.2 Layers and Features

In very rare cases, a cap comprises a single soil layer over waste material.Typically however, a cap is the unique combination of soils placed in layers ontop of each other and in certain order that create the desired effect This sectionbrieﬂy outlines the general performance objective of each potential cap layer

• Ground surface layer — The top few inches of any surface soil mayneed to be treated as a unique soil region since, due to desiccation anddrying effects, this zone generally has a much higher hydraulic con-ductivity than the soil a few inches below surface This zone is espe-cially important to include when simulating inﬁltration through coversystems using numerical models

• Vegetation layers — It is common to include a vegetation growthlayer that may or may not be part of another cover layer In manycases, the vegetation can be a key to cap performance, but based onAccording to energy balance accounting, the sum of actual evaporationand transpiration are always less than the potential evaporation Thismeans that for near-surface processes, the availability of water limitsevapo-transpiration, and water that is not transpired through vegetation

is removed through evaporation In other words, if vegetation were notpresent, actual evaporation would remove a similar amount of water.The transpiration process becomes important when it is necessary todraw water from deeper beneath the surface, particularly when actualevaporation has signiﬁcantly diminished at the surface due to drying

of soils Vegetation is also critical for stability purposes on slopedcovers, as well as erosion control

• Capillary break layers — These layers are generally created withcoarse materials next to fine materials because, at a common negativewater pressure, two different soils have different water contents Cap-illary breaks can be used in caps for various purposes When placedbeneath a compacted layer, the capillary break limits percolationthrough the compacted material When placed above a compacted layer,the capillary break limits the evaporative drying of the compacted layer,because water cannot readily be drawn up in its liquid phase throughthe coarser capillary break layer when it is dry For this type of coverdesign, a model that includes coupled vapor flow should be used toassess the impact of vapor flux on barrier layer drying in the event thatupward liquid phase flow has shut down

• Barrier layers — Barrier layers are generally made of pacted, low-permeability ﬁne-grained soils A barrier layer should not

well-com-be placed directly at the surface, or it will well-com-be subjected to effects such

as extreme drying, desiccation, and freeze-thaw It is common to place

the analysis presented in Section 1.4.1, this should not be assumed

Trang 5

a barrier layer over a coarser layer to create a capillary break effect,and then place it beneath a vegetation growth layer It is not desired

to have the root zone of the plant species extend into the barrier layerwhere damage can occur While long-term barrier layer performance

is unknown and cannot be predicted with precision, the use of dense,well-graded materials for these layers has shown the best resistance tolong-term performance deterioration (Wilson, 2002)

• Storage layers — These layers are generally made of loose well gradedmaterials such that the hydraulic conductivity is sufficient to allowwater to infiltrate and subsequently be drawn back out by evaporationand/or roots The thickness of a storage layer becomes a critical ques-tion in its functionality The cover must be thick enough to keep near-surface wetting and drying processes from interacting with the waste,and to withstand long-term erosion If the cover is to limit gas fluxes

as well, there must be a zone of continual near-saturation within thislayer over time and over prolonged dry periods; either that, or thestorage layer must protect a deeper near-saturation barrier layer Long-term storage layer performance can be affected by coarse materialbreakdown, which can result in permeability loss

2.2.2 C URRENT S TATE OF P RACTICE FOR M ODELING

P ERFORMANCE OF C APS

Water movement through soils can be thought of as a three-component systemconsisting of the soil-atmosphere interface, the near-surface unsaturated zone,and the deeper saturated zone In the past, groundwater modeling has primarilyfocused on the saturated zone, which creates a discontinuity in the natural systembecause the unsaturated zone and the soil-atmosphere interface are not repre-sented Advances in unsaturated soil technology during the past decade have led

to the development of routine modeling techniques for saturated and unsaturatedsoil systems However, modeling techniques for the third component, involvingthe detailed evaluation of the ﬂux boundary condition imposed by the atmosphere,are not routinely available This section discusses some of the available codesthat can be used for the predictive modeling of processes associated with capperformance A summary of the codes considered, and some of the key featuresdifferent available software tools and their main solution processes, as well asfeature overviews and source availability Table 2.2 lists the individual program’ssolution options and features that are built into the various codes

2.2.2.1 Water Balance Method

The estimation of the amount of water inﬁltrating through a cap is essentially theestimation of the water balance for the cap The net percolation through the cap isthe remainder from precipitation after run off, surface storage, evapo-transpiration,

and solution techniques are provided in Tables 2.1 and 2.2 Table 2.1 lists several

Trang 6

Coupled, Simultaneous, nonlinear

Pre- and post-processor included; code unavailable

VADOSE/W 2D, transient

and

steady-state FEM

Pressure, temperature, vapor pressure; can be linked with slope stability software and contaminant transfer software

Coupled, simultaneous, nonlinear

Enhanced pre and processor included; climate and soils database included;

post-user support included;

commercially developed for cover/cap design

Full CAD data input and mesh generation;

Microsoft certiﬁed for XP and lower OS

www.geo-slope.com

Oxygen or radon diffusion, dissolution, decay

Subsequent linear

Earthquake seismic analysis using VADOSE/W generated pore pressure data

Supplemental Integrated with program

QUAKE/W

Slope stability analysis using VADOSE/W generated pore pressure data

Supplemental Integrated with program

SLOPE/W

Trang 7

Subsequent nonlinear

Integrated with program CTRAN/W

HELP 1D, quasi 2D,

Analytical

Water balance Analytical Climate and soil database

included; not physically based; limited design application; assumes unit gradient

Text in editor or Windows dialogues

www.wes.army.mil/el/

elmodels/helpinfo.html

UNSAT-H 1D, transient

FEM

Pressure with vapor Nonlinear Pre- and post-processor

available but excluded Code available

Text in editor or Windows dialogues

Nonlinear Pre- and post-processor

included; CAD mesh generation add-on

CAD and Windows dialogues

www.ussl.ars.usda.gov/

models/hydrus2d.HTM

Temperature Subsequent linear

Contaminant transfer Subsequent

Coupled, Simultaneous, nonlinear

Limited pre- and processor available from independent suppliers Code available; users can customize

post-Limited CAD and text in editor

www-esd.lbl.gov/

TOUGH2

Trang 8

multi-can solve contaminant ﬂow as advection/

dispersion or particle tracking

Coupled, simultaneous, nonlinear

Limited pre- and processor with 3D grid generator available from independent sources Unix or

post-PC based; code included;

user can customize; USA only

Limited CAD with text input

Linear Can automatically optimize

layer thickness

Text entry RAECOM-cloned

calculator available on the web:

wise/uranium/ctc.html

Coupled, physical coupling between equations; simultaneous, more than one equation solved at same time (must be coupled); subsequent, more than one equation

solved one after the other at each time step; supplemental, data from completed analysis used in separate analysis; linear, material properties not a function of variable

http://www.antenna.nl/

being solved; nonlinear, material properties change with variable being solved, so iterations required; analytical, no partial differential equation, one pass solution.

Trang 9

Evapor -ation

ation

Transpir- ing

Freez-Run Off

ing

Pond-Soil Properties

SoilCover Pressure, temperature, vapor pressure with

pseudo gas

VADOSE/W Pressure, temperature, vapor pressure; can

be linked with slope stability software and

contaminant transfer software

VADOSE/W generated pore pressure data

Slope stability analysis using VADOSE/W

generated pore pressure data

Trang 10

Evapor -ation

ation

Transpir- ing

Freez-Run Off

ing

Pond-Soil Properties

FEHM Multi-phase, multi-component heat, mass,

gas, air including double porosity ﬂow;

can solve contaminant ﬂow as

advection/dispersion or particle tracking

RAECOM Radon-gas concentration and ﬂux through

a multi-layer system

RP, rigorous physically based with assumptions limited to current understanding of real physical processes; RE, rigorous physically based but

with empirical components or built-in limiting assumptions; SE, semi-empirical, equation based but user sets limits or there are limited built-in

assumptions; E, empirically based, extreme limiting assumptions and little physical bases for generated data; A, analytically based — no partial

differential equations; FF, free-form functions, user can customize; CF, closed-form functions, curve-ﬁt parameters.

Trang 11

and soil storage are considered The ﬁrst method used for water balance lations was developed by Thornthwaite and Mather (1957) This method was used

calcu-by Fenn et al (1975) to analyze leachate generation at municipal solid wastelandﬁlls

Typically, the water balance method is based on monthly climatic variables.The monthly inﬁltration, I (cm), into a cover is given by:

where P is precipitation (cm) and R is surface run off (cm) Surface storage wasnot considered by Fenn et al (1975) Run off is calculated from precipitationusing a run off coefﬁcient, C:

Fenn et al (1975) provided values of C for different soil types and slopes, withvalues ranging from 0.05 for sand with less than a 2% slope, to 0.35 for a steeplysloped (>7%) clay layer

Thornthwaite and Mather (1957) also provided tables for determining tial evapo-transpiration (PET) as a function of mean temperature, heat index, andhours of sunlight When PET exceeds infiltration, moisture storage in the cap isexpected to decrease unless the cap was already dry PET cannot exceed the waterstored in the cap plus the infiltration for the month When infiltration exceedsPET, evapo-transpiration is equal to PET, and excess infiltration increases themoisture storage in the cap to field capacity Excess infiltration above the fieldcapacity of the cap percolates through the cap

poten-2.2.2.2 HELP

The hydrologic evaluation of landﬁll performance (HELP) model was developed

by the United States Army Engineer Waterways Experimentation Station for theUnited States Environmental Protection Agency (USEPA) in 1984 The currentversion of the model, Version 3, was released in 1993

The HELP model is essentially a water balance model that includes face water routing It simulates both model cap and liner behavior in a landfillsystem The model is referred to as a quasi-two-dimensional model, as it simulatesvertical flow in barrier and waste layers (assuming unit hydraulic gradient), andhorizontal flow in drainage layers (using an analytical solution of the Boussinesqequation) Calculations are performed on a daily basis, and changes in soilmoisture and surface storage are tracked (Peyton and Schroeder, 1993) The HELPmodel considers both rain and snow infiltration and accounts for interception byvegetation, surface evaporation, and surface storage

subsur-Evapo-transpiration is modeled based on a square root of time calculationand the energy available for evaporation The type and stage of vegetative growth

is also considered in evapo-transpiration calculation

Trang 12

2.2.2.3 UNSAT-H

UNSAT-H (WinUNSAT-H) is a model for calculating water and heat ﬂow inunsaturated media The model was developed at Paciﬁc Northwest NationalLaboratory in Richland, Washington, to assess the water dynamics of near-surface, waste disposal sites The code is primarily used to predict deep drainage

as a function of environmental conditions such as climate, soil type, and tion UNSAT-H is a one-dimensional model that simulates the dynamics processes

vegeta-of inﬁltration, drainage, redistribution, surface evaporation, and uptake vegeta-of waterfrom soil by plants It uses a ﬁnite-difference approximation to solve the one-dimensional vertical form of Richards’ equation, which governs unsaturatedmoisture movement UNSAT-H was designed for use in water balance studiesand has capabilities to estimate evaporation resulting from meteorological surfaceconditions and plant transpiration

The parameters required for each material type are saturated hydraulic ductivity, volumetric moisture content at saturation, irreducible moisture content,air entry head, and inverse pore size distribution index

or an unsaturated soil surface on the basis of atmospheric conditions, vegetationcover, and soil properties and conditions The Penman–Wilson formulation isused to compute the actual rate of evaporation from the soil-atmosphere boundary,which is critical to modeling of evapo-transpirative caps (Wilson, 1990; Wilson

et al., 1994)

The primary features and modeling capabilities of SoilCover are as follows:

• Speciﬁcation of detailed climate data, including minimum and mum air temperature, net radiation, minimum and maximum relativehumidity, and wind speed

maxi-• Speciﬁcation of reduced climate data, including air temperature, tive humidity, and potential evaporation (wind speed is optional)

rela-• Multi-layered soil proﬁles

• Optional speciﬁcation of an internal liquid source/sink node

• Optional specification of oxygen diffusion coefficients for monitoringoxygen flux and the concentration between soil surface and seconduser-specified node

Trang 13

• User-deﬁned or SoilCover-predicted thermal and hydraulic soil erty functions

prop-• Internal adaptive time stepping scheme for daily simulations

• Relative convergence criteria for suction and temperature applied atevery node

• Output data ﬁles providing daily proﬁles of volumetric and gravimetricwater content, degree of saturation, matrix suction, total head, temper-ature, ice content, hydraulic conductivity, oxygen concentration, andvapor pressure

• Daily reporting of potential evaporation, surface flux, base flux, totalevaporation, total run off, root flux, user-selected internal node fluxand user selectable on-screen graphics during program execution show-ing continuous daily or cumulative fluxes in chart and table formatplus daily updates of temperature and degree of saturation profilesThe program user interface occurs in Microsoft Excel using dialogue boxesand custom menus, and the solver is a 32-bit Fortran executable file

simu-to account for water uptake by plant roots The heat transport equations considertransport due to conduction and convection with flowing water The solute trans-port equations consider convective-dispersive transport in the liquid phase, aswell as diffusion in the gaseous phase The transport equations also includeprovisions for nonlinear nonequilibrium reactions between the solid and liquidphases, linear equilibrium reactions between the liquid and gaseous phases, zero-order production, and two first-order degradation reactions: one independent ofother solutes and one that provides coupling between solutes involved in thesequential first-order decay reactions

The user interface includes data pre-processing and graphical presentation ofthe output results in the Microsoft Windows 95, 98, and NT environments Datapre-processing involves specification of a flow region of arbitrary continuousshape by means of lines, arcs and splines, discretization of domain boundaries,and subsequent automatic generation of an unstructured finite element mesh Analternative structured mesh for relatively simple transport domains defined byfour boundary lines can also be considered Graphical presentation of the outputresults consists of simple two-dimensional x–y graphs, contour and spectral maps,velocity vectors, as well as animation of both contour and spectral maps Graphsalong any cross sections or boundaries can be readily obtained A small catalog

of soil hydraulic properties is also part of the interface

Trang 14

2.2.2.6 VADOSE/W

VADOSE/W is a commercially developed two-dimensional ﬁnite element codethat accounts for precipitation; evaporation; snow accumulation/melt/run off;groundwater seepage; freeze-thaw; ground vapor ﬂow; actual transpiration fromplants; and gas diffusion, dissociation, and decay It solves the same primary heatand mass differential equations as the SoilCover model except in two dimensions.The gas diffusion equation is solved at the completion of each time step oncewater contents and temperatures are known throughout the domain

VADOSE/W uses the Penman–Wilson method (Wilson, 1990; Wilson et al.,1994) method for computing actual evaporation at the soil surface such that actualevaporation is computed as a varying function of potential evaporation dependent

on soil pore water pressure and temperature conditions and independent of soiltype and drying history The fully coupled heat and mass equations with vaporﬂow in VADOSE/W permit the necessary parameters at the soil surface to beavailable for use in the Penman–Wilson method VADOSE/W is currently theonly numerical two-dimensional cap design model capable of calculating actualevaporation based on ﬁrst-principle physical relationships, not empirical formu-lations that are developed for unique soil types, soil moisture conditions, orclimate parameters

VADOSE/W can be used wherever accurate surface boundary conditions arerequired Typical applications include designing single or multi-layered soil cov-ers over mine waste and municipal landﬁll disposal sites; obtaining climate-controlled soil pore pressures on natural slopes or man-made covered slopes foruse in stability analysis; and determining inﬁltration and evaporation as well asplant transpiration from agricultural irrigation projects

VADOSE/W comes with a built-in soil property database as well as full-yeardetailed climate data for over 40 sites worldwide Climate data can be easilyscaled to suite speciﬁc conditions or the user can input speciﬁc climate data

2.2.2.7 TOUGH2

Transport of unsaturated groundwater and heat (TOUGH2) is a multi-dimensionalnumerical simulator that simulates the transport of air, water, and heat in porousand fractured media (Pruess, 1991) Mass and energy balances for air, water, andheat are solved simultaneously in TOUGH2 using the integrated finite differencemethod The integrated finite difference formulation of TOUGH2 allows for theconstruction of nonuniform elements that can be used to represent irregulardomains The development of this code was originally motivated by problemsinvolving heat-driven flow, although this code is now used in a wide range ofproblems involving unsaturated flow For example, Ho and Webb (1998) usedTOUGH2 to simulate the effects of heterogeneities on capillary barrier perfor-mance in landfill caps A multi-phase approach was used to describe the move-ment of gaseous and liquid phases, their transport of latent and sensible heat, andphase transitions between liquid and vapor Water vapor and air, which generally

Trang 15

constitute the gas phase, are tracked and simulated separately Liquid- and eous-phase ﬂow can occur under pressure, viscous, and gravity forces according

gas-to Darcy’s Law, and interference between the phases is represented throughrelative permeability functions

A number of variations of the TOUGH2 code have been developed to includeadditional capabilities of modeling additional species, modeling fluctuating atmo-spheric boundary conditions, and inverse modeling The model parameters, initialconditions, and boundary conditions are typically entered into the code throughtext entry into a file that is read by the code Post-processing within TOUGH2 islimited and is typically performed by third-party software The source code forTOUGH2, written in standard FORTRAN77, is available from the United StatesDepartment of Energy (USDOE) Office of Scientific and Technical InformationEnergy Science and Technology Software Center in Oak Ridge, Tennessee

2.2.2.8 FEHM

Finite element heat and mass (FEHM) is a numerical simulation code for surface transport processes (Zyvoloski et al., 1997) It models three-dimensional(3-D), time-dependent, multi-phase, multi-component, nonisothermal, reactiveflow through porous and fractured media It can represent complex 3-D geologicmedia and structures and their effects on subsurface flow and transport FEHMuses a finite-element formulation to solve the governing equations of heat andmass transport Simulation of additional species (e.g., organics, radionuclides)can be performed simultaneously with the solution of heat, air, and water trans-port In addition, a particle-tracking module is also included that provides a morecomputationally efficient procedure to the solution of contaminant transport.Millions of particles can be simulated that represent the effects of advection,diffusion, dispersion, and fracture-matrix interactions on transport

sub-The entry of model parameters, boundary conditions, and initial conditionsinto FEHM is performed through the creation of text ﬁles that are read by thecode FEHM does not perform any direct post-processing of the data for visual-ization, but the user has the option to output the data in formats that can be read

by third-party software FEHM can be obtained free of charge in the United States

2.2.2.9 RAECOM

Radiation attenuation effectiveness and cover optimization with moisture effectsthrough a multi-layer cover (Rogers et al., 1984) Material properties, dimensions,and diffusion coefficients can vary among the different layers, and activity andemanation coefficients can be specified An online calculator that provides the

for most applications via the web site http://ees-www.lanl.gov/EES5/fehm/

http://www.antenna.nl/wise/uranium/ctc.html

(RAECOM) is a code that simulates steady, one-dimensional radon gas diffusion

same functional calculations as RAECOM is provided at the following web site:

Trang 16

2.2.3 M ODELING L IMITATIONS AND R ESEARCH N EEDS FOR C APS

There are many limitations to modeling the performance of caps, including dataneeds; lack of quality assurance and control of models and model usage; and lack

of veriﬁcation, validation, and calibration This section discusses these limitationsand the associated research needs, as well as the role of modeling in designingcaps

2.2.3.1 Role of Modeling

There is often a misperception of what a model can and cannot do It is critical

to get all stakeholders to understand and agree on the objectives of using themodel Many believe that if the predictions arise from a sophisticated computercode that incorporates the fundamental physics as it is currently understood, theanswer must be correct In fact, at best, the model output is a scientiﬁcallydefensible, although not necessarily accurate, prediction of system behavior Thisbelief in modeling leads to the development and use of more sophisticated modelsthat advance the state of the science, but do not necessarily provide more defen-sible predictions

In modeling cover system performance, the objective is to provide a measure

of the ability of the cover to prevent water infiltration to the waste zone over longperiods of time (i.e., tens of years to hundreds of years) It is not possible toprecisely predict infiltration over long time periods due to the large number ofuncontrolled variables (e.g., weather conditions, burrowing animals, root growth),heterogeneities in the physical properties of the system, and lack of preciseunderstanding of the flow physics (e.g., hysteresis effects and soil characteristiccurves are empirical relationships based on data) Therefore, the modelingapproach should aim to demonstrate that the cover system limits infiltration to

an acceptable level over a range of potential conditions This lends itself naturally,although not exclusively, to probabilistic modeling

2.2.3.2 Data Needs

The data required for modeling cap behavior depends on the model being used.The simplest models such as the water balance method of Thornthwaite andMather (1957) and Fenn et al (1975) require monthly climatic data such asprecipitation, mean temperature, heat index, and hours of sunlight Soil types andcap slopes are also required to allow estimation of run off

More comprehensive water balance models such as the HELP model allowfor more complex cap conﬁgurations and, thus, require speciﬁcation of the dif-ferent cap layers The HELP model also simulates the surface processes in greaterdetail and therefore requires additional climatic data and soil properties Theclimate data input to the HELP model include daily precipitation, daily meantemperature, daily solar evaporation, maximum leaf area index, growing season,and evaporative zone depth (Peyton and Schroeder, 1993) The soil properties

Trang 17

required include porosity, ﬁeld capacity, wilting point, hydraulic conductivity,and the United States Soil Conservation Society curve number for the surfacelayer The HELP model contains a list of default soil properties, and a database

of climate data for a large number of North American cities (Peyton andSchroeder, 1993)

Other more rigorous models such as UNSAT-H, HYDRUS-2D, andVADOSE/W simulate unsaturated water flow by solving Richards’ equation.Simulation of unsaturated water flow with Richards’ equation requires parameterspecification of the soil characteristic curves for hydraulic conductivity and mois-ture content as a function of suction pressure, typically represented by empiricalrelationships such as those developed by van Genuchten (1980) or Fredlund andXing (1994) These parameters are required for each unique soil layer in the coversystem Saturated hydraulic conductivity and porosity are also required for eachmaterial Other parameters, such as the air entry pressure head, residual saturationvalue, vertical and horizontal saturated conductivity, and anisotropy parametersmay be required depending on the model

Some of the models (i.e., TOUGH2, FEHM, VADOSE/W, HYDRUS-2D) alsosolve the heat transport equation to track evaporation and water vapor transport.Therefore, these models require additional information regarding soil propertiesrelated to heat transport for the gas and liquid phase Parameters that are typicallyneeded include thermal conductivity, speciﬁc heat capacity, latent heat of vapor-ization, surface tension, and parameters that describe the interactions betweengas and liquids under ﬂowing conditions (e.g., relative permeability)

2.2.3.3 Code Quality Assurance and Quality Control

Numerical models are nothing more than tools that solve mathematical equationsthat cannot be solved with conventional techniques Typical geotechnical com-puter models have thousands of lines of code; it is easy to inadvertently introducemistakes that can cause unpredictable behavior When source code is made avail-able to end users to change and compile, unique versions of the code that onlysolve specific problems commonly result, and the original verification of theoriginal model may not apply to the slightly changed version For this reason,regulatory authorities should consider developing a standard set of benchmarktests that model developers can use to verify and validate their codes If smallchanges to the code are made, all benchmark tests must be resolved to ensurethat no undesirable errors have been introduced Benchmark testing would includesolving some simple steady-state and transient seepage examples using fixedmaterial properties where known solutions for the equation exist to validate thenumerical solution of the code under the most basic conditions More advancedbenchmark tests should be available where individual theoretical components ofthe models could be tested in isolation from other factors (e.g., actual evaporationcan be computed and compared against rigorously controlled laboratory experi-ments) All input data, including material properties, would be listed in the

Trang 18

benchmark test documentation as would the required results If the new model

cannot perform the basic benchmark tests, it is not acceptable for use in ﬁeld

design

A ﬁnal point to consider is the establishment of a group of individuals who

can assemble the benchmark tests and can review and update the tests as new

and more advanced physics are introduced

2.2.3.4 Verification, Validation, and Calibration

The veriﬁcation, validation, and calibration of numerical models are key

compo-nents in the modeling process and are often the most poorly implemented and

misunderstood

The key questions to ask when looking at models are what equations are

being solved, what assumptions have been applied to the equations, and how are

the equations being solved? For example, just because a model computes

evap-oration does not mean that it does so based on sound physical relationships or

that, if it is based on sound physics, the equations are solved properly After a

model user has an understanding of the theory and physics incorporated into a

numerical code, they should satisfy themselves that the numerical solution for

that set of equations is correct This is the veriﬁcation stage of the modeling

process and is usually carried out by the model developer Veriﬁcation has nothing

to do with site data and everything to do with correct solution of the mathematics

Veriﬁcation and validation go together; where veriﬁcation addresses solution

techniques and validation is the process of obtaining conﬁdence that the model

applies to real situations represented by the theoretical formulations applied in

the model Validation tests if the model theories actually apply to speciﬁc real

observations — whether they are laboratory experiments or ﬁeld studies

It is absolutely critical to validate a model based on known closed-form

solutions, known physical observations, and laboratory tests where all parameters

can be controlled and adjusted individually Models cannot be validated using

ﬁeld data alone because there is no direct control over or monitoring of all major

model parameters For example, if a model is validated using site data where

precipitation, run off, change in water storage, and bottom drain ﬂuxes are

mea-sured but actual surface evaporation and transpiration are not meamea-sured, then the

source of discrepancies between measured and computed results cannot be

deter-mined There could be error in the model estimate of evaporation, or there could

be error in the ﬁeld measurement of particular parameters The most appropriate

use for ﬁeld data in modeling is calibration of a previously validated model

Calibration of a model involves making small adjustments to measured or

predicted model input parameters to obtain better matches between measured and

computed results data at more than one instance in time In the ideal case, once

a model is calibrated for a site, it will give reliable results for the same site if

external parameters at that site change For example, if precipitation is doubled

or halved, the change in soil responses can be predicted using a calibrated model

for that site only The problem with calibration is that it only works if the model

Trang 19

physics truly represent the real physical processes in the ground If the model is

rigorous enough and calibrated properly, then all physical processes measured in

the ground and predicted by the model should match Calibration of nonrigorous

models such as HELP must be interpreted with caution because, in many cases,

the calibration can be achieved by adjusting only a single model parameter When

this is done, the predicted and measured data only match for a single instance in

time There is no guarantee that the adjustments made to the model to ﬁt measured

data represent the true physical properties in the ﬁeld It may be possible to

calibrate HELP to match measured percolation data, but it is very unlikely that

parameters such as the temperature, water pressure, water stored in the soil, and

the root depth match ﬁeld conditions at the same instance or at some other instance

in time

2.2.4 U NRESOLVED M ODELING C HALLENGES

There are many challenges facing model developer users These challenges

include the difﬁculties in modeling systems with time-varying properties and

processes, the problems encountered in modeling inﬁltration at arid sites, and the

role of heterogeneities in modeling

2.2.4.1 Time-Varying Material Properties and Processes

A major challenge facing modelers of cap performance is the time-varying nature

of climate, vegetation, and soil properties All models of cap performance require

extensive climatic data, including precipitation, temperature, and solar radiation

to determine inﬁltration and evapo-transpiration Although historic data are

avail-able for many locations, methods for estimating extreme values of these variavail-ables

are not well developed

Physical deterioration of caps is commonplace, as they are easily impacted

by surface and climate processes Changes in vegetation have an effect on run

off generation and evapo-transpiration Establishment of shrubs and trees on caps

can lead to cap penetration by roots, creating high conductivity pathways for water

inﬁltration Similarly, burrowing animals can create high conductivity conduits

through a cap Erosion and subsidence can seriously impact cap performance

The cracking of clay layers in caps due to freeze-thaw cycles or desiccation

(e.g., Albrecht and Benson, 2001) can signiﬁcantly increase the effective

hydrau-lic conductivity of caps, leading to greatly increased water inﬁltration or vapor

escape Albrecht and Benson (2001) found that clay hydraulic conductivities

increased by factors as high as 500 upon desiccation Subsequent resaturation

did not lead to complete healing of dessication-induced cracks Although cap

modeling can predict soil moisture levels in the cap, reliable models for changes

in cap hydraulic properties due to dessication or freeze-thaw have not been

developed

Many caps are expected to provide environmental protection for decades or

centuries Studies of cap stability and soil and geomembrane property stability

Trang 20

over these long periods of time have not been conducted In addition, accurate

predictions of long-term climate changes and the occurrence and impact of

extreme events (e.g., earthquakes, ﬂoods, hurricanes, tornadoes) are not possible

2.2.4.2 Infiltration at Arid Sites

Arid sites are characterized by levels of precipitation that are almost balanced by

loss mechanisms such as evaporation, transpiration, and run off For water balance

models, the recharge is estimated by subtracting the losses from the predicted

production Thus, small errors in either estimate can lead to large errors in

recharge estimates

A second issue at arid sites is that evapo-transpiration models used in the

water balance models for disposal cells that are sparsely vegetated are not accurate

and tend to overpredict evapo-transpiration and underpredict recharge The use

of physically based evapo-transpiration models (e.g., SoilCover, VADOSE/W)

that are formulated to shut down actual evaporation as ground surfaces dry greatly

improves inﬁltration estimates at arid sites

2.2.4.3 Role of Heterogeneities

The most commonly used models for estimating ﬂow through cover systems

assume uniform hydraulic and thermal properties for each layer of the cover

system In practice, local heterogeneities are likely to be responsible for a large

portion of the ﬂow through cover systems The heterogeneities can arise naturally

and their impact on ﬂow does not exist For example, desiccation cracking is

known to occur in clay barriers and leads to increased ﬂow However, the

capa-bility to predict crack formation; the density of cracks; the changes in hydraulic

conductivity that occur due to cracking and subsequently rewetting; and, more

importantly, the change in ﬂow through the layer does not exist

For ﬁeld performance, localized failure will often control inﬁltration through

the cover system This leads to the need to develop procedures to adequately

represent these local failures using gross average properties for the layers

2.3 PRBS

In recent years, PRBs have evolved from the realm of an experimental

method-ology to standard practice for containment and treatment of a variety of

contam-inants in groundwater Like any remedial technology, the decision to use PRBs

is conditioned by the characteristics of the natural system, target contaminants,

and treatment objectives More than 60 sites have implemented this technology

in the last few years to treat chlorinated solvents, fuel hydrocarbons, and various

inorganic contaminants in groundwater As with any technology used to treat or

Currently, the capability to predict the occurrence of local heterogeneities

evolve in time (Section 2.3.5.1)

due to improper construction (e.g., leaks at seams, improper compaction) or

Trang 21

extract contaminants in the subsurface, successful implementation is contingent

on effective site characterization, design, and construction Recent studies on

long-term PRB performance at a number of sites emphasize the following key

issues for successful use of PRBs:

• Performing adequate site characterization on the scale of the

PRB — Site characterization approaches, typical of Resource

Conser-vation and Recovery Act (RCRA) facility investigations (RFIs), are not

adequate Performing additional localized characterization of the

plume distribution in three spatial dimensions and with time,

under-standing the local hydrogeology, and knowing the site geochemistry is

required

• Understanding site hydrology to achieve successful implementation

— PRBs must be located correctly to intercept the plume because once

located in the subsurface, they cannot be moved It is therefore

imper-ative that the PRB captures the plume at the present time and in the

future allowing for variations in ﬂow direction, velocity, and

concen-trations of contaminants over time

• Developing contingency plans for failure to meet design objectives —

It is surprising that site owners and regulators often fail to explicitly

develop contingency plans Contingency plan development requires

speciﬁcation of design criteria and performance objectives and

deter-mination of what constitutes a failure in order to clearly trigger

con-tingency plan activation

2.3.1 F EATURES , E VENTS , AND P ROCESSES A FFECTING

P ERFORMANCE OF PRB S

Design of PRBs requires consideration of groundwater hydraulics, geochemical

processes, and reaction kinetics and the interaction between these processes

2.3.1.1 Groundwater Hydraulics

As with any groundwater remediation technology, an understanding of the

direc-tion and rate of groundwater ﬂow spatially and temporally is critically important

for successful design Groundwater hydraulics are particularly crucial for PRBs

because the treatment system is immovable and passive yet must intercept the

contaminant plume for effective treatment

Groundwater ﬂow is well understood, and groundwater modeling is a mature

technology (e.g., Bear and Verruijt, 1987; Anderson and Woessner, 1992) Many

computer models are available in the public and commercial domains that can

be utilized to perform quantitative predictions of transient 3-D groundwater ﬂow

given appropriate input The key difﬁculty in modeling groundwater hydraulics

is that critical variables that control groundwater ﬂow typically exhibit a high

degree of variability spatially and temporally These variables are difﬁcult to

Trang 22

92 Barrier Systems for Environmental Contaminant Containment & Treatmentcharacterize with precision and sufﬁcient resolution given physical and budgetaryconstraints.

To assess PRB system performance, information is needed on groundwatervelocities through and near the planned PRB On the simplest level, these valuescan be estimated from observed hydraulic gradients and measured or estimatedhydraulic conductivities Alternatively, groundwater velocities can be determinedwith a numerical groundwater ﬂow model based on estimated hydraulic propertydistributions and hydrologic boundary conditions (i.e., water levels and/or ﬂuxes

on model boundaries and recharge and extraction rates), which can vary rally In many cases, it is important to consider the effects of temporal changes

tempo-in ﬂow direction and velocity due to variations tempo-in recharge, pumptempo-ing of adjacentwells, or other disturbances It is not uncommon to observe changes in ﬂowdirection on the order of 30˚ or more over time due to transient boundaryconditions Furthermore, the PRB permeability itself can change markedly overtime in some situations (e.g., due to biological fouling or chemical precipitation

in or near the PRB), which can substantially impact the hydraulic regime.Understanding site stratigraphy and lithology is crucial to understanding andpredicting groundwater hydraulics If a low permeability layer exists at the site,the PRB can be keyed into this layer If one does not exist, then a hanging walldesign can be employed, but uncertainty regarding plume capture may increase

If the site has low permeability layers through which the PRB must be constructed,care must be taken during construction to avoid smearing of such layers, whichcould impact hydraulic contact between the formation and reactive media Athorough understanding of site stratigraphy is important when choosing a partic-ular construction method For example, the use of sheet piling to construct areactive gate may not be a good choice where low permeability layers existbecause of smearing potential

2.3.1.2 Geochemical Processes

The nature and extent of geochemical processes occurring within a PRB to alarge degree determine the long-term treatment performance of the barrier Thedetails of these processes are site speciﬁc and associated with chemical, physical,and biological factors such as the following:

• Reactive media type (e.g., zero-valent iron (ZVI), other metals, zeolite,organic materials)

• Inﬂuent groundwater chemistry (e.g., pH; amounts of cations, anions,and target contaminants)

• Microbiological environment within and around the PRB

• Physical conditions (e.g., temperature)

• The 3-D characteristics of groundwater ﬂow within and near the PRBThere are several good sources that provide information about pilot and full-scale PRB installations worldwide Although new PRBs continue to be deployed,

Trang 23

summaries provided by the Air Force Research Laboratory (AFRL, 2000) and

on the Remediation Technologies Development Forum (RTDF) web sitethat have been installed Of these, the vast majority (approximately 85%) useZVI as the reactive medium Other types of reactive media that have been inves-tigated include other metallic materials (Gillham and O’Hannesin, 1992; Korte

et al., 1995; Muftikian et al., 1995; Orth and McKenzie, 1995; Bostick et al.,1996; Hayes and Marcus, 1997), zeolite (Bowman et al., 2001; Rabideau andVan Benschoten, 2002), various organic materials (Benner et al., 1997), apatite(Conca et al., 2000; Fuller et al., 2002), and sodium dithionite injected as asolution (Fruchter et al., 1997) The AFRL (2000) summarizes different PRBmedia that have been investigated The AFRL (2000) and the RTDF web site alsodocument the range of contaminants that are being treated by PRBs Chlorinatedsolvents such as trichloroethylene (TCE) and perchloroethylene (PCE) are thedominant target contaminants, but others include metals and radionuclides[e.g., Cr(VI), U(VI), Tc(VII)], other inorganics (e.g., NO3, SO42–), and otherorganics (e.g., pesticides, toluene)

Because of the dominance of ZVI as a reactive medium in PRBs, the followingdiscussion focuses exclusively on geochemical processes occurring within it ZVIfunctions as a redox medium and treats contaminants by chemical reduction Atthe same time, the iron is sacriﬁcially oxidized progressively from Fe(0) to Fe2+

and, ﬁnally, Fe3+ The oxidized species of iron potentially can react with othercomponents in the groundwater to precipitate a variety of amorphous and crys-have been formed by reactions occurring in ZVI PRBs

The reaction of groundwater with ZVI causes several major compositionalchanges that drive the formation of these reaction products ZVI begins to dissolveaccording to the following reactions:

2Fe0 + 2H2O + O2 (aq) = 2Fe2+ + 4OH–

Fe0 + 2H2O = Fe2+ + H2 (aq) + 2OH–

The first reaction involves the scavenging of dissolved oxygen by ZVI and isknown to be a fast reaction because column and field studies show the completeabsence of dissolved oxygen within a few centimeters of the influent face of aPRB The second reaction prevails once the oxygen is gone and is slower Bothreactions result in a significant decrease in redox potential and a dramatic rise in

pH, both of which are observed in typical ZVI PRBs The magnitude of change

in pH depends on the detailed chemistry of the inﬂuent groundwater, its bufferingcapacity, and the rate of groundwater ﬂow through the barrier For example, highalkalinity groundwater is more resistant to a change in pH However, the largeavailable mass of ZVI in PRBs tends to overwhelm any redox buffering capacity

of the groundwater

talline phases as described below Table 2.3 lists secondary phases that reportedly(http://www.rtdf.org/public/permbarr/prbsumms/) identify more than 50 PRBs

Trang 24

The oxidation of ZVI (and associated decrease in groundwater redox tial) and the dramatic pH rise are the two principal factors that result in theformation of new solid phases, many of which are iron bearing (Table 2.3) Some

poten-of these phases that contain either Fe2+ (e.g., amorphous ferrous oxyhydroxides,FeS, FeCO3) or mixed Fe2+ and Fe3+ (e.g., Fe3O4, green rust) also are effectivereducing agents for metals, radionuclides, and organics in groundwater Conse-quently, the formation of these reduced iron phases does not necessarily signiﬁ-cantly diminish the reactivity of the barrier media However, not all phases formed

in a PRB are iron-bearing For example, the increase in pH can also lead toprecipitation of various carbonate minerals (e.g., calcite, aragonite) if the influentwater has sufficient amounts of dissolved alkalinity and calcium The mix of solidphases formed and their order of precipitation depend on influent groundwaterchemistry, the complex interplay of changing redox potential and pH in the system

as ZVI dissolves, reaction rates, factors affecting the nucleation of phases, andgroundwater ﬂow rate The ability to predict these reactions and estimate theirOne concern associated with secondary mineral formation in PRBs is thatthese phases passivate the ZVI media, decreasing its reactivity and ability to treatcontaminated groundwater Farrell et al (2000) reported an example of ZVIpassivation with results of long-term column experiments in which they observed

an over six-fold decrease in the reactivity of ZVI to TCE in the two-year experiment

TABLE 2.3

Examples of Precipitated Minerals Found in Fe(0)

Field-Installed PRBs and Column Studies

Iron oxides and oxyhydroxides Goethite ( α-FeOOH)

Akaganeite (β-FeOOH) Lepidocrocite ( γ-FeOOH) (Maghemite (Fe2O3)) Magnetite (Fe3O4) Amorphous iron oxyhydroxides Iron sulﬁdes Mackinawite (Fe9S8)

Amorphous ferrous sulﬁde (FeS) Carbonates Aragonite (CaCO3, orthorhombic)

Calcite (CaCO3, hexagonal) Siderite (FeCO3)

Green Rusts GR-I (CO32– ) (Fe42+ Fe23+ (OH)12)(CO3⋅2H 2 O)

GR-I (Cl – ) (Fe32+ Fe 3+ (OH)8Cl) GR-II (SO42– ) (Fe42+ Fe23+ (OH)12)(SO4⋅2H 2 O)

Source: Liang, L., Sullivan, A.B., West, O.R., Kamolpornwijit, W and Moline,

G.R., 2003 Predicting the precipitation of mineral phases in permeable reactive

barriers Environ Eng Sciences Vol 20(6): p 635.

impact on PRB performance is discussed in Section 2.4.4

Trang 25

The authors found that the degree of passivation was related to the adheringability of secondary minerals and not the overall mass of these phases formed

A number of PRBs have been cored and the media examined to understandthe formation of secondary minerals (e.g., Puls et al., 1999a; Vogan et al., 1999;Phillips et al., 2000; Roh et al., 2000) Typically, cores are obtained by angledrilling through the vertical inﬂuent face of the barrier to provide a cross sectionextending into the PRB interior, capturing the precipitation that is expected to bethe most signiﬁcant at the sediment–ZVI interface Analytical methods such as

X-ray diffraction (XRD) and scanning electron microscopy (SEM) typically are

used to examine the solid phases that have formed Table 2.4 illustrates differences

in groundwater chemistry and resultant secondary minerals observed in PRBs atthe Canadian Forces Base Borden in Ontario, Canada (O’Hannesin and Gillham,1998) and the USDOE Y-12 plant in Oak Ridge, Tennessee (Phillips et al., 2000).The low dissolved solids groundwater at the Borden site has resulted in littleformation of new solid phases over a period of four years, and most precipitation

Goethite Maghemite Amorphous iron oxide Green rust

Mackinawite (Iron media cemented extensively at inﬂuent face) 4040_book.fm Page 95 Wednesday, September 14, 2005 12:43 PM

Trang 26

appears to be restricted to a very thin zone at the inﬂuent face of the PRB Incontrast, the highly mineralized water from the Y-12 plant resulted in much moreextensive formation of secondary phases, illustrated by the cementation of reac-tive media (Figure 2.3) The presence of NO3 at high concentrations in the Y-12plant groundwater is an important factor in determining the degree of reactionoccurring in this PRB because NO3 is readily reduced to NH3 as iron is oxidizedand, therefore, is very corrosive to ZVI

In terms of groundwater treatment, the geochemical reactions between ZVIand the target contaminants are of primary importance within PRBs Currently,most PRBs are deployed to treat groundwater contaminated with chlorinatedsolvents such as TCE, PCE, and their daughter products A discussion of severalpossible degradation reaction pathways for TCE is provided in AFRL (2000) Thefollowing reaction illustrates the overall reductive dechlorination process for TCE:

3Fe0 + C2HCl3 + 3H+ = 3Fe2+ + C2H4 + 3Cl–

As noted in AFRL (2000), there may be a number of reaction pathwaysresulting in a variety of potential intermediates, but experimental and ﬁeld studiesindicate that the net reaction is one of iron oxidation coupled with reductivedechlorination, leading to the production of dissolved ethene (and ethane) andchloride

FIGURE 2.3 Section of cemented core from a PRB at the USDOE Y-12 plant in Oak

Trang 27

PRBs with ZVI also can be used to treat groundwater contaminated withsome redox-sensitive toxic metals For example, dissolved species such as hexava-lent chromium, pertechnetate, and uranyl ions are known to react with ZVI.Examples that conceptually illustrate these reactions are as follows:

Fe0 + 6H+ + CrO42– = Fe3+ + Cr(OH)2 + 2H2O

Fe0 +4H+ + TcO4 = Fe3+ + TcO2 + 2H2O2Fe0 + 3UO22+ = 2Fe3+ + 3UO2The reduction of these metals tends to make them less soluble and less mobilethan the oxidized forms Because these contaminants generally are present insuch low concentrations in groundwater, it has not been possible to identifyspeciﬁc solid phases where they are located within PRBs For example, Phillips

et al (2000) did not observe uranium-bearing phases at the Y-12 PRB However,Fiedor et al (1998), and Gu et al (1998, 2002a) were able to confirm that U(VI)readily reduced to U(IV) in laboratory experiments with ZVI Gu et al (1998)also confirmed that precipitation, not sorption, was the overwhelmingly dominantprocess for immobilizing uranium For the PRB in Elizabeth City, North Carolina,Puls et al (1999b) report that the chromate contaminant in groundwater wasreduced to the chromic (Cr3+) form in an insoluble mixed Cr–Fe hydroxide,although that assumption is based on the sharp decrease in chromium concentra-tions within the PRB rather than characterization of specific chromium-bearingphases Based on their work at this PRB, Mayer et al (2001) also assert thatchromic hydroxide is the likely Cr(III)-bearing phase formed There is no doubtthat ZVI reduces these metals, but whether they are immobilized in the form of

a separate reduced solid state, sorbed in phases such as iron oxyhydroxides, precipitated with other metals, or a combination of all of these processes has notbeen thoroughly studied

co-The ﬁnal geochemical process deserving consideration is related to the tial for contaminant remobilization from an aging PRB For contaminants such

poten-as metals and radionuclides where sorption and/or precipitation poten-associated withreduction is the dominant process occurring, the contaminants can be releasedeventually through ion exchange, desorption, reoxidation, or colloidal transport.Beyond recognition of these possibilities, little formal study of this potential hasbeen performed Of more immediate concern, however, is that some regulatoryagencies may insist on eventual excavation and removal of reactive media whenprecipitation and/or sorption are the dominant processes of contaminant seques-tration Currently, there is no information concerning the risk associated eitherwith leaving the PRB in the ground or excavating it Considering the cost asso-ciated with excavating, transporting, and disposing of spent reactive media, thisdeserves further investigation

Trang 28

2.3.1.3 Reaction Kinetics

Reaction kinetics must be considered for chemical reactions occurring withinPRBs From a design perspective, the target contaminants (and any toxic daughterproducts) must have adequate residence time within the barrier to react sufﬁciently

so that efﬂuent concentrations meet design expectations at the downgradient point

of compliance (POC) Typical values of half-lives for common organic inants with commercial iron tend to range from less than one to approximately

contam-50 hours, depending on the contaminant and the source of iron used Tabulatedsummaries of measured half-lives for many compounds are given by Gillham(1996) and AFRL (2000); however, in the design phase of a PRB, site-speciﬁchalf-lives of target contaminants usually are determined experimentally.Residence times are also partially dependent on the architecture of the ground-water ﬂow system within the PRB Flow heterogeneities (e.g., preferential path-ways) generally exist and can permit target contaminants to migrate through thePRB more rapidly than designed, resulting in inadequate concentration reduction

In addition, ZVI corrosion reactions and precipitation of secondary phases within

a PRB can lead to progressive clogging of the media, resulting in localized flowdiversion Tracer tests have been conducted at several PRBs and illustrate theheterogeneous nature of groundwater flow that exists (e.g., Battelle, 1998).Although the impacts of heterogeneities on groundwater flow and contaminantresidence times in PRBs have received only limited attention, Benner et al (2001)modeled heterogeneous aquifer-barrier systems and provided some insights onhow the effects on preferential pathways and residence time can be minimized

2.3.2 I MPACTS ON D OWNGRADIENT B IODEGRADATION P ROCESSES

At many sites, the rate of natural biodegradation is not sufﬁcient to meet remedialgoals, and intervention is required in the form of additional treatment to accelerate

or enhance the degradation rate ZVI treatment and natural biodegradation arecompatible treatment processes for many chlorinated solvents Both are reductiveprocesses that follow ﬁrst-order reaction kinetics, and both involve the generation

of partially dechlorinated daughter products with reaction rates that are typicallyslower than those of the parent compound Under appropriate circumstances, thetwo treatment processes may be synergistic in that the ZVI treatment can enhance

or accelerate downgradient biodegradation rates by creating geochemical tions more suitable to anaerobic bacterial metabolism A variety of mechanismsmay be operative that stimulate biological processes

condi-2.3.2.1 Enhancement of Geochemical Conditions

Conducive to Anaerobic Biodegradation

ZVI PRBs remove any dissolved oxygen or nitrates present in the upgradientgroundwater The removal of these inorganic electron acceptors lowers the oxi-dation/reduction potential of the groundwater, creating more favorable conditions

Trang 29

for reductive biological processes In addition, many of the organisms involved

in chlorinated solvent biodegradation are obligate anaerobes that cannot survive

in the presence of oxygen

2.3.2.2 Overall Contaminant Concentration Reduction

In cases where a PRB does not achieve complete treatment of the parent pounds or reaction daughter products, partial treatment reduces the total loading

com-of chlorinated contaminants in the downgradient aquifer Such incomplete ment can be helpful to the downgradient biological processes in a number of ways

treat-In aquifers that are electron donor limited, the PRB can bring down the tration of chlorinated solvents to a point where they can be fully dechlorinated

concen-by the available electron donor supply

Some chlorinated compounds are known to inhibit reductive dechlorinationprocesses when present above a threshold concentration An example of this isthe observed inhibitory effect of chloroform (CF) on the reductive dechlorination ofchlorinated ethenes (Maymo-Gatell et al., 2001) This effect has been observed at

CF concentrations above a few parts per million 1,1,1-trichloroethane (1,1,1-TCA)and carbon tetrachloride have also been observed to inhibit methanogenesis andthe dechlorination of chloroethenes, although not a severely as CF (Adamson andParkin, 2000) The PRB can create more favorable conditions for dechlorination

by reducing the concentration of such compounds to below the level where theyare inhibitory

A third beneﬁcial effect of incomplete treatment is that the effects of competingelectron acceptors can be reduced or eliminated When mixtures of chlorinatedsolvents are present in groundwater, the dechlorinating bacteria preferentially usethe electron acceptors that yield the most energy for their metabolism Themetabolic energy available from a given half-reaction is expressed as the Gibbsfree energy of reaction, in units of kilojoules per mole The greater the Gibbs freeenergy available from dechlorinating a given compound, the more likely it is thatthe dechlorinating bacteria will preferentially use that compound as an electronacceptor This effect can be signiﬁcant in plumes with mixtures of differentchlorinated compounds

to persist far downgradient of the PRB, the enhanced biological activity in this

Trang 30

100 Barrier Systems for Environmental Contaminant Containment & Treatmenthydrogen-rich zone can have effects that extend further Dechlorinating organisms,

notably Dehalococcoides ethenogenes, are known to be mobile in groundwater

systems and can be carried downgradient with the groundwater ﬂow (Ellis et al.,2000) As such, the hydrogen-rich zone immediately downgradient of the PRBcan act as a robust source of dechlorinating organisms for the downgradientplume However, increased microbial activity can result in PRB biofouling(Gu et al., 2002b)

2.3.2.4 Electron Donor Production

The ZVI reaction products from chlorinated solvent treatment include fully andpartially dechlorinated simple organic compounds that can serve as electrondonors for downgradient biological dechlorination processes Examples includethe production of formate from carbon tetrachloride and ethene from PCE andTCE Similarly, the treatment of 1,1,1-TCA with ZVI yields a signiﬁcant amount

of ethane, with lesser amounts of ethene, cis-2-butene, and 2-butyne (Fennelly

and Roberts, 1998)

Typically, the daughter products of chlorinated solvents treated with ZVI arepartially dechlorinated and therefore less highly oxidized than the parent com-pound Some of these partially reduced daughter products can be used as electrondonors in downgradient biodegradation processes, particularly in aquifers thatare not strongly reduced An example is the conversion of carbon tetrachloride

to dichloromethane (DCM) in a PRB DCM is known to biodegrade rapidly underboth aerobic and anaerobic conditions Under aerobic conditions, DCM can bebiologically oxidized to carbon dioxide and hydrochloric acid Under anaerobicconditions, DCM can be converted to acetate by fermentation (Freedman andGossett, 1991) The generated acetate can then serve as an electron donor Otherexamples of partially dechlorinated compounds that can serve as electron donors

are cis-1,2-dichloroethylene (cis-1,2-DCE) and vinyl chloride, the daughter

prod-ucts of PCE and TCE (Bradley and Chapelle, 2000), and 1,1-DCE, the daughterproduct of 1,1,1-TCA In aerobic aquifers, the biological transformation of fullychlorinated compounds such as PCE or carbon tetrachloride does not occur Undersuch conditions, a PRB can convert these compounds into lesser chlorinatedspecies, which subsequently can be biologically oxidized

2.3.2.5 Direct Addition of Dissolved Organic Carbon

The addition of organic carbon to an aquifer can signiﬁcantly accelerate reductivebiodegradation processes by providing a ready supply of an electron donor In somecases, the introduction of organic carbon to the aquifer may be the sole purpose

of the PRB and no other reactive media is involved Such applications are referred

to as biobarriers or bio-walls and involve the emplacement of a slow release source

of organic carbon such as compost or vegetable oil It is a matter of discussionwhether such treatment should be considered a permeable barrier technology orwhether it should be considered purely in the context of bioremediation

Trang 31

Modeling of Fluid Transport through Barriers 101

More commonly, the introduction of dissolved organic carbon can be anancillary effect from ZVI emplacement, as in the case of biopolymer trenching

or guar-based injection methods such as hydraulic fracturing or jetting Whentrenching is used as the construction method for a PRB, a high-density biode-gradable slurry is often used to hold the trench open during excavation andemplacement of the reactive media Typically, the slurry-ﬁlled trench is ﬁlledwith the granular iron or sand/iron mixture from the bottom up using a tremie.After the reactive media is emplaced, the biopolymer slurry is typically brokenwith an enzyme that converts it into simple soluble sugars Complete breaking

of the slurry is essential to ensure that the completed PRB will have the desiredpermeability The resultant simple sugars are then dissolved in the groundwaterand carried downgradient in the PRB efﬂuent

Guar is used to suspend the granular iron during injection-based PRBemplacement methods such as jetting and hydraulic fracturing Guar is a highlysoluble food-grade starch which, when chemically cross-linked, forms a highlyviscous gel This viscous gel serves as a carrier ﬂuid for the iron during the jetting

or fracturing process As in the case of biopolymer slurry trenching, a breakerenzyme is added to the gel/iron slurry as it is injected into the subsurface, resulting

in the transformation of the cross-linked slurry into simple soluble sugars.Dissolved organic carbon introduced as a by-product of PRB construction istransient and completely consumed after a period of time This period of time,however, may last as long as a few years depending on site-speciﬁc conditionssuch as groundwater velocity and the amount of biological activity During thisperiod, the introduced organic carbon can signiﬁcantly impact downgradientgroundwater quality, particularly during the transition period from pre-PRB con-ditions to the new post-PRB steady-state In fact, the organic carbon can shortenthe period of time necessary to reach steady-state by accelerating biodegradationand consequently accelerating the rate of desorption in the downgradient aquifer

2.3.3 PRB S YSTEM D YNAMICS

After a PRB is installed, treated groundwater begins to displace untreated water in the downgradient aquifer Notwithstanding this inﬂux of treated water,contaminants continue to be present in the downgradient aquifer for some timeafter PRB installation, largely due to the slow desorption of contaminants fromthe aquifer solids This desorption occurs as the downgradient aquifer transitionsinto a new equilibrium with the treated PRB efﬂuent Eventually, the reservoir

ground-of sorbed contaminants is depleted and downgradient contaminant concentrationsare no longer replenished by the dissolution of sorbed contaminants This processcan take several years, depending on site-speciﬁc factors such as aquifer grain

size, fraction of organic carbon (FOC), groundwater velocity, and initial

contam-inant concentrations

The rate of desorption is a function of the relative amounts of the compound

in the dissolved phase and the sorbed phase Like dissolution, the rate of tion is largely driven by the concentration gradient Desorption rate can be

desorp-4040_book.fm Page 101 Wednesday, September 14, 2005 12:43 PM

Trang 32

102 Barrier Systems for Environmental Contaminant Containment & Treatment

modeled using the Langmuir isotherm and is often represented as a graph ofdissolved phase concentration vs time In groundwater systems, time is equivalent

to the number of pore volumes of clean water that have passed through the aquifermaterial A typical desorption curve is shown in Figure 2.4

Sorbed phase contaminants are largely unavailable for biodegradation Ascontaminants desorb from aquifer solids and re-enter the dissolved phase, theybecome available to dechlorinating organisms in the downgradient aquifer Bio-degradation can accelerate the desorption process by suppressing contaminantconcentrations in the dissolved phase Downgradient biodegradation processescan be accelerated by the presence of organic carbon from PRB construction andother processes discussed later

Both the desorption of parent compounds and the subsequent biodegradation

of these compounds into fully or partially dechlorinated daughter compoundsimpacts downgradient groundwater quality until steady-state is reached Datafrom groundwater monitoring during this transient period can be confusing andcan lead to erroneous conclusions about the performance of the PRB itself Forexample, the persistent presence of parent compounds such as TCE or PCE due

to desorption in the downgradient aquifer can be interpreted as breakthrough orleakage of these compounds through the PRB because of faulty construction such

as PRB holes or gaps Similarly, the presence of cis-1,2-DCE or vinyl chloride

resulting from partial biodegradation of desorbed PCE or TCE can be incorrectlyinterpreted as daughter products in the PRB efﬂuent resulting from insufﬁcientresidence time in the reactive treatment zone It is therefore useful to model these

FIGURE 2.4 Representative desorption curve.

Theoretical Desorption Curve

Trang 33

processes to develop an estimate of the amount of time needed for the aquifer toreach steady-state at a given distance downgradient of the PRB

Figure 2.5 shows a conceptual model of a treatment train consisting of a PRB

in combination with natural biodegradation processes at steady-state The centration of the target constituent (vertical axis) is shown vs distance in the

con-downgradient direction (horizontal axis) C0 represents the initial concentration

in the upgradient area, which decreases in the downgradient direction at theintrinsic rate of natural biodegradation at the site This intrinsic rate can be slow

or negligible at sites where additional treatment has been deemed warranted C t isthe target concentration that needs to be achieved prior to reaching the downgra-

dient POC It is not necessary to design the PRB to achieve C t at the downgradientedge of the PRB, if space is available downgradient of the PRB for biodegradation

to further reduce the concentration prior to reaching the POC C d is the PRBdesign concentration selected to achieve the target concentration at the POC,taking downgradient biodegradation into account

Consider the case of a treatment train consisting of ZVI treatment in a PRBfollowed by natural biological degradation for treatment of a carbon tetrachlorideplume When treated by ZVI, carbon tetrachloride can be completely transformedinto the fully dechlorinated end products carbon monoxide and formate, with apartial yield of trichloromethane (TCM) (approximately 40% on a molar basis).The generated TCM can then be completely transformed into a mixture of fullydechlorinated end products and DCM, which is not further treated by the ZVI.The end result of this reaction series is the complete transformation of carbontetrachloride and TCM, with production of DCM in the amount of approximately

FIGURE 2.5 Conceptual model of chlorinated solvent treatment using a PRB coupled

with natural biodegradation.

Compliance point Permeable

barrier

Design basis

Trang 34

104 Barrier Systems for Environmental Contaminant Containment & Treatment20% of the parent carbon tetrachloride on a molar basis Biodegradation can be

an effective means of treating the generated DCM daughter product, therebyachieving complete dechlorination of the parent carbon tetrachloride to nonchlo-rinated end products prior to reaching the POC DCM is known to be rapidlybiodegradable under a variety of environmental conditions, both aerobic andanaerobic (Cox et al., 1998)

In addition to carbon tetrachloride, the design approach shown conceptuallychlorinated solvents where daughter products are generated that can be fullytransformed in the presence of ZVI, but at a slower rate than the parent com-pounds The slower reaction rates of the daughter products necessitate a longerresidence time in the reactive iron zone to achieve full dechlorination, resulting

in a thicker PRB and correspondingly greater cost Examples of such compoundsinclude PCE and TCE In the case of PCE and TCE, the daughter compounds

cis-1,2-DCE and vinyl chloride are generated as part of the ZVI-driven

degrada-tion reacdegrada-tions Both of these compounds have slower reacdegrada-tion rates than the parent

PCE or TCE The generated vinyl chloride and cis-1,2-DCE can be fully

dechlo-rinated in a ZVI PRB given a sufﬁcient thickness of iron and correspondingresidence time However, signiﬁcant cost savings can be realized by using theavailable aquifer space downgradient of the PRB as a natural bioreactor to degrade

residual cis-1,2-DCE and vinyl chloride rather than increasing PRB thickness.

This approach is only practicable at sites where the biodegradation rates of thesedaughter compounds is sufﬁciently rapid and where sufﬁcient space and residencetime is available downgradient of the PRB prior to reaching the POC

2.3.4 G EOCHEMICAL M ODELING

The discussion in the previous section illustrates that a variety of chemicalreactions occur when groundwater, with a complex mix of cations, anions, andcontaminants, passes through a ZVI PRB One method for evaluating these types

of reactions and predicting their effects is through the use of geochemical models.There are several different kinds of geochemical models that can be applied toPRBs Speciation models evaluate the state of thermodynamic equilibrium ofgroundwater in a static, closed system Reaction path models add to speciationmodeling the capability of considering the step-wise reaction of the water with

a medium such as ZVI These models progressively compensate for groundwaterthat becomes oversaturated with respect to specified phases by allowing them toprecipitate to maintain a state of chemical equilibrium Redissolution of an earlyformed phase also is possible if the changing groundwater composition leads toundersaturation Coupled flow, transport, and reactive transport models potentiallydevelop a more realistic picture of reaction processes in dynamic systems byincorporating groundwater flow, solute transport (by advection, dispersion, anddiffusion), and reaction kinetics into the modeling Inverse or mass balancemodeling differs from other types of geochemical models in that it does notinvolve thermodynamic considerations, but rather attempts to link two related

in Figure 2.5 can be used to optimize the design of PRB remedies for other

Trang 35

groundwater compositions by the dissolution and/or precipitation of phases from

a user-speciﬁed list of candidate phases

There are several general resources that discuss geochemical modeling indetail (Bethke, 1996; Paschke and van der Heijde, 1996; Zhu and Anderson,2002) Those individuals interested in the chemical, mathematical, and numericalmethods behind geochemical models and information about the limitations of thethermodynamic databases commonly used should consult these sources In addi-tion, these sources provide listings and references to many of the most commongeochemical models in use today AFRL (2000), Yabusaki et al (2001), and Mayer

et al (2001) are examples of studies that focus on geochemical modeling as

applied to PRBs with ZVI as the reactive medium The speciﬁc examples in thefollowing discussion are largely summarized from these and several additionalsources

2.3.4.1 Speciation Modeling

Speciation models utilize the composition of groundwater (e.g., concentrations

of dissolved species, pH, redox state) and temperature as input data to examine

a large number of chemical reactions that potentially interrelate the chemicalconstituents of the water These models use the law of mass action to relate thevarious chemical species; identify the state of saturation of mineral phases; anddetermine the distribution of dissolved constituents among several different spe-cies (e.g., Ca2+, CaHCO3), including conversion of redox-sensitive species tomore stable forms consistent with the redox state of the system (e.g., NO3 to

NH3) At the heart of speciation models is a database containing the dynamic properties for elements, ions in solution, solid phases, and gases fromwhich the state of groundwater equilibrium is computed Speciation models alsoincorporate corrections for the effects of ionic activity and temperature.The saturation state of a speciﬁc phase in groundwater is determined by the

thermo-saturation index (SI), which is deﬁned by the relationship:

SI = log (IAP/K)

where K is the equilibrium constant of the reaction controlling formation of the phase and IAP is the ion activity product for the reaction (i.e., using the actual

activities of the reaction species in groundwater) A state of equilibrium for the

phase in the system is deﬁned by SI = 0; SI > 0 indicates oversaturation and SI < 0

deﬁnes a condition of undersaturation

In general, speciation models predict some mineral phases will be rated in groundwater; however, many of these phases will never be found pre-cipitating from the water Slow kinetics is one reason Likewise, some dissolved

oversatu-ions are known to inhibit the precipitation of certain phases, and this phenomenon

is not captured by geochemical models Therefore, natural groundwaters quently are in meta-stable equilibrium with respect to some phases For example,

fre-4040_book.fm Page 105 Wednesday, September 14, 2005 12:43 PM

Định dạng
Số trang	71
Dung lượng	1,47 MB