United States Office of Office of Solid Waste EPA/540/4-89/005Environmental Protection Research and and EmergencyAgency Davelopment Response

E P A Ground Water Issue

Performance Evaluations ofPump-and-Treat Remediations

Joseph

One of the most commonly used ground-water remediationtechnologies is to pump contaminated waterto the surfacefortreatment, Evacuating the effectiveness of pump-and-treatrernediations at Superfund sites is an issue identified by theRegional Superfund Ground Water Forum as a concern ofSuperfunddecision-makers. The Forum is a group of ground-water scientists, representing EPA’s Regional SuperfundOffices, organized to exchange up-to-date information relatedto ground-water remediation at Superfund sites.

Recent research has led to a better understanding of thecomplex chemical and physical processes controlling themovement of contaminants through the subsurface, and theabilitytopump such contaminantsto the surface. Understandingthese processes permits the development and use of bettersite characterization technology and the design andimplementation of more effective and efficient site remediationprograms.

This document is an interim product of a research project thatis developing a protocol for evacuating the effectiveness ofground-water remediations. It has been reviewed by membersof EPA’s Robert S. Kerr Environmental Research Laboratory.

For further information contact Marion R. (Dick) Scalf, Chief,Applications and Assistance Branch, RSKERL-Ada, FTS743-2312, or Randall R. Ross, Project Officer, RSKERL-Ada,FTS 743-2355.

Summary

Pump-and-treat remediations are complicated by a variety offactors. Variations in ground-water flow velocities anddirections are imposed on natural systems by remediationwellfields, and these variations complicate attempts to

F. Keely

evaluate the progress of pump-and-treat remediations. Thisis in part because of the tortuosity of the flowlines that aregenerated and the concurrent re-distribution of contaminantpathways that occurs. An important consequence of alteringcontaminant pathways by remediation wellfields is that historicaltrends of contaminant concentrations at local monitoringwells may not be useful for future predictions about thecontaminant plume.

An adequate understanding of the true extent of a contaminationproblem at a site may not be obtained unless the site’sgeologic, hydrologic, chemical, and biological complexitiesare appropriately defined. By extension, optimization of theeffectiveness and efficiency of a pump-and-treat remediationmay be enhanced by the utilization of sophisticated sitecharacterization approaches to provide more complete, site-specific data for use in remediation design and managementefforts.

Introduction

Pump-and-treat remediations of ground-water contaminationare planned or have been initiated at many sites across thecountry. Regulatory responsibilities require that adequateoversight of these remediations be made possible by structuringappropriate monitoring criteria for monitoring and extractionwells. These efforts are nominally directed at answering thequestion: What can be done to show whether a remediationis generating thedesired control of the contamination? Recently,other questions have come to the forefront, brought on by therealization that many pump-and-treat remediations may notfunction as well as has been expected: What can be done todetermine whether the remediation will meet its timelines?and What can be done to determine whether the remediationwill stay within budget?

Superfund Technology Support Centers for Ground Water

Robert S. Kerr EnvironmentalResearch Laboratory

Ada, OK

Conventional wisdom has it that these questions can beanswered by the use of sophisticated data analysis tools,such as computerized mathematical models of ground-waterflow and contaminant transport. Computer models canindeed be used to make predictions about future performance,but such predictions are highly dependent on the quality andcompleteness of the field and laboratory data utilized. Thisis also true of models used for performance evaluations ofpump-and-treat remediations. In most instances an accurateperformance evaluation can be made simply by comparingdata obtained from monitoring wells during rernediation tothedatagenerated priortothe onset of remediation. Historicaltrends of contaminant levels at local monitoring wells areoften not useful for comparisons with data obtained duringthe operation of pump-and-treat remediations. This is aconsequence of complex flow patterns produced locally bythe extraction and injection wells, where previouslythere wasa comparatively simple flow pattern.

Complex ground-waterflow patterns present great technicalchallenges in terms of characterization and manipulation(management) of the associated contaminant transportpathways. In Figure I, for example, waters moving along theflowline that proceeds directly into a pumping well fromupgradient are moving the most rapidly, whereas thosewaters at the limits of the capture zone move much moreslowly. One result is that certain parts of the aquifer areflushed quite well and other parts poorly. Another result ofthe pumpage is that previously uncontaminated portions ofthe aquifer at the outer boundary of the contaminant plumemay become contaminated by the operation of an extractionwell that is located too close to the plume boundary, becausethe flowline pattern extends downgradient of the well.

The latter is not a trivial situation that can be avoided withoutrepercussions by simply locating the extraction well farenough inside the plume boundary so that its flowline patterndoes not extend beyond the edge of the plume. Such actionswould result In very poor cleansing of the aquifer between theextraction well and the plume boundary, because of the

stagnation of flow that occurs downgradient of the well.Detailed field investigations are required during remediationto determine the locations of the various flowlines generatedbyapump-and-treat operation. Consequently, there maybea need for more data to be generated during the site remediation(especially inside the contamination plume boundaries) thanwere generated during the site investigation, and forinterpretations of those data to require highly sophisticatedtools. For most settings, it is Iikelythat interpretations of thedata that are collected during a pump-and-treat remediationwill require the use of mathematical and statistical models toorganize and analyze those data.

Contaminant Behavior and Plume Dynamics

Ground water flows from recharge zones to discharge zonesin response to the hydraulic gradient (the drop in hydraulicpressure) along that path. The hydraulic gradient may beobtained from water-level elevation contours for ground waterthat has constant fluid density, but it must be obtained fromwater pressure contours when the fluid density varies. Thisis because hydraulic pressure is created by the combinedeffects of elevation, fluid density, and gravity. Additions to thedissolved solids content of a fluid increase its density. Forexample, synthetic seawater can be prepared by addingmineral salts to fresh water. Landfill Ieachate is often so ladenwith dissolved contaminants that its density approaches thatof seawater.

As ground water flows through the subsurface It may dissolvesome of the materials it contacts and may also transportviruses and small bacteria. This gives rise to natural waterquality -- a combined chemical, biological, and physical statethat may, or may not, be suitable for man’s uses. Brines andbrackish waters are examples of natural ground waters thatare unsuitable for man’s use. It is this same power of waterto solubilize minerals and decayed plant and animal residuesthat causes contamination when ground water Is brought intocontact with manmade solids and liquids (Figure 2). Oncecontaminated, ground water also provides a medium for

.

. ,

F

Figure 1. Flowline pattern generatad by an extraction well.Ground-water within the bold line will be captured by the well.Prior to pumping, the flowlines were straight diagonals.

Figure 2. Above-ground spill of chemicals from storage drums.Spilled fluids initiaily fill the upper most soil pores. As much as halfof the fluids remain in each pore after drainage.

2

potentially destructive interactions between contaminantsand subsurface formations, such as the dissolution of limestonearid dolomite strata by acidic wastewaters. Contaminatedground water is a major focus of many hazardous waste sitecleanups, At these sites, a large number of EPA’s Records ofDecision (ROD’s) call for pump-and-treat remediations.

The mechanism by which a source introduces contaminantsto ground water has a profound effect on the duration andareal extent of the resulting contamination. Above-groundspills (Figure 2) are commonly attenuated over short distancesby the moisture retention capacity of surface soils. Bycontrast, there is much less opportunityfor attenuation whenthe contaminant is introduced below the surface, such asoccurs through leaking underground storage tanks, injectionwells, and septic tanks.

The hydraulic impacts of some sources of ground-watercontamination, especially injection wells and surfaceimpoundments, may impart a strongly three-dimensionalcharacter to local flow directions. The water-table moundingthat takes place beneath surface impoundments (Figure 3),for instance, is often sufficient to reverse ground-water flowdirections locally and commonly results in much deeperpenetration of contaminants into the aquifer than wouldotherwise occur. Interactions with streams and other surfacewater bodies may also impart three-dimensional flowcharacteristics to contaminated ground water (e.g., a losingstream creates local mounding that forces ground-waterflowdownward). In addition, contaminated ground water maymove from one aquifer to another through a leaky aquitard,such as a tight silt Iayer that is sandwiched between two sandor gravel aquifers.

II GREAT HYDRAULIC IMPACTSFROM HIGH RATES OF RELEASE

Figure 3. Hydraulic impacts of contarninant sources. Injectionwells and surface impoundments may release fluids at a high rate,resulting in local mounding of the water table.

As ground water moves, contaminants are transported byadvection and dispersion (Figure 4). Advection, or velocity,estimates can be obtained from Darcy’s Law, which statesthat the amount of water flowing through porous sedimentsin a given period of time is found by multiplying together

values of the hydraulic conductivity of the sediments, thecross-sectional area through which flow occurs, and thehydraulic gradient along the flowpath through the sediments.The hydraulic conductivities of subsurface sediments varyconsiderably over small distances. It is primarily this spatialvariability in hydraulic conductivity that results in a correspondingdistribution of flow velocities and contaminant transport rates.

travel by advection alone

additionel spreading caused by dispersion

Figure 4. Bird’s-eye view of contarninant plume spreading. Advectioncauses the majority of plume spreading in most cases Dispersion addsonly marginally to the spreading.

The plume spreading effects of spatially variable velocitiescan be confused with hydrodynamic dispersion (Figure 5), Ifthe details of the velocity distribution are not adequatelyknown. Hydrodynamic dispersion results from the combinationof mechanical and chemical phenomena at the microscopiclevel.The mechanical component of dispersion derives from velocity

Figure 5. Cross-sectional view of contaminant spreading.Permeability differences between strata cause comparabledifferences in advection and, hence, plume spreading.

3

variations among water molecules traveling through thepores of subsurface sediments (e.g., the water moleculesthat wet the surfaces of the grains that bound each pore movelittle or not at all, whereas water molecules passing throughthe center of each pore move most rapidly) and from thebranching of flow into the accessible pores around each grain.By contrast, the chemical component of dispersion is theresult of molecular diffusion. At modest ground-water flowvelocities, the chemical (or diffusive) component of dispersionis negligible and the mechanical component creates a smallamount of spreading about the velocity distribution. At veryslow ground-waterflow velocities, such as occur in clays andsilts, the mechanical component of dispersion is negligibleand contaminant spreading occurs primarily by moleculardiffusion.

In some geologic settings, most of the ground-water flowoccurs through fractures in low permeability rock formations.The flow in the fractures often responds quickly to rainfallevents and other fluid inputs, whereas the flow through thebulk matrix of the rock is extremely slow -- so slow thatcontaminant movement by molecular diffusion may be muchquicker by comparison. On the other end of the ground-waterfiow velocity spectrum is the flow in karst aquifers, since it mayoccur mostly through large channels and caverns. In thesesituations, ground-water flow is often turbulent, and theadvection and dispersion of dissolved contaminants are notadequately describable by Darcy’s Law and other porousmedia concepts. Dye tracers have been used to studycontaminant transport in fractured rock and karst aquifers,but such studies have yet to yield relationships that can betransferred from the study site to other sites.

Regardless of the character of ground-water flow, contaminantsmay not be transported at the same rate as the water itself.Variations in the rate of contaminant movement occur as aresult of sorption, ion-exchange, chemical precipitation, andbiotransformation. The movement of a specific contaminantmay be halted completely by precipitation or biotransformation,because these processes alter the chemical structure of thecontaminant. Unfortunately, the resulting chemical structuremay be more toxic and more mobile than the parent compound,such as in the anaerobic degradation of tetrachloroethene(PCE), which yields, successively, trichloroethene (TCE),dichloroethene (DCE), and monochloromethene (vinyl chloride).

Sorption and ion-exchange (Figure 6), conversely, arecompletelyreversibie processes that release the contaminantunchanged after temporarily holding it on or in the aquifersolids. This effect is commonly termed retardation and isquantified by projecting or measuring the mobility of thecontaminant relative to the average flow velocity of theground water. Projections of retardation effects on themobility of contaminants are baaed on equations that incorporatephysical (e.g., bulk density) and chemical (e.g., partitioncoefficients) attributes of the real system. Direct measurementof the effective mobility of contaminants can be made byobservations of plume composition and spreading overtime.Alternatively, samples of soils or sediments from thecontamination site may be used in laboratory studies todetermine the effective partitioning of contaminants betweenmobile (water) and immobile (solids) phases.

Retardation effects can be short-circuited by facilitated transport,

Figure 6. Contaminant transport facilitated by particles. Sorptionof organics (e.g., PCB’S) or metals (e.g., Pb) onto particles maybe effective in increasing their transport.

a term that refers to the combined effects of two or morediscrete physical, chemical, or biological phenomena that actin concert to materially increase the transport of contaminants,Examples of facilitated transport include particle transport,cosolvation, and phase shifting.

Particle transport (Figure 7) involves the movement of colloidalparticles to which contaminants have adhered by sorption,ion-exchange, or other means. Contaminants that otherwiseexhibit moderate to extreme retardation may travel far greaterdistances than projected from their nominal retardation values.Pumping often removes many colloidal particles from thesubsurface. This fact can complicate remediations, and isalso relevant to public water supply concerns.

Figure 7. Retardation of metals by ion exchange. Metal ionscarrying positive charges are attracted to negatively chargedsurfaces, where they may replace existing ions.

4

Cosolvation is the process by which the volubility and mobilityof one contaminant are increased by the presence of another(Figure 8), usually a solvent present at levels of a few percent(note: 1 percent= 10,000 parts per million). Such phenomenaare most likely to occur close to contamination sources,where pure solvents and high dissolved concentrations areoften found.

Figure 8. Conceptualization of transport by cosolvation. Insolublecontaminants may dissolve in ground water that contains solventsat high concentrations.

Those who design treatment strategies should anticipate theneed to remove from ground water certain contaminants thatare normally immobile, if the groundwater is to be extractedin areas that are close to a source of contamination. Thosewho make health risk estimates should attempt to factor in theincreased mobility and exposure potential generated bycosolvation.

Shifts between chemical phases (Figure 9) involve a largechange in the pH or redox (reaction) potential of water, andcan increase contaminant solubilties and mobilities by ionizingneutral compounds, reversing precipitation reactions, formingcomplexes with other chemical species, and limiting bacterialactivity. Phase shifts may occur as the result of biologicaldepletion of the dissolved oxygen normally present in groundwater, or as the result of biological mediation of oxidation-reduction reactions (e.g., oxidation of iron Il to iron Ill). Phaseshifts may also result from raw chemical releases to thesubsurface.

Some ground-water contaminants are conponents of immisciblesolvents, which may be either floaters or sinkers (Figure 10).The floaters generally move along the upper surface of thesaturated zone, although they may depress this surfacelocally, and the sinkers tend to move downward under theinfluence of gravity. Both kinds of immiscible fluids leaveresidual portions trapped in pore spaces by capillary tension.This is particularly troublesome when an extraction well isutilized to control local gradients such that free product(drainable gasoline) flows into its cone of depression.

Figure 9. Facilitated transport by phase diagram shifts, Releasesof acidic contaminants, or depletion of oxygen by biota, maysolubilize precipitated metals or ionize organics.

Figure 10. Dynamics of immiscible floater and sinker plumes.Buoyant plumes migrate Iaterally on top of the water saturatedzone. Dense plumes sink and follow bedrock slopes.

5

The point of concern is that the cone of depression will containtrapped residual gasoline below the water-table (Figure 11).That residual will become a continuous some of contamination,which will persist even when the extraction well is turned off.The extent of the contamination that is generated by theresidual gasoline in the cone of depression may exceed thatgenerated by the gasoline resting in place above the saturatedzone prior to the onset of pumping.

Figure 11. Zone of contaminant residuals caused by pumping.Pumping creates a cone of depression to trap gasoline for removalbut also leaves residues below the water table.

Reliable prediction of the future movement of contaminantplumes under natural flow conditions is difficult because of theneed to evaluate property the many processes that affectcontaminant transport in a particular situation. Remediationevacuations are even more difficult because of extensiveredirection of pre-remediation transport pathways by pump-and-treat wellfields. Hence, to prepare for remediation, it isimportant to determine the potential transport pathways duringthe site investigation.

Monitoring for Remediation PerformanceEvaluations

Ground-water data are collected during rermediations to evaluateprogress towards goals specified in a ROD. The key controlson the quality of these data are the monitoring criteria that areselected and the locations at which those criteria are to beapplied. Ideally, the criteria and the locations would beselected on the basis of a detailed site characterization, fromwhich transport pathways prior to remediation could be identified,and from which the probable pathways during remediationcould be predicted.

The monitoring criteria and locations should also be chosen insuch a way as to provide information on what is happeningboth downgradient of the plume boundary and inside theplume. Monitoring within the plume makes it possible todetermine which parts of the plume are being effectivelyrernediated and how quickly. This facilitates management of

the rernediation wellfield for greatest efficiency; for example,by reducing the flowrates of extraction wells that pump fromrelatively clean zones and increasing the flowrates of extractionwells that pump from zones that are highly contaminated. Bycontrast, the exclusive use of monitoring points downgradientof the plume boundary does not allow one to gain anyunderstanding about the behavior of the plume duringremediation, except to indicate out-of-control conditions whencontaminants are detected.

There are many kinds of monitoring criteria and locations inuse today. The former are divided into three categories:chemical, hydrodynamic, and administrative control. Chemicalcriteria are based on standards reflecting the beneficial usesof ground water (e.g., MCL’S or other health-based standardsfor potential drinking water). Hydrodynamic monitoring criteriaare such things as:

(1) prevention of infiltration through theunsaturated zone,

(2) maintenance of an inward hydraulic gradientat the boundary of a plume of ground-water contamination, and

(3) providing minimum flows in a stream.

Administrative controls maybe codified governmental rulesand regulations, but also include:

(1) effective implementation of drilling bansand other access-limiting administrativeorders,

(2) proof of maintenance of site security, and(3) reporting requirements, such as frequency

and character of operational and post-operational monitoring.

Combinations of chemical, hydrodynamic, and administrativecontrol criteria are generally necessary for specific monitoringpoints, depending on location relative to the source ofcontamination.

Natural Water Quality Monitoring Points

Natural water quality (or “background”) sampling locationsare the most widely used monitoring points, and are usuallypositioned a short distance downgradient of the plume. Theexact location is chosen so that:

(1) it is neither in the plume nor in adjacentareas that may be affected by theremediation,

(2) it is in an uncontaminated portion of theaquifer through which the plume wouldmigrate if the remediation failed, and

(3) its location minimizes the possibility ofdetecting other potential sources ofcontamination (e.g., relevant to the targetsite only).

Data gathered at a natural water quality monitoring pointindicate out-of-control conditions when a portion of the plumeescapes the remedial action. The criteria typically specifiedfor this kind of monitoring point are known natural waterquality concentrations, usually established with water qualitydata from wells located upgradient of the source.

6

Publlc Supply Monitoring Points

Public water supply wells located downgradient of a plume areanother kind of monitoring point. The locations of these pointsare not negotiable; they have been drilled in locations that aresuitable for water supply purposes, and were never intendedto serve as plume monitoring wells. The purpose of samplingthese wells is to assure the quality of water delivered toconsumers, as related to specific contaminants associatedwith the target site. The criteria typically specified for this kindof monitoring point are MCL’S or other health-based standards.

Gradient Control Monitoring Points

A third kind of off-plume monitoring point frequently estabiishedis one for deterrninations of hydraulic gradients. This kind maybe comprised of a cluster of small diameter wells that havevery short screened intevals, and iS usually located justoutside the perimeter of the plume. Water ievel measurementsare obtained from wells that have comparable screenedintervals and are then used to prepare detailed contour mapsfrom which the directions and magnitudes of local horizontalhydraulic gradients can be determined. it is equally importantto evaluate vertical gradients, by comparison of water levelmeasurements from shallow and deeper screened intevals,because a remediation wellfield may control only the uppermostportions of a contaminant plume if remediation wells are tooshallow or have insufficient flow rates.

[Internai] Piume Monitoring Points

Less often utilized is the kind of monitoring point representedby monitoring wells located within the perimeter of the plume.Most of these are installed during the site investigation phase,prior to the remediation, but others maybe added subsequentto implementation of the remediation; they are used to monitorthe progress of the remediation within the plume. These canbe subdivided into on-site plume monitoring points locatedwithin the property boundary of the facility that contains thesource of the contaminant plume, and off-site plume monitoringpoints located beyond the facility boundary, but within theboundary of the contamination plume.

interdependencies of Monitoring Point Criteria

Each kind of monitoring point has a specific and distinct role toplay in evaluating the progress of remediation. The informationgathered is not limited to chemical identities and concentrations,but includes other observable or measurable items that relateto specific remedial activities and their attributes. in choosingspecific locations of monitoring points, and criteria appropriateto those locations, it iS essential to recognize the interdependencyof the criteria for different locations.

In addition to the foregoing, one must decide the following:Should evacuations of monitoring data incorporate allowancesfor statistical variations in the repotted values? if so, thenwhat cut-off (e.g., the average value plus two standard deviations)should be used? Should evaluations consider each monitoringpoint independently or use an average? Finally, what methodshould be used to indicate that the maximum clean-up hasbeen achieved? The zero-slope method, for example, holdsthat one must demonstrate that contaminant levels havestabilized at their lowest values prior to cessation of remediation

. . and that they will remain at that level subsequently, asshown by a fiat (zero-slope) plot of contaminant concentrationsversus time,

Limitations of Pump-and-Treat Remediations

Conventional remediations of ground-water contaminationoften involve continuous operation of an extraction-injectionwellfield. In these remedial actions, the level of contaminationmeasured at monitoring wells maybe dramatically reduced ina moderate period of time, but low levels of contaminationusually persist. In parallel, the contaminant load dischargedby the extraction wellfield declines over time and graduallyapproaches a residual level in the latterstages (Figure 12). Atthat point, large volumes of water are treated to remove smallamounts of contaminants.

Figure 12. Apparent dean-up by pump-and-treat remediation.Contamination concentrations in pumped water decline overtime,to an apparently irreducible level.

Depending on the reserve of contaminants within the aquifer,this may cause a remediation to be continued indefinitely, orit may lead to premature cessation of the remediation andclosure of the site. The latter is particularly troublesomebecause an increase in the Ievel of ground-water contaminationmay follow (Figure 13) if the remediation is discontinued priorto removal of all residual contaminants,

There are several contaminant transport processes that arepotentially responsible for the persistence of residualcontamination and the kind of post-operational effect depictedin Figure 13. To cause such effects, releases of contaminantresiduals must be slow relative to pumpage-induced watermovement through the subsurface.

Transport processes that generate this kind of behaviorduring continuous operation of remediation wellfield include:

(1) diffusion of contaminants in low permeabilitysediments,

(2) hydrodynamic isolation (’dead spots’) withinwellfields,

7

(3) resorption of contaminants from sedimentsurfaces, and

(4) Iiquid-liquid partitioning of immisciblecontaminants.

Figure 13. Contaminant increases after remediation stops. Con-taminant levels may rebound when pump-and-treat operationscease, because of contaminant residuals.

Advection vs. Diffusion

Lccalized variations in the rate of ground-waterflow (advection)arise in heterogeneous settings because of interlayering ofhigh-and low-permeabiiity sediments. When operating aremediation wellfield, these advection variations result inrapid cleansing of the higher permeability sediments, whichconduct virtually all of the flow (Figure 14). By contrast,contaminants are removed from the lower permeabilitysediments very slowly, by diffusion. The specific rate atwhich this diffusive release occurs is dependent on thedifference in contaminant concentrations within and externalto the low permeability sediments.

When the higher permeability sediments are cleaned up, thechemical force drawing contaminants from the lowerpermeability sediments is at its greatest. This force isexhausted only when the chemical concentrations are nearlyequal everywhere.

Low permeability sediments have orders-of-magnitude greatersurface area per volume of material than high permeabilitysediments. Much greater amounts of contaminants may thusaccumulate in a given volume of low permeability sediments,as compared with contaminant accumulations in a like voiumeof high permeability sediments. The thicker the low permeabilltystratum, the more contaminant reserves it can hold, and themore diffusion controls contaminant movement overall. Thusthe majority of contaminant reserves in heterogeneous settingsmay be available only under just such diffusion-controlledconditions.

Figure 14. Permeability variations Iimit remediations. Highpermeability sediments conduct moat of the flow; low permeabilitysediments act as leaky contaminant reservoirs.

The situation iS similar, though reversed, for in-situ remediationsthat require the injection and deliveryof nutrients or reactantsto the zone of intended action: access to contaminants in lowpermeability sediments is restricted to that provided bydiffusion.

Hydrodynamic lsolation

The operation of any wellfield in an aquifer containing movingfluid results in the formation of stagnation zones downgradientof extraction wells and upgradient of injection wells. Thestagnation zones are hydrodynamically isolated from theremainder of the aquifer, so mass transport into or out of theisolated water may occuroniy by diffusion. If remedial actionwells are located within the bounds of a contaminant plume,such as for the removal of contaminant hot-spots, the portionof the plume lying within their associated stagnation zone(s)will not be effectively remediated. The flowline pattern mustbe altered radically, by major changes in the locations ofpumping wells, or by altering the balance of flowrates amongthe existing wells, or both, if the original stagnation zone(s)are to be remediated.

Another form of hydrodynamic isolation is the physicalcreation of enlarged zones of residual hydrocarbon (Figure11). This occurs when deep wells are used to create conesof drawdown into which underground storage tank andpipeline leaks of gasoline can flow, so that skimmer pumpscan remove the accumulated product. When the deep waterpump is turned off, the watertable will rise to its pre-pumpingposition. This will allow the aquifer waters that then fill thecone of depression, and any subsequent ground-waterflowthrough the former cone of depression, to become highlycontaminated with BTEX compounds (benzene, toluene,the ethyl benzenes, and the xylenes) as a result of contactwith the gasoline remaining there on the aquifer solids.Gasoline in residual saturation may occupy as much as 40percent of the pore space of the sediments.

8

Sorption Influences

The number of pore volumes of contaminated water to beremoved during a remediation depends on the sorptivetendencies of the contaminant. The number of pore volumesto be removed also depends on whether ground-waterflowvelocities during remediation are too rapid to allow contaminantlevels to build up to equilibrium concentrations locally (Figure15). If insufficient contact time is allowed, the affected wateris advected away from sorbed contaminant residuals prior toachieving a state of chemical equilibrium and is replaced byfresh water from upgradient.

TIME

Figure 15. Sorption limitation of pump-and-treat remediations.Increased flow velocities caused by pumpage may not allow forsufficient time to reach chemical equilibrium locally.

Hence, continuous operation of pump-and-treat remediationsmay result in steady releases of contaminants at substantiallyless than their chemical equilibrium levels. With lesscontamination being removed per volume of water broughtinto contact with the affected sediments, it is clear that largevolumes of mildly contaminated water are recovered, wheresmall volumes of highly contaminated water would otherwisebe recovered.

Unfortunately, this is all too likely to occur with conventionalpump-and-treat remediations and with those in-situ remediationsthat depend upon injection wells for delivery of nutrients andreactants. This is because ground-waterflow velocities withinwellfields may be many times greater than natural (non-pumping) flow velocities. Depending on the sorptive tendenciesof the contaminant, the time to reach maximum equilibriumconcentrations in the ground water may simply be too great

compared with the average residence time in transit throughthe contaminated sediments.

Liquid-Liquid Partitioning

Subsequent to gravity drainage of free product that has beendischarged to the subsurface, immiscible or non-aqueousphase liquids (NAPL’s) remain trapped in the pores ofsubsurface sediments by surface tension to the grains thatbound the pores. Liquid-liquid partitioning controls thedissolution of NAPL residuals into ground water.

As with sorbing compounds, flow rates during remediationmay be too rapid to allow aqueous saturation Ievel ofpartitioned NAPL residuals to be reached locally (Figure 16).If insufficient contact time is allowed, the affected water isadvected away from the NAPL residuals prior to reachingchemical equilibrium and is replaced by fresh water fromupgradient. “

ADVECTION

GROUND-WATER VELOCITY

Figure 16. Partitioning limits pump-and-treat effectiveness. Lessthan solubility levels of contaminants maybe released from trappedsolvents if pumpage increases flow velocities.

Again, this process generates large volumes of mildlycontaminated water where small volumes of highly contaminatedwater would other wise result, and this means that it will benecessary to pump and treat far more water than wouldotherwise be the case. The efficiency loss Is generally two-fold, because much of the pumped water will contain contaminant

9

concentrations that are below the level at which optimaltreatment is obtained.

Design and Analysis Complications

Contaminant concentrations and ground-waterflow velocitiescan be highly variable throughout the zone of action, which isthat portion of an aquifer actively affected by the remediationwellfield. Consequently, monitoring strategies should befocussed on detection of rapid, sporadic changes in contaminantconcentrations and flow velocities at any specific point in thezone of action. In practice, this means that tracking theeffectiveness of pump-and-treat remediations by chemicalsamplings is quite complicated.

Decisions regarding the frequency and density of chemicalsamplings should take into account the detailed flowpathsgenerated by the remediation wellfield, including changes incontaminant concentrations that result from variations In theinfluences of transport processes along those flowpaths. Theneed to reposition extraction wells occasionally, to remediateportions of the contaminated zone that were previously subjectto slow flowlines, means that the chemical samplings maygenerate results that are not easily understood. It also meansthat it may be necessary to move the chemical monitoringpoints during the course of a remediation.

Evaluations of the hydrodynamic performance of remediationwellfields are also data intensive. For example, It is usuallyrequired that an inward hydraulic gradient be maintained atthe periphery of a contaminant plume undergoing pump-and-treat remediation. This requirement Is imposed to ensure thatno portion of the plume is free to migrate away from the zoneof action. To assess this performance adequately, thehydraulic gradient should be measured accurately in threedimensions between each pair of adjacent pumping or injectionwells. The design of an array of piezometers (small diameterwells with very short screened Intervals, that are used tomeasure the hydraulic head of selected positions in anaquifer) for this purpose is not as simple as one might firstimagine. Many points are needed to define the convolutedwater-table surface that develops between adjacent pumpingor Injection wells. Not only are there velocity divides in thehorizontal dimension near active wells, but in the verticaldimension, too, because the hydraulic Influence of each wellextends to only a limited depth in practical terms.

Innovations in Pump-and-Treat Remediations

One of the promising innovations in pump-and-treat remediationsis pulsed pumping. Pulsed operation of hydraulic systems isthe cycling of extraction or injection wells on and off in activeand resting phases (Figure 17). The resting phase of apulsed-pumping operation can allow sufficient time forcontaminants to diffuse out of low permeability zones and intoadjacent high permeability zones, until maximum concentrationsare achieved In the higher permeability zones. For sorbedcontaminants and NAPL residuals, sufficient time can beallowed for equilibrium concentrations to be reached In localground water. Subsequent to each resting phase, the activephase of the cycle removes the minimum volume ofcontaminated ground water, at the maximum possibleconcentrations, for the most efficient treatment. By occasionallycycling only select wells, stagnation zones may be broughtinto active flowpaths and remediated.

Figure 17. Pulsed pumping removal of residual contaminants.Repetitive removal of pulses of highly contaminated water ensureseffective depletion of contaminant residuals.

Pulsed operation of remediation wellfields incurs certainadditonal costs and concerns that must be compared with itsadvantages forsite-specific applications. During the restingphase of pulsed-pumping cycles, peripheral gradient controlmay be needed to ensure adequate hydrodynamic control ofthe plume. In an ideal situation, peripheral gradient controlwould be unnecessary. Such might be the case where thereare no active wells, major streams, or other significant hydraulicstresses nearby to influence the contaminant plume while theremedial action wellfield is In the resting phase. The plumewould migrate only a few feet during the tens to hundreds ofhours that the system was at rest, and that movement wouldbe rapidly recovered by the much higher flow velocities backtoward the extraction wells during the active phase.

When significant hydraulic stresses are nearby, however,plume movement during the resting phase may be unacceptable.Irrigation or water-supply pumpage, for example, might causeplume movement on the order of several tens of feet per day.it might then be impossible to recover the lost portion of theplume when the active phase of the pulsed-pumping cyclecommences. In such cases, peripheral gradient controlduring the resting phase would be essential. If adequatestorage capacity is available, it may be possible to providegradient control in the resting phase by injection of treatedwaters downgradient of the remediation wellfield. Regardlessof the mechanics of the compensating actions, their capitaland operating expenses must be added to those of theprimary remediation wellfield to determine the complete cost.

Pump-and-treat remediations are underway today thatIncorporate some of the principles of pulsed pumping. Forinstance, pumpage from contaminated bedrock aquifers andother low permeability formations results in Intermittent wellfieldoperations by default; the wells are pumped dry even at lowflow rates. In such cases, the wells are operated on demandwith the help of fluid-level sensors that trigger the onset andcessation of pumpage. This simultaneously accomplishesthe goal of pumping ground water only after it has reached

10

chemical equilibrium, since equilibrium occurs on the sametime frame as the fluid recharge event in low permeabilitysettings. in settings of moderate to high permeability, theonset and cessation of pumpage could be keyed to contaminantconcentration Ievels in the pumped water, independent offlow changes required to maintain proper hydrodynamiccontrol. Peripheral hydrodynamic controls mayor may not benecessary during the resting phase of the cycle.

Other strategies to improve the performance of pump-and-treat remediations include:

(1) scheduling of wellfield operations to satisfysimultaneously hydrodynamic control andcontaminant concentration trends or otherperformance criteria,

(2) repositioning of extraction wells to changemajor transport pathways, and

(3) integration of wellfield operations with othersubsurface technologies, such as barrierwalls that limit plume transport and minimizepumping of fresh water, or infiltration pondsthat maintain saturated flow conditions forflushing contaminants from previouslyunsaturated soils and sediments.

Flexible operation of a mediation wellfield, such as occasionallyturning off individual pumps, allows for some flushing ofstagnant zones. That approach may not baas hydrodynamicallyefficient as one that involves permanently repositioning oradding pumping wells to new locations at various timesduring remediation. Repositioned and new wells requireaccess for drilling, however, and that necessarily precludescapping of the site until after completion of the pump-and-treat operations. The third approach cited above, combiningpump-and-treat with subsurface barrier walls, trenching, orin-situ techniques, ail of which may occur at any time duringremediation, may also require postponement of capping untilcompletion of the remediation.

The foregoing discussion may raise latent fears of lack ofcontrol of the contaminant source, something almost alwaysmitigated by isolation of the contaminated soils and subsoilsthat remain long after manmade containers have been removedfrom the typical site. Fortunately, vacuum extraction ofcontaminated air/vapor from soils and subsoils has recentlyemerged as a potentially effective means of removing volatileorganic compounds (VOC’S). Steam flooding has shownpromise for removal of the more retarded organics, and in-situ chemical fixation techniques are being tested for theisolation of metals wastes.

Vacuum extraction techniques are capable of removingseveral pounds of VOC’S per day, whereas air stripping ofVOC’S from comparable volumes of contaminated groundwater typically results in the removal of only a few grams ofVOC’S per day because VOC’S are so poor insoluble in water.Similarly, steam flooding is an economically attractive meansof concentrating contaminant residuals, as a front leading theinjected body of steam. Steam flooding or chemical fixationhave potential for control of fluid and contaminant movementin the unsaturated zone and should thus be considered apotentially significant addition to the list of source controloptions. in addition, soils engineering and landscape

maintenance techniques can minimize infiltration of rainwaterin the absence of a multilayer RCRA-style cap.

In terms of evacuation of the performance of a remediation,the presence of a multliayer RCRA-styled cap could posemajor limitations. The periodic removal of core samples ofsubsurface solids from the body of the plume and the sourcezone, with subsequent extraction of the chemical residues onthe solids, is the only direct means of evacuating the truemagnitude of the residuals and their depletion rate. Since thismust be done periodically, capping would conflict unlessPostponed until closure of the site. Ifcappingcan be postponedor foregone, great flexibility for management of pump-and-treat remediations (e.g., concurrent operation of a soil vaporextraction wellfield, and sampling of subsoils) can be used toimprove effectiveness.

Modeling as a Performance Evaluation Technique

Subsurface contaminant transport models incorporate a numberof theoretical assumptions about the natural processes governingthe transport and fate of contaminants, in order for solutionsto be made tractable, simplifications are made in applicationsof theory to practical problems. A common simplification forwellfield simulations is to assume that air flow is horizontal, sothat a two-dimensional model can be applied, rather than athree-dimensional model, which is much more difficult tocreate and more expensive to use. Two-dimensional modelrepresentations are obviously not faithful to the true complexitiesof real world pump-and-treat remediations since most ofthese are in settings where three-dimensional flow is the rule.Moreover, most pump-and-treat remediations use partiallypenetrating wells, which effect significant vertical flowcomponents, whereas the two-dimensional models assumethat the remediation wells are screened throughout the entiresaturated thickness of the aquifer, and therefore do not causeupconing of deeper waters.

Besides the errors that stem from simplifying assumptions,applications of mathematical models to the evacuation ofpump-and-treat remediations are also subject to considerableerror where the study site has not been adequately characterized.It is essential to have appropriate field determinations ofnatural process parameters and variables (Figure 18), becausethese determine the validity and usefulness of each modelingattempt. Errors arising from inadequate data are not addressedproperly by mathematical tests such as sensitivity analyses orby the application of stochastic techniques for estimatinguncertainty, contrary to popular beliefs, because such testsand stochastic simulations assume that the underlyingconceptual basis of the model is correct. One cannot properlychange the conceptual basis (e.g., from an isolated aquifer toone that has strong interaction with a stream or anotherunderlying aquifer) without data to justify the change. Thehigh degree of hydrogeological, chemical, and microbiologicalcomplexity typically present in field situations requires site-specific characterization of various natural processes bydetailed field and laboratory investigations.

Hence, both the mathematics that describe models and theparameter inputs to those models should be subjected torigorous quality control procedures. Otherwise, results fromfield applications of models are likely to be quatitatively, aswell as quantitatively, incorrect, if done properly, however,

11

Figure 18. Grid of points for a contaminant transport modal.Known values of water level and other inputs are used to predictconcentration changes at each grid intersection (node).

mathematical modeling may be used to organize vast amountsof disparate data into a sensible framework that will providerealistic appraisals of which parts of a contaminant plume arebeing effectively cleansed, when the remediation will meettarget contaminant reductions, and what to expect in terms ofirreducible contaminant residuals. Models may also be usedto evaluate changes in design or operation, so that the mosteffective and efficient pump-and-treat remediation can beattained.

Other Data Analysis Methods for PerformanceEvacuations

Mathematical models are by no means the only methodsavailable for use in evaluating the performance of pump-and-treat remediations. Two other major fields of analysis arestatistical methods and graphical methods. The potentialpower of statistical methods has been tapped infrequently inground-water contamination investigations, aside from theiruse in quality assurance protocols. The uses are many,however, as shown in Table 1. While data interpretation andpresentation methods vary widely, many site documents lackstatistical evaluations; some also present inappropriatesimplifications of datasets, such as grouping or averagingbroad categories of data, without regard to the statisticalvalidities of those simplifications.

Graphical methods of data presentation and analysis havebeen used heavily in both ground-waterflow problems (e.g.,flowline plots and flownets) and water chemistry problems(e.g., Stiff kite diagrams). Figure I, for example, is a flowlineplot for a single well. From analysis of such plots, it is possibleto estimate the number of pore volumes that will be removedover a set period of time of constant pumpage, at differentlocations in the contaminant plume. Figure 9 is a chemicalphase diagram for iron, which may be used to relate pH andredox measurements to the most stable species of iron.

Figure 19 presents one means of producing readily recognizablepatterns of the major ion composition of a water sample, sothat it maybe differentiated from other water types. At sites

Analysis of Variance (ANOVA) Teohniques

ANOVA techniques maybe used to segregate errors due to chemicalanalyses from those errors that are due to sampling procedures andfrom the intrinsic variability of the contaminant concentrations at eachsampling point.

Oorrelation Coefflcients

Correlation coefficients can be used to provide justification forlumping various chemicals together (e.g., total VOC’S), or for using asingle chemical as a class representative, or to Iink sources by similarchemical behavior.

Regression Equatlons

Regression equations may be used to predict contaminant loadsbased on historical records and supplemental data, and may be used totest cause-and-effect hypotheses about sources and contaminantrelease rates.

Surface Trend Analysis Technique

Surface trend analysis techniques maybe used to identify recurringand intermittent (e.g., seasonal) trends in oontour maps of ground-waterIevels and contaminant distributions, which may be extrapolated tosource locations or future plume trajectories.

Table 1. Statistical methods useful in performance evaluatlons.

of subsurface contamination, the major ion composition oftendiffers greatly from the natural quality of water in adjacentareas. Water quality specialists have used such plots fordecades to differentiate zones of brackish water from zonesof potable water, in studies of saltwater intrusion into coastalaquifers, and in studies of the upconing of saline water fromshales during pumping of overlying sandstone aquifers.

TOTAL

Figure 19. Pie chart of major ions in aground-water sample.The mini-equivalence values of the major ions are computedand plotted to generate patterns specific to the source.

12

Figure 20 illustrates another means of producing readilyrecognizable patterns of the milli-quivalence values of majorcations and anions in aground-water sample. Geochemicalprospectors have used this graphical technique as an aid inthe identification of waters associated with mineral deposits.These graphical presentation techniques have been adaptedrecently to the display of organic chemical contaminants. Forexample, a compound of interest such as trichloroethene(TCE) maybe evaluated interms of its contribution to the totalorganic chemical contamination load, or against other specificcontaminants, so that some differentiation of some contributionsto the overlll plume can be obtained.

(a)

(b)

Figure 20. Stiff diagrams of major ions in two samples.The concentrations of the ions are plotted in the manner shownin (a); the uniqueness of another water type is shown by (b).

Perspectives for Site Characterizations

Concepts pertinent to investigating and predicting the transportand fate of contaminants in the subsurface are evolving.Additional effort devoted to site-specific characterizations ofpreferential pathways of contaminant transport and the naturalprocesses that affect the transport behavior and ultimate fateof contaminants may significantly improve the timeliness andcost-effectiveness of remedial actions at hazardous wastesites.

Characterzation Approaches

To underscore the latter point, it is useful to examine theprincipal activities, benefits, and shortcomings of increasinglysophisticated levels of site characterization approaches:conventional (Table 2), state-of-the-art (Table 3), and state-of-the-science (Table 4). The conventional approach to sitecharacterizations is typified by the description given in Table2.

Each activity of the conventional approach can be accomplishedwith semi-skilled labor and off-the-shelf technology, withmoderate to iow costs. It may not be possible to characterizethoroughly the extent and probable behavior of a subsurfacecontaminant plume with the conventional approach.

Key management uncertainties regarding the degree of healththreat posed by a site, the selection of appropriate remedialaction technologies, and the duration and effectiveness of theremediations all should decrease significantly with theimplementation of more sophisticated site characterizationapproaches.

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Actions Typically Taken

install a few dozen shallow monitoring wellsSample ground-water numerous times for 129+priority pollutantsDefine geology primarily by driller’s logs and drillcuttingsEvaluate local hydrology with water level contourmaps of shallow wellsPossibly obtain soil and core samples for chemicalanalyses

Benefits

Rapid screening of the site problemsCosts of investigation are moderate to lowfield and laboratory techniques used are standardData analysis/interpretation is straightforwardTentative identification of remedial alternatives ispossible

Shortcomings

True extent of site problems maybe misunderstoodSelected remadial alternatives may not be appropriateOptimization of final remediation design may not bepossibleClean-up costs remain unpredictable, tend to excessivelevelsVerification of compliance is uncertain and dlfficult

Table 2. Conventional approach to site characterization.

it will probably cost substantiality more to implement state-of-the-art and state-of-the-science approaches in sitecharacterizations, but the increased value of the informationobtained is likely to generate offsetting cost savings by way ofimprovements in the technical effectiveness and efficiency ofthe site clean-up.

Obviously, it is not possible to test these conceptual relationshipsdirectly, because one cannot carry site characterization andremediation efforts to fruition along each approachsimultaneously. One can infer many of the foregoing discussionpoints, however, by observing the changes in perceptions,decisions, and work plans that occur when more advancedtechniques are brought to bear on a site that has alreadyundergone a conventional Ievel of characterization. The lattersituation is a fairly common occurrence, because many firstattempts at site characterization turn up additional sources ofcontamination or hydrogeologic complexities that were notsuspected when the initial efforts were budgeted.

Hypothetical Example

It is helpful to examine possible scenarios that might resultfrom the different site investigation approaches just outlined.

13

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Recommended Actions

Install depth-specific clusters of monitoring wellsInitially sample for 129+ priority pollutants, be selectivesubsequentlyDefine geology by extensive coring/sediment samplingsEvaluate local hydrology with well clusters andgeohydraulic teatsPerform limited tests on sediment samples (grain size,day content, etc.)Conduct surface geophysical surveys (resistivity,EM, ground-penetrating radar)

Benefits

Conceptual understandings of the site problems aremore completeProspect are improved for optimization of remedialactionsPredictability of remediation effectiveness is increasedClean-up costs maybe lowered, estimates are morereliableVerification of compliance is more soundly based

shortcomings

Characterization costs are higherDetailed understandings of site problems are stilldifficultFull optimization of remediation is still not likelyField tests may create secondary problems (disposalof pumped waters)Demand for specialists is increased, shortage is akey limiting factor

Table 3. State-of-the-art approach to site characaterizatlon.

Figure 21 depicts a hypothetical ground-water and soilcontamination situation located in a mixed residential andlight industry section of a town in the Northeast. As illustrated,there are three major plumes: an acids plume (e.g., fromelectrolytic plating operations), a phenols plume (e.g., from acreosoting operation that used large amounts ofpentachlorophenol), and a volatile organics plume (e.g., fromsolvent storage leaks). in addition, soils onsite are heavilycontaminated in one area with spilled pesticides, and inanother area with spilled transformer oils that containedPCB’S in high concentrations.

The hydrogeologic setting for the hypothetical site is a productivealluviai aquifer composed of an assortment of sands andgravels that are interfingered with clay and silt remnants of oldstreambeds and floodplain deposits. The latter have beencontinually dissected by a central river as the valley maturedover geologic time. The deeper portion of the sediments ishighly permeable and is the zone most heavily used formunicipal and industrial supply wells, whereas the shallowportion of the sediments is only moderately permeable sinceit contains many clay and silt lenses. The predominantground-water flow direction in the deeper zone parallels theriver (which is also parallei to the axis of the valley), except inIocallzed areas around municipal and industrial wellfields.The predominant direction of flow in the shallow zone is

Figure 21. Hyothetical contamination site in an alluvial setting.Ground-water contamination at the site includes a variety of plumes;the setting is complex geologically and hydrologically.

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

Idealized Approach

Assume state-of-the-art as starting pointConduct soil vapor surveys for volatile% fuelsConduct tracer teak and borehole geophysical surveys(neutron and gamma)Conduct karat stream tracing and recharge studies, ifappropriate to the settingConduct bedrock fracture orientation and intercon-nectivity studies, if appropriateDetermine the percent organic carbon and cation exchangecapacity of solidsMeasure redox potential, pH, and dissolved oxygen levelof subsurfaceEvaluate sorption-desorption behavior by laboratorycolumn and batch studiesAssess the potential for biotransformation of specificcompounds

Benefits

Thorough conceptual understandings of site problemsare obtainedFull optimization of the remediation is possiblePredictability of the effectiveness of remediation ismaximizedClean-up cost maybe lowered significantly,estimates are reliableVerification of compliance is assured

Shortcoming

Characterization costs may be much higherFew previous applications of advanced theoriesand methods have been completedfield and laboratory techniques are specializedand are not easily masteredAvallabiiity of specialized equipment is lowNeed for specialism is greatly increased (it maybe the key limitation overall)

Table 4. state-of-the-science approach to site characterization.

14

seasonally dependent, having the strongest component offlow toward the river during periods of low flow in the river, andbeing roughly parallel to the river during periods of high flowin the river.

Strong downward components of flow carry water from theshallow zone to the deeper zone throughout municipal andindustrial weilfields, as well as along the river during periodsof high flow. Slight downward components of flow existelsewhere due to local recharge by infiltrating rainwaters.

Conventional Characterization Scenario

A conventional site characterization would define the horizontalextent of the most mobile, widespread plume. However, itwould provide only a superficial understanding of variations inthe composition of the sediments. An average hydraulicconductivity would be obtained from review of previouslypublished geologic reports and would be assumed to representthe entire aquifer for the purpose of estimating flow rates. Thekind of clean-up that would Iikely result from a conventionalsite investigation is illustrated in Figure 22. The volatileorganics plume would be the most important to remediate,since it is the most mobile, and an extraction system would beinstalled. Extracted fluids would be air-stripped of volatilesand then passed through a treatment plant for removal ofnon-volatile residues, probably by relatively expensive filtrationthrough granular activated carbon.

Extraction wells would be placed along the downgradientboundary of the VOC plume to withdraw contaminated groundwater. A couple of injection wells would be placed upgradientand would be used to return a portion of the extracted andtreated waters to the aquifer. The remainder of the pumpedand treated waters would be discharged to the tributary underan NPDES permit.

Figure 22. Conventional dean-up of the hypothetical site. EW’sare extraction wells. IW’s are injection wells; all are screened atthe same elevation and have identical flowrates.

information obtained from the drilling logs and samples of themonitoring wells wouldbe inadequate to do more than position

all of the screened sections of the remediation wells at thesame shallowdepth. The remediation wellfield would operatefor the amount of time needed to remove a volume of watersomewhat greater than that estimated to reside within thebounds of the zone of contamination, allowing for averageretardation values (from the scientific literature) for contaminantsfound at the site. The PCB-laden soils would be excavatedand sent to an incinerator or approved waste treatment anddisposal facility. The decision makers would have based theirapproval on the presumption that the plume had been adequatelydefined, and that if it had not, that the true magnitude of theproblem does not differ substantially, except for the possibilityof perpetual care.

State-of-the-Art Characterization Scenario

incorporation of some of the more common state-of-the-ailsite investigation techniques, such as pump tests, installationof vertically-separated clusters of monitoring wells (shallow,intermediate, and deep) and river stage monitors, and chemicalanalysis of sediment and soil samples would Iikely result inthe kind of remediation illustrated in Figure 23. Since adetailed understanding of the geology and hydrology wouldbe obtained, optimal selection of well locations, well screenpositions and flowrates (the values in the parentheses inFigure 23, in gallons per minute) for the remediation wellscould be determined, A special program to recover the acidplume and neutralize itwould be instituted. A special programcould also be instituted for the pesticide plume. This approachwould probably lower treatment costs overall, despite theneed for separate treatment trains for the different plumes,because substantially lesser amounts of ground water wouldbe treated with expensive carbon filtration for removal of non-volatile contaminants.

The screened intervals of the extraction wells would beplaced at deeper positions towards the river, if water qualitydata from monitoring well clusters show that the plume ismigrating beneath shallow accumulations of clays and silts to

Figure 23. Moderate state-of-the-art remediatlon. Clusters ofvertically-separated monitoring wails and an aquifer test are usedto tailor the remedy to the site hydrogeology.

15

the deeper, more permeable sediments. All or most of theextracted and treated ground water could be returned to theaquifer through injection wells. The well screens would needto be positioned (e.g., deeper) to avoid diminishing theeffectiveness of nearby extraction wella. As in the conventionalremedy, the remediation weilfield would operate for theamount of time needed to remove a volume of water that isdetermined from average contaminant retardation valuesand the rate of flushing of groundwater residing in the zoneof contamination. The detailed geologic and hydrologicinformation acquired would result in an expectation of morerapid cleansing of portions of the contaminated zone thanothers.

One could conclude that this remediation is optimized to thepoint of providing an effective clean-up, and decision makerswould be reasonably justified in giving their approval. Oneshould note, however, that the efficiency (esp., duration) ofthe remediation maybe less than optimal.

Advanced State-of-the-Art CharacterizationScenario

if ail state-of-the-art investigation too ls were used at the site,there would be an opportunity to evaluate the desirability ofusing a subsurface barrier wall to enhance remediationefforts (Figure 24). The wall would not entomb the plumes,but would limit pumping to contaminated fluids, rather thanhaving the extracted waters diluted with fresh waters, as wastrue of the two previous approaches. The volume to bepumped could be lowered because the barrier wall willincrease the drawdown at each well by hydraulic interferenceeffects, thereby maintaining the same effective hydrodynamiccontrol with less pumping (note the lower flowrates for eachwell in Figure 24). Treatment costs should be less, also,because the pumped waters should contain higherconcentrations of contaminants and treatment efficienciesare often greatest at moderate to high concentrations. soilwashing techniques could be used on the pesticide contaminatedarea minimize future source releases to ground water.

Both the effectiveness and the efficiency of this remediationmight therefore appear to be optimizable, but that is aperception that is based on the presumption that thecontaminants will be released readily. Given the potentiallimitations to pump-and-treat remediations discussed previously,it is doubtful that even this advanced state-of-the-art siteinvestigation precludes further improvement. Much attentioncould be devoted to the chemical and biological peculiarities,just as has been given to the geologyand the hydrology. Forexample, detalledevaluation of sorption or other contaminantretardation processes at this site, rather than the use ofaverage retardation values from the literature, should generateadditional improvements ineffectiveness and efficiencyof theremediation. Likewise, detailed examination of the potentialfor biotransformation would be expected to lead to improvedeffectiveness and efficiency.

State-of-the-Science Characterization Scenario

At the state-of-the-science level of site characterization,tracer tests could be undertaken which would provide goodinformation on the potential for diffusive restrictions in lowpermeability sediments and on anisotropic biases in the flowregime, Sorption behavior of the VOC’S could be evaluatedin part by determining the total organic carbon content ofselected subsurface sediments. Similarly, the cation exchangecapacities of subsurface sediment sampes could be determinedto obtain estimates of release rates and nobilities of toxicmetals. The stabilities of various possible forms of elementsand compounds could be evaluated with measurements ofpH, redox potential, and dissolved oxygen. Contemporaryresearch indicates that the acids plume and the phenolsplume might be better addressed with such measurements(e.g., chemical speciation modeling). Finally, if state-of-the-science findings regarding potential biotransformations couldbe taken advantage of, it might be possible to effect in-situdegradation of the phenols pluma, and remove volatile residuestoo (Figure 25).

Figure 24. Advanced state-of-the-art remediation. Subsurfacebarrier walls or other technologies can be integrated with pumpand-treat operations.

16

Figure 25. State-of-the-science remediafion. Bioremediation andother emerging technologies could be tested and Implemented withreasonable oertainty of effects.

Additional Considerations

The foregoing discussion highlights genetic gains In effectivenessand efficiency of remediation that should be expected bybetter defining ground-water contamination problems andusing that information to develop site-specific solutions.

Because the complexities of the subsurface cannot fully bedelineated even with “state-of-the-science” data collectiontechniques and many of these techniques are not fully developednor widely available at this time, it Is important to proceed withremediation in a phased process so that information gainedfrom initial operation of the system can be incorporated intosuccessive stages of the remedy. Some considerations thatmay help to guide this process include the following:

1. In many cases, it maybe appropriate to initiate aresponse action to contain the contaminant plumebefore the remedial investigation is completed.Containment systems (e.g., gradient control) canoften be designed and implemented with less informationthan required for full remediation. In addition topreventing the contamination from migrating beyondexisting boundaries, this action can provide valuableInformation on aquifer response to pumping.

2. Early actions might also be considered as a way ofobtaining information pertinent to design of the finalremedy. This might consist of installing an extractionsystem in a highly contaminated area and observingthe response of the aquifer and contaminant plumeas the system is operated.

3. The remedy itself might be Implemented In a stagedprocess to optimize system design. Extraction wellsmight be installed incrementally and observed fora period of time to determine their range of influence,This will help to identify appropriate locations foradditional wells and can assure proper sizing ofthe treatment systems as the range of contaminantconcentrations in extracted ground water isconfirmed.

4. In many cases, ground water response actionsshould be initiated even though it is not possibleto assess the restoration time frames or ultimateconcentrations achievable. After the systemshave been operated and monitored over time,it should be possible to better define the finalgoals of the action.

Selected References

Abriola, L.M. and G.F. Pinder. 1985. A Multiphase Approachto the Modeling of Porous Media Contamination by OrganicCompounds. Water Resources Research 21(1):11-18.

Baehr, A, L., G.E. Hoag, and M.C. Marley. 1989. RemovingVolatile Contaminantsf rom the Unsaturated Zone by InducingAdvective Air-Phase Transport. Journal of ContaminantHydrology 4(1 ):1 -26.

Barker, J. F., G.C. Patrick, and D. Major, 1987. NaturalAttenuation of Aromatic Hydrocarbons in a Shallow SandAquifer. Ground Water Monitoring Review 7(1):64-71.

Borden, R., M. Lee, J.M. Thomas, P. Bedient, and C.H. Ward.1989. In Situ Measurement and Numerical Simulation ofOxygen Limited Biotransformation. Ground Water MonitoringReview 9(1):63-91.

Bouchard, D.C., A.L. Wood, M.L. Campbell, P. Nkedi-Kizza,and P.S.C. Rae. 1988. Sorption Nonequllibrium duringSolute Transport. Journal of Contaminant Hydrology2(3):209-223.

Chau, T.S. 1988. Analysis of Sustained Ground-WaterWithdrawals by the Combined Simulation-optimizationApproach. Ground Water 26(4):454-463.

Cheng, Songlin. 1988. Computer Notes - Trilinear DiagramRevisited: Application, Limitation, and an Electronic spreadsheetProgram. Ground Water 26(4):505-510,

Curtis, G. P., P.V. Roberts, and Martin Reinhard. 1986. ANatural Gradient Experiment on Solute Transport in a SandAquifer: 4. Sorption of Organic Solutes and its Influence onMobility. Water Resources Research 22(13):2059-2067.

Enfield, C.G. and G. Bengtsson. 1986. MacromolecularTransport of Hydrophobic Contaminants in AqueousEnvironments. Ground Water 26(l) :64-70.

Faust, C.R. 1985. Transport of Immiscible Fluids Within andBelow the Unsaturated Zone: A Numerical Model. WaterResources Research 21(4):587-596.

Feenstra, S., J.A. Cherry, E.A. Sudicky, and Z. Haq. 1984.Matrix Diffusion Effects on Contaminant Migration from anInjection Well in Fractured Sandstone. Ground Water22(3):307-316.

Flathman, P., D. Jerger, and L. Bottomley. 1989. Remediationof Contaminated Ground Water Using Biological Techniques.Ground Water Monitoring Review 9(1 ):105-1 19.

Gelhar, L.W. 1986. Stochastic Subsurface Hydrology fromTheory to Applications. Water Resources Research 22(9):135S145S.

Goltz, M.N. and P.V. Roberts. 1986. Interpreting OrganicSolute Transport Data from a Field Experiment using PhysicalNonequilibrium Models. Journal of Contaminant Hydrology1 (1/2):77-94.

Goltz, M.N. and P.V. Roberts. 1988. Simulations of PhysicalNonequilibrium Solute Transport Models: Application to aLarge-Scale Field Experiment. J. Contaminant Hydrology3(1):37-64.

Guven, O. and F.J. Molz. 1986. Deterministic and stochasticAnalyses of Dispersion in an Unbounded Stratified PorousMedium. Water Resources Research 22(1 1):1565-1574.

17

Hinchee, R. and H.J. Reisinger. 1987. A Practical Applicationof Multiphase Transport Theory to Ground Water ContaminationProblems. Ground Water Monitoring Review 7(1):84-92.

Hossain, M.A. and M.Y. Corapcioglu. 1988. Modifying theUSGS Solute Transport Computer Model to Predict High-Density Hydrocarbon Migration. Ground Water 26(6):717-723.

Hunt, J. R., N. Sitar, and K.S. Udell. 1988. NonaqueousPhase Liquid Transport and Cleanup: 1. Analysis of Mechanisms.Water Resources Research 24(8):1 247-1258.

Hunt, J. R., N. Sitar, and K.S. Udell. 1988. NonaqueousPhase Liquid Transport and Cleanup: 2. Experimental Studies.Water Resources Research 24(8):1 259-1269.

Jensen, B. K., E. Arvln, and A.T. Gundersen. 1988.Biodegradation of Nitrogen- and Oxygen-Containing AromaticCompounds in Groundwater from an Oil-Contaminated Aquifer.Journal of Contaminant Hydrology 3(1):65-76.

Jorgensen, D.G., T. Gogel, and D.C. Signor. 1982.Determination of Flow in Aquifers Containing Variable-DensityWater. Ground Water Monitoring Review 2(2):40-45.

Keely, J.F. 1984. Optimizing Pumping Strategies forContaminant Studies and Remedial Actions. Ground WaterMonitoring Review 4(3):63-74.

Keely, J. F., M.D. Piwonl, and J.T. Wilson. 1986. EvolvingConcepts of Subsurface Contaminant Transport. JournalWater Pollution Control Federation. 58(5):349-357.

Keely, J. F.and C. F. Tsang. 1983. Velocity Plots and CaptureZones of Pumping Centers for Ground Water Investigations.Ground Water 22(6):701 -714.

Kipp, K. L., K.G. Stollenwerk, and D.B. Grove. 1986.Groundwater Transport of Strontium 90 in a Glacial OutwashEnvironment. Water Resources Research 22(4):519-530.

Konikow, L.F. 1986. Predictive Accuracy of a Ground-WaterModel - Lessons from a Postaudit. Ground Water24(2):173-184.

Kueper, B.H. and E.O. Frind. 1988. An Overviewof ImmiscibleFingering in Porous Media. Journal of Contaminant Hydrology2(2):95-110.

Macalady, D. L., P.G. Tratnyek, and T.J. Grundl. 1986.Abiotic Reduction Reactions of Anthropogenic OrganicChemicals in Anaerobic Systems:A Critical Review. Journalof Contaminant Hydrology 1(1/2):1-28.

Mackay, D. M., W.P. Ball, and M.G. Durant. 1986. Variabilityof Aquifer Sorption Properties in a Field Experiment onGroundwater Transport of Organic Solutes: Methods andPreliminary Results. Journal of Contaminant Hydrology 1(1/2):119-132.

Mackay, D. M., D.L. Freyberg, P.V. Roberts, and J.A. Cherry.1986, A Natural Gradient Experiment on Solute Transport ina Sand Aquifer 1. Approach and Overview of Plume Movement.Water Resources Research 22(1 3):201 7-2029.

Major, D.W., Cl. Mayfield, and J.F. Barker. 1988.Biotransformation of Benzene by Denitrification In AquiferSand. Ground Water 26(l):8-14.

Mercado, Abraham. 1985. The Use of HydrogeochemicalPatterns in Carbonate Sand and Sandstone Aquifers toIdentify Intrusion and Flushing of Saline Water. GroundWater 23(5):635-645.

Molz, F.J., O. Guven, J.G. Melville, and J,F. Keely. 1986.Performance and Analysis of Aquifer Tracer Tests withimplications for Contaminant Transport Modeling. U.S. EPAOffice of Research and Development: EPA/600/2-86-062.

Molz, F.J., O. Guven, J.G. Melville, J.S. Nohrstedt, and J.K.Overholtzer. 1988. Forced-Gradient Tracer Tests and InferredHydraulic Conductivity Distributions at the Mobile Field Site.Ground Water 26(5):570-579.

Molz, F.J., M.A. Widdowson, and L.D. Benefield. 1986.Simulation of Microbial Growth Dynamics Coupled to Nutrientand Oxygen Transport in Porous Media. Water ResourcesResearch 22(8):1207-1216.

Novak, S.A, and Y. Eckstein. 1988. HydrogeochemicalCharacterization of Brines and Identification of BrineContamination in Aquifers. Ground Water 26(3):317-324.

Osiensky, J. L., K.A. Peterson, and R.E. Williams. 1988.Solute Transport Simulation of Aquifer Restoration after InSitu Uranium Mining. Ground Water Monitoring Review8(2):137-144.

Osiensky, J.L., G.V. Winter, and R.E. Williams. 1984. Monitoringand Mathematical Modeling of Contaminated Ground-WaterPlumes in Fluvial Environments. Ground Water 22(3):298-306.

Ophori, D. U., and J. Toth. 1989. Patterns of Ground-WaterChemistry, Ross Creek Basin, Alberta, Canada. GroundWater 27(1):20-26.

Pinder, G.F. and L.M. Abriola. 1986. On the Simulation ofNonaqueous Phase Organic Compounds in the Subsurface.Water Resources Research 22(9):109S-1 19S.

Pollock, D.W. 1988. Semianalytical Computation of PathLines for Finite Difference Models. Ground Water26(6):743-750.

Roberts, P.V., M.N. Goltz, and D.M. Mackay. 1986. ANatural Gradient Experiment on Solute Transport In a SandAquifer: 3. Retardation Estimates and Mass Balances forOrganic Solutes. Water Resources Research 22(13):2047-2058.

Satlin, R.L. and P.B. Bedient. 1988. Effectiveness of VariousAquifer Restoration Schemes Under Variable HydrogeologicConditions. Ground Water 26(4):488-499.

Siegrist, H. and P.L. McCarty. 1987. Column Methodologiesfor Determining Sorption and Biotransformation Potential forChlorinated Aliphatic Compounds in Aquifers. Journal ofContaminant Hydrology 2(1 ):31 -50.

18

Spain, J. C., J.D. Milligan, D.C. Downey, and J.K. Slaughter.1989. Excessive Bacterial Decomposition of H2O2 DuringEnhanced Biodegradation. Ground Water 27(2):163-167.

Spayed, S.E. 1985. Movement of Volatile Organics Througha Fractured Rock Aquifer. Ground Water 23(4):496-502.

Srinivasan, P. and J.W. Mercer. 1988. Simulation ofBiodegradation and Sorption Processes in Ground Water.Ground Water 26(4):475-487.

Staples, CA and S.J. Geiselmann. 1988. Cosolvent Influenceson Organic Solute Retardation Factors. Ground Water26(2):192-198.

Starr, R.C., R.W. Gillham, and E.A. Sudicky. 1985. ExperimentalInvestigation of Solute Transport in Stratified Porous Media:2. The Reactive Case. Water Resources Research 21(7):1043-1050.

Steinhorst, R.K. and R.E. Williams. 1985. Discrimination ofGroundwater Sources Using Cluster Analysis, MANOVA,Canonical Analysis, and Discriminant Analysis. WaterResources Research 21(8):1 149-1156.

Stover, Enos. 1989. Co-produced Ground Water Treatmentand Disposal Options During Hydrocarbon Recovery Operations.Ground Water Monitoring Review 9(1):75-82.

Sudicky, E.A., R.W. Gillham, and E.O. Frind. 1985. ExperimentalInvestigation of Solute Transport in Stratified Porous Media:1. The Nonreactive Case. Water Resources Research21 (7):1 035-1041.

Testa, S. and M. Paczkowski. 1989. Volume Determinationand Recoverability of Free Hydrocarbon. Ground WaterMonitoring Review 9(1):120-128.

Thomsen, K., M. Chaudhry, K. Dovantzis, and R. Riesing.1989. Ground Water Remediation Using an Extraction,Treatment, and Recharge System. Ground Water MonitoringReview 9(1):92-99.

Thorstenson, D.C. and D.W. Pollock. 1989. Gas Transportin Unsaturated Porous Media: The Adequacy of Fick’s Law.Reviews of Geophysics 27(1):61-78.

Usunoff, E.J., and A. Guzman-Guzman. 1989. MultivariateAnalysis in Hydrochemistry: An Example of the Use of Factorand Correspondence Analyses. Ground Water27(l):27-34.

Valocchi, A.J. 1988. Theoretical Analysis of Deviations fromLocal Equilibrium during Sorbing Solute Transport throughideallzed Stratified Aquifers. Journal of Contamination Hydrology2(3):191-208.

Watson, [an. 1984. Contamination Analysis - Flow Nets andthe Mass Transport Equation. Ground Water 22(1):31-37.

Zheng, C,, K.R. Bradbury, and M.P. Anderson. 1988. Roleof Interceptor Ditches in Limiting the Spread of Contaminantsin Ground Water. Ground Water 26(6):734-742.

19