November 2002

Allocating Responsibility for Groundwater Remediation Costs

Barbara J. Graves
Traverse City, Michigan

David Jordan, Dominique Cartron, Daniel B. Stephens
Albuquerque, New Mexico

Micheal A. Francis
Los Angeles, California

Daniel B. Stephens Expert Witness Profile on Experts.com

One of the challenges facing environmental attorneys and their clients is the development of an equitable allocation of responsibility for cleanup costs associated with groundwater contamination from several potentially responsible parties. This article provides a survey of several scientific approaches for the allocation of such responsibility.

Recent allocation processes under the Comprehensive Environmental Response, Compensation and Liability Act,₍₁₎ also referred to as CERCLA and Superfund, have caused the development of several scientific methods for sharing the responsibility for groundwater cleanup costs between potentially responsible parties (PRPs). Originally, the Environmental Protection Agency (EPA) directed the nonbinding allocation of responsibility (NBAR) to encourage settlement by PRPs; now, as part of settlement, allocation processes are commonly managed by PRP groups that may consider a number of approaches including the Gore factors, weighted site attributes, and an analysis of groundwater plume mass, volume, or a combination of mass/volume.₍₂₎ These scientific approaches are designed to be representative of site conditions and contaminant fate and transport so that an equitable allocation can be developed.

A PRP can be held liable for 100 percent
of a cleanup of a disposal site, even though
the disposal of the hazardous substance,
at the time of its disposal, was legal and there
were other contributors to the disposal site.

REGULATORY DRIVERS AND THE ALLOCATION PROCESS

CERCLA was enacted in 1980 in response to the discovery of abandoned and orphan hazardous chemical disposal sites, most notably Love Canal and the Valley of the Drums. Congress was compelled to act because there was no existing federal law to manage and force the cleanup of these problematic waste disposal sites.₍₃₎ Four years earlier, Congress had enacted the Resource Conservation and Recovery Act, often referred to as RCRA.₍₄₎ However, RCRA is a forward-looking statute, which set standards for the management, storage, treatment, and disposal of newly generated hazardous wastes. RCRA did not directly address the historical, preexisting abandoned and/or orphan disposal sites that were being identified across the country.₍₅₎

Consequently, Congress hastily drafted CERCLA to provide a mechanism for the EPA to use to manage and force the cleanup of these abandoned waste sites.₍₆₎ This haste on the part of Congress has made CERCLA a controversial law and spawned years of litigation.₍₇₎ CERCLA has proven to be controversial because, among other things, it established a funding mechanism by which the EPA, state governments, and private parties who have performed remedial work can sue PRPs to recover costs.₍₈₎ CERCLA is a strict liability scheme causing PRPs to sue each other to reduce their alleged responsibility.

CERCLA was amended with the enactment of the Superfund Amendments and Reauthorization Act of 1986 (SARA).₍₉₎ SARA added a provision allowing PRPs who participated in a site cleanup to obtain contribution from other PRPs.₍₁₀₎ SARA also added provisions affecting the EPA's powers (including a rigid legal framework for settlement,₍₁₁₎ restricting the EPA's discretion in remedy selection and imposing complicated cleanup standards,₍₁₂₎ and imposing public participation requirements.₍₁₃₎ Subsequently, the EPA has embarked on administrative reform programs designed to streamline and expedite CERCLA cleanups. These administrative reforms attempt to address some of the problems associated with remedy selection and the requirements that all cleanups meet all applicable or relevant and appropriate legal requirements (ARARs), which hinder the implementation of risk-based cleanup standards. However, these administrative reforms are not consistently applied and may be subject to legal challenges for not being authorized.

The Potentially Responsible Parties

The CERCLA-defined universe of PRPs, which are liable for the response and cleanup actions, comprises four classes of parties: (1) the current owner or operator of a vessel or a facility; (2) any person who, at the time of disposal of any hazardous substance, owned or operated any facility where such hazardous substances were disposed; (3) any person who arranged for disposal or treatment, or arranged with a transporter for transport to disposal or treatment, of hazardous substances owned or possessed by such person at any facility or incineration vessel owned or operated by another party and containing such hazardous substances; and (4) any person who accepts or accepted any hazardous substances for transport to disposal or treatment facilities, incineration vessels or sites selected by such person.₍₁₄₎

As CERCLA has been interpreted, these classes of PRPs, the "owners/operators," "generators," and "transporters," are strictly liable for the cleanup of the waste site.₍₁₅₎ Liability is not only strict, but may be joint and several if the "harm" is indivisible.₍₁₆₎ Thus, a PRP can be held liable for 100 percent of a cleanup of a disposal site, even though the disposal of the hazardous substance, at the time of its disposal, was legal and there were other contributors to the disposal site. The scope of CERCLA-imposed liability extends to (with limited exceptions) all cleanup costs, natural resource damages, and all health assessment and health effects study costs.₍₁₇₎ Congress provided very limited defenses to CERCLA liability.₍₁₈₎ The CERCLA defenses are triggered when the release or threatened release of a hazardous substance, and the resulting damages, were caused solely by an act of God, an act of war, or an act or omission of a third party.

>Of the three CERCLA defenses, the third-party defense is probably the most viable. However, the third-party defense is not available if a contractual relationship exists with the liable third party. The term "contractual relationship" is specifically defined to exclude an "innocent landowner" who has no reason to know of the hazardous substance release.₍₁₉₎

Identified PRPs have few choices other than responding to the EPA's demands. Moreover, the courts have determined that a PRP may not seek judicial review of the EPA's CERCLA enforcement actions until after the demanded response actions are performed by the PRPs.₍₂₀₎ That is, a PRP has no right to immediately challenge a CERCLA action or assertion of the EPA. Rather, the PRP must respond to and act upon the EPA's CERCLA demands and reserve its right to challenge the EPA action or assertion until after completing the CERCLA response actions. If a PRP does not respond to the EPA's demands, it subjects itself to significant further liability. The typical Superfund site involves a landfill, recycling facility, or other similar type of facility called a "waste-in" site, where hazardous substances were brought in and stored, treated, recycled, and/or disposed of. At such sites, records frequently reveal the amount and nature of the hazardous substances brought in by each PRP. Therefore, the allocation of CERCLA liability among the identified and participating PRPs generally focuses on the amount of hazardous substances brought to the site and the nature of such substances. In the settlement process, the viable PRPs may choose to allocate liability among themselves, based upon their relative contributions of the hazardous substances, and adjust the allocation based upon factors such as the contributed substances' toxicity or mobility.

However, other Superfund sites involve regional contamination plumes resulting from a variety of different sources over the years. For example, in the greater Los Angeles area, the EPA has identified two groundwater basins as separate Superfund sites. They are known as the San Fernando Valley Superfund Site (SFVSS) and the San Gabriel Valley Superfund Site (SGVSS). These two sites present rather unique challenges to the PRPs in their respective allocations of liability because, unlike the typical waste-in Superfund site, there are no records of hazardous substances sent to the sites. Instead, the alleged sources of the releases or threatened releases of hazardous substances are current and historical manufacturing, industrial, and commercial sites located in the regions. These alleged sources have, in many cases, operated for many years, if not decades.

The courts have determined that a PRP
may not seek judicial review of the EPA's
CERCLA enforcement actions until after
the demanded response actions are performed by the PRPs.

Moreover, these manufacturing businesses did not, as a matter of course, plan for or intentionally discharge hazardous substances, as opposed to a typical "waste-in" site situation. What occurred, in many cases, were sudden and accidental releases due to spills or equipment failures. In other cases, drips, leaks, or spills of hazardous substances may have occurred over the years. In other words, the businesses did not plan or intend to dispose or discharge hazardous substances from their manufacturing operations. Thus, there is a lack of information about the quantity of a substance that might have been released. Accordingly, the task of allocating cleanup and/or response costs among the PRPs is not as simple as referring to the waste records to allocate shares. In certain operable units, many PRPs simply distribute the CERCLA response costs on a per capita basis to expediently allocate liability on an interim basis. A per capita allocation of future cleanup costs may be unfair because the relative contributions of the PRPs across the regional site and the location of the PRPs, relative to the selected remedy, are different. The PRPs may be spread over large areas across the "sites." These differences have a significant bearing on the cost of the response actions.

PRPs quickly recognize the potential unfairness of a per capita allocation and search for a fair alternative. This search, driven by the EPA's demands upon the PRPs to perform work, usually gives rise to the development of an alternative dispute resolution (ADR) process. These ADR processes can involve mediation, arbitration, or a combination of both mediation and arbitration. Recent experiences reveal that ADR can effectively resolve the difficult allocation issues without costly litigation. However, developing an ADR process that is fair and acceptable to all of the concerned PRPs can be a difficult and costly undertaking. As with any agreement, the PRPs must negotiate competing interests.

PRPs must also clarify the costs that will be allocated. This article focuses on scientific methods for allocating remediation costs. However, other costs typically associated with ADR include past response costs for completion of an EPA remedial investigation (RI) by a PRP group, EPA oversight costs (past and future), and basin-wide monitoring costs. These costs may need to be allocated by a different method than contribution alone or, at the least, different factors than contribution may need to be considered. For instance, a PRP distal to a remedy may incur RI, monitoring, and oversight costs, but it may have only a small volume and mass to remedy.

PRPs, once they have decided to use ADR as a means of allocating shares among themselves, must make many important decisions regarding their ADR. Those decisions, which become part of the PRPs' ADR agreement, include the following: Will they use mediation or arbitration? If the PRPs choose arbitration, will the arbitration decision be binding? Will the PRPs use mediation followed by arbitration if the mediation fails? Will the mediator also be used as the arbitrator? Who will be the mediator or the arbitrator? Will the mediator or the arbitrator have a technical background, a legal background, or will there be a team with a combination of technical and legal expertise? What right to appeal, if any, will there be? Who will be vested with the responsibility of interpreting the scope of the arbitration or mediation agreement? What is the procedure for discovery, if discovery is permitted? What is the procedure for presenting a case to the mediator or arbitrator? Once the PRPs have resolved these issues and completed the ADR agreement, the allocation process can move forward.

A number of scientific methods have been developed in an attempt to provide a fair, consistent, and representative characterization of PRPs' contributions to groundwater contamination at a CERCLA site and, thereby, establish allocations acceptable to the PRPs and/or the selected allocation arbitrators. Some of these scientific methods, discussed in the following section, include nonbinding allocations of responsibility (NBAR), Gore factors, weighted site attributes, and individual plume analysis of mass, volume, or a combination of mass and volume.

ALLOCATION APPROACHES

Nonbinding Preliminary Allocations of Responsibility

NBARs were created as a tool for the EPA to encourage settlement by PRPs.₍₂₁₎ An NBAR is a preliminary allocation of the "percentage of total response costs at a facility."₂₂ Pursuant to the NBAR guidelines, the EPA will allocate all of the response costs to PRPs. Percentages allocated to nonviable or insolvent parties are reallocated to solvent parties. Similarly, any percentages allocated to unknown parties are reallocated to known parties based on volume.₍₂₃₎ An NBAR theoretically promotes settlement among PRPs because it affords the PRP an opportunity to view EPA data on volume contributed and allows the PRPs to offer to implement or finance cleanup.₍₂₄₎ From a practical standpoint, an NBAR allows a PRP to respond to the EPA's demands for response action, knowing what its share of the cost will be. In many cases, though, PRPs are willing to sign up to perform EPA-demanded work without first knowing what their individual exposure will be.

An NBAR allocates responsibility according
to volume of waste contributed, as well as
"settlement factors."

An NBAR is performed by the EPA region where the site/operating unit/facility is located. The EPA regional administrator is responsible for the NBAR, which is conducted by the EPA's technical and legal staff. The regional EPA staff performing the NBAR rely on the documentation available at the time. If a volumetric ranking and waste-in list has been completed, it will be used as part of the NBAR.

Allocation Method Used in NBAR. An NBAR allocates responsibility according to volume of waste contributed, as well as "settlement factors."₍₂₅₎ The first step in an NBAR is to gather the maximum amount of information about waste types and volumes for each generator PRP at a facility. Although certain sites may allow for allocation of specific waste removal costs to a particular PRP, most sites have commingled waste. The EPA will generally base its allocation of waste on volume and type of waste.

Once the initial volume-based allocation is complete, the EPA will adjust the allocation by considering other factors, including ability to pay, toxicity, mobility, inequities, and aggravating factors.₍₂₆₎ At this point, the EPA determines whether nonviable parties or unknown parties have been allocated a percentage of the total. If so, that percentage is reallocated to the other known and financially solvent PRPs.

In the case of multiple owners and operators of a particular site, allocation is based on relative length of ownership of the property.₍₂₇₎ The EPA states that allocations for transporters should also be based on volume and refers to the development of future guidelines to provide more specific detail on the allocation of responsibility for transporters.₍₂₈₎ However, the EPA has issued no additional NBAR guidance since 1987.

Evolution of NBAR. The EPA guidelines make it clear that the EPA will undertake an NBAR solely at the request of the PRPs and then only if it appears the NBAR will promote settlement. Generally, the EPA prefers that the parties agree among themselves about the allocation of costs.

Although the NBAR was created as a mechanism to promote settlement, other approaches such as alternative dispute resolution have proven more successful in reaching settlements. The EPA has been critical of the NBAR because it is resource intensive and does not necessarily result in a successful settlement. The EPA has also been criticized for failing to use the settlement tools made available in the SARA amendments (such as NBAR and mixed-funding agreements).₍₂₉₎ Consequently, the NBAR no longer has a primary role in Superfund settlements, nor do PRPs frequently request that the EPA conduct an NBAR.₍₃₀₎

Gore Factors

The "Gore factors" were proposed as an amendment to CERCLA by Al Gore while he served in Congress.₍₃₁₎ Although Congress did not pass the amendment, the Gore factors have become a framework for addressing allocation issues. The Gore factors consist generally of (1) the ability of the party to distinguish its contribution to the discharge, release, or disposal of a hazardous waste; (2) the volume of waste involved; (3) the toxicity of the waste; (4) the involvement of the party in hazardous waste generation, treatment, storage, and disposal; (5) the waste management practices of the party; and (6) the degree of cooperation with regulatory agencies.

CERCLA § 113(f)(1) provides that courts may "allocate response costs among liable parties using such equitable factors as the court determines are appropriate."₍₃₂₎ Therefore, the courts have frequently used the Gore factors as "equitable factors" for determining how liability for remediation costs is allocated among parties.₍₃₃₎

Use of the Gore factors often results
in dissimilar evaluations between parties
at a Superfund site.

Despite the widespread use of the Gore factors, no clear guidance is provided by the factors for developing an equitable standard for cost allocation among parties. Without clear guidance from the EPA, in addition to a lack of clear judicial approach, use of the Gore factors often results in dissimilar evaluations between parties at a Superfund site. Therefore, the Gore factors may be considered a simple framework or a basic starting point for further assessment and development of an approach that incorporates guidelines that scientifically represent environmental impacts related to each party.

The successful use of the Gore factors is further limited by the variability of technical data available from PRPs. The use of the Gore factors is most effective for characterized sites, especially as this relates to applying the factors of distinguishability, volume, and toxicity of waste. Certainly, a PRP's ability to distinguish its waste from that of other PRPs assumes the existence of a tool or methodology of distinction. For other than "waste-in" sites, distinguishing between wastes is usually based on scientific studies of the fate and transport of the contaminants from source areas. These studies rely on an adequate number of scientific sampling points to characterize the geology and hydrogeology and, frequently, an application of environmental forensics to "fingerprint" contamination sources. Field studies completed by PRPs show considerable variability, often resulting in disparate amounts and quality of technical data. Although this inherently limits all allocation processes, it is particularly troublesome when no unified guidance exists on the application of the Gore factors.

Another issue is the subjective nature of two of the Gore factors: waste management practices and the degree of cooperation with regulatory agencies. These factors are intended to aid PRPs with responsible environmental management practices. Although criteria may be established to evaluate these factors, it is difficult to make distinctions between parties whose responses to these factors fall in the average range. Stellar housekeeping practices along with unquestionably poor practices may be easy to evaluate, but the bulk of a PRP's operations probably fall somewhere in the middle of these extremes. Ranking them becomes a much more subjective and, perhaps, unequal task.

This is also true for the evaluation of the degree of cooperation with regulatory agencies, where a simple and legitimate disagreement over scientific approaches or data could be construed as lack of cooperation.

When [the WSAM] method is applied,
an allocation percentage assignment for each
PRP is developed, based on a set of ranking
criteria applied uniformly to each site.

Weighted Site Attributes

The weighted site attributes model (WSAM) is a semiquantitative approach that considers the relative ranking of sites in an allocation based on site characteristics, such as contaminant of concern (COC) use and handling practices, and soil and groundwater data. This method has some of the same limitations as the Gore factors, but it can be fashioned to a more comprehensive list of attributes for evaluation, and it has a systematic application. When this method is applied, an allocation percentage assignment for each PRP is developed, based on a set of ranking criteria applied uniformly to each site. Given a set of site attributes for each PRP, an equation can be developed that assigns different weights to each site attribute to produce a numerical score for each site. This allocation method may be especially useful in developing a screening-level relative ranking for sites where few site characterization data are available. It can also be used for an initial "rough justice" allocation.

The WSAM can be applied based on an evaluation of site-specific files and records and is usually based on the responses to an agreed-upon set of questions. Typical site attributes include:

• years of COC use
• quantity of COCs purchased
• average COC usage rate
• quality of COC handling and management practices
• history of releases at the site
• depth to groundwater
• permeability of onsite soils
• area of property
• concentration of COCs in soil and/or soil-gas
• probability that onsite groundwater and/or soil contamination originates from an upgradient source
• lowest, highest, and average (geometric mean) groundwater concentration for COCs
• toxicity of COCs used at the site
• equity term, which considers good-faith efforts on the part of the PRP
• adequacy of site characterization
• whether site-specific remediation has been done.

A simplified example, using four site attributes, is presented below. The attributes considered in the example are:

• years of COC use (A)
• depth to groundwater (B)
• soil permeability (C)
• COC toxicity (D).

An equation is then developed that includes a weighting factor and results in a score that is calculated from the four attributes (where w_n are the weighting factors):

[IMAGE CANNOT BE RENDERED]

For example, for factors with a potentially large degree of variability, such as years of COC usage, the logarithm of the value is taken to reduce the variability of the total score and facilitate comparison with other contributing factors. Factors C and D (soil permeability and toxicity) are assigned values corresponding to their relative magnitude. Soil permeability is assigned a 1, 2, or 3, based on whether it is low, medium, or high, respectively. In this simplistic example, it is assumed that only basic geologic data are available, rather than actual measurements of permeability. If actual measurements of permeability are available, this term might be treated differently. Likewise, toxicity is arbitrarily assigned a value of 1, 2, or 3, based on whether it is low, medium, or high, respectively. Tables 1A and 1B on the following page present an example calculation using three hypothetical sites, each with different attributes and three different weightings.

Tables 1A and 1B.
Sample Calculation of Weighting Scheme Method of Allocaiton

A. Four Factors at Three Sites

B. Three Weighting Schemes with Results

From the tables, it is clear that the allocation percentages assigned to a given site are highly dependent on the weighting assigned to each site attribute. In this analysis, using the equation arbitrarily developed as an allocation tool, depending on the choice of weighting factors, the allocation percentages vary by little at site 1 to a factor of nearly 3 at site 2. For remedies costing tens of millions of dollars, any variability is highly significant to the property owners.

The primary advantage of the WSAM method is that it provides a screening-level allocation in the absence of a large degree of site-specific data. The screening-level allocation, typically, allows a relative ranking of sites to be performed. However, a major drawback of the method is that it may produce widely variable results, based on how the scoring equation is set up and how each site attribute is weighted. Variability can be controlled by careful selection of the weighting factors.

Application of Defined Plumes to Allocate Costs

The individual plume approach (IPA) uses observed concentrations to develop stand-alone plumes. It incorporates site history, source location, and hydrogeology for plume development. Where observed data are sparse, additional qualitative support from analytical or numerical modeling may be used to fill in gaps. Allocations for PRPs can then be based on plume volume, mass, or some weighted combination of mass and volume.

Substantial site and regional data must exist to develop individual PRP plumes. These plumes should be based on relevant site and regional data. Plumes developed using the IPA can be verified using an analytical or numerical advection-dispersion model. The IPA is based on a number of assumptions:

• An individual plume can be constructed for each site (that is, groundwater data are available).
• For sites with no groundwater data, a plume can be inferred, based on equilibrium partitioning of soil contamination into groundwater.
• The volume, mass, or weighted sum of the mass and volume associated with each plume can be used to assign a share to each PRP.

Delineation of individual plumes is based on primary principles of contaminant fate and transport. Observed concentrations from on-site and off-site monitor wells, when available, are the preferred sources of data. In areas where plumes from two or more facilities overlap, care should be taken that composite plume concentrations do not exceed observed concentrations at the monitor wells for each site. An evaluation of plume length should be made based on simple travel-time calculations, which consider the observed groundwater flow velocity, as well as existing monitor well data, which constrain an individual plume downgradient of a facility.

Delineation of individual plumes is based
on primary principles of contaminant fate
and transport.

The alignment of a plume emanating from a particular facility is controlled primarily by the local direction of groundwater flow at that facility. Where off-site data are unavailable, qualitative support for plume length can be developed using analytical or numerical models.

In developing individual plumes, it is important to consider the basic hydrogeological and transport principles, which control the impacts that a particular facility may have on underlying groundwater. These include mass intercepted by pumping wells or other remedial activities, heterogeneity in the subsurface, perched aquifers, inadequate site monitoring, overlapping plumes from multiple facilities, orphan plumes, monitor wells' depths, and source location, concentration, and duration. Consideration of all of these factors is important in developing an accurate and reasonable plume for each PRP facility.

Individual plumes can be contoured in a variety of ways. For example, the outer edge of a plume might be defined by a contour that is defined by nondetects, the EPA maximum contaminant level (MCL) of the COC, or some arbitrary cleanup level. The selection of the outer boundary of the plume is important in defining the extent of the plume for the particular allocation issues under consideration.

Often, there is a question of which data to use. Sampling data, typically, are not available from all sampling points for the same date, which forces the analyst to use sampling data from a series of sampling events over a specified time period. Given that some locations will have been sampled several times during any time period, while other locations will have been sampled only once, it is usually necessary to take an average or maximum value of concentration at each sampling location. Taking an average or maximum value of several data introduces some bias and raises additional questions, such as how to treat nondetects (as zero, the detection limit, or half of the detection limit?).

Figure 1.
Concentration Contour Developed by Interpolation

[IMAGE CANNOT BE RENDERED]

Since contouring is a somewhat subjective exercise, bias may be introduced in a variety of ways. For example, the position of the outermost contour of the plume, which defines the full extent of a plume, may be subjectively placed due to lack of data to define the plume's extent, or simply due to misinterpretation of the data. Concentration contours are developed by interpolation between well observation point values of concentration (see Figure 1). The contour can be drawn by a linear fit to the data by eye or by measurement, or simply based on professional judgment or knowledge of the underlying processes which control the transport of contaminants at a particular site.

The most important aspect of contouring is that it forms the basis for allocation by actual impacts to groundwater from individual sites. A clear understanding of the factors that control contaminant distribution is essential to produce accurate contour maps. A common problem in almost any allocation is the issue of commingled plumes of one chemical, such as trichloroethene (TCE). That is, a particular contaminant plume may be made up of TCE from a variety of facilities, and the challenge is to determine what portion of the plume came from a particular facility. This is the essence of any allocation. A key methodology for separating commingled plumes is the use of tracer constituents unique to specific PRPs. Ideally, one would be able to isolate a specific chemical (used only at a particular facility) that may or may not be a contaminant of interest from a regulatory or remediation standpoint. The signature of that chemical would be used to identify the portion of a plume that came from the particular facility. Examples of such signature chemicals include chloride, 1,1-dichloroethene (1,1-DCE), perchlorate, and methyl tertiary-butyl ether (MTBE).

Groundwater is not a two-dimensional medium, so there may be a three-dimensional element to almost any contaminant plume. It is important to evaluate depth-specific data, where available. For example, contamination from an upgradient facility may travel downward as it moves downgradient and not be detected in shallow downgradient monitor wells consistently screened across the water table. Therefore, it is important to consider that deep contamination beneath a facility may actually be from an upgradient source, but evaluation of the available data, with respect to depth, is required to determine this.

Numerical or analytical groundwater flow and contaminant transport models may also be used to develop individual plumes for allocation. A key input parameter to determine is the source term that will be used for each facility. This parameter quantifies the timing and mass of the contaminant that is put into the groundwater during the simulation. The source term for a facility can be estimated using a variety of methods, including using a known concentration history from observed data, an average concentration of observed data, or an estimate of groundwater concentrations calculated from soil concentrations. Another method is to assume a "unit plume" emanating from each site, scale each plume based on observed data, and develop an automated optimization scheme.

Any model used should be calibrated to real data. Calibration is the process of comparing simulated results to observed data. Calibration quantifies how well a model fits the real data. The calibration process proceeds by comparing simulated water levels and contaminant concentrations to observed water levels and contaminant concentrations. A model that is well calibrated should provide reasonable predictions of contaminant levels for areas and time periods where no field data are available.

Models used for allocation are typically either analytical or numerical. An analytical model is an exact solution to a mathematical equation, or set of equations, describing groundwater flow and transport of contaminants. Analytical models are typically relatively straightforward to use, but they are based on a definite set of limiting assumptions, usually characterized by an overly simplistic geometry. Usually, analytical models may only approximate hydrogeologic behavior.

A numerical model can be used to solve the equations of groundwater flow and contaminant transport over an area of interest by dividing the area into grid cells and solving each equation at each grid cell. Numerical models provide much more flexibility in terms of specifying the geometry of the problem, and they approximate hydrogeologic behavior more closely than analytical models.

Groundwater is not a two-dimensional medium,
so there may be a three-dimensional element
to almost any contaminant plume

However, numerical models have their own set of associated problems. For example, using a grid cell size that is too large causes a phenomenon known as numerical dispersion. If a small facility sits on a large grid cell, the contaminant mass from that facility is immediately spread out over the whole of the grid cell, thus dispersing the mass over an unrealistically large area (hence, the term numerical dispersion).

Another problem that plagues numerical and analytical models alike is the lack of observed data throughout the domain of interest. Typically, only relatively sparse point data from monitor wells are available, and the modeler is left with using some sort of interpolation scheme to fill in the gaps between. In addition, a variety of model input parameters, such as effective porosity (which has a significant effect on contaminant velocity), recharge, and dispersivity (a measure of contaminant spreading), can only be estimated.

Once each PRP's groundwater plume has been defined using a combination of the methods described above, a variety of allocation approaches can rely on the defined plumes. Use of the defined plumes for the area, volume, mass, and combination mass/volume allocation approaches is described in the following sections.

Area-of-Contribution Method

The area of contamination in an aquifer is a surrogate for the volume of contamination. Individual plume areas are best used to estimate contribution where aquifers are thin, contamination is well mixed vertically, and groundwater flow is horizontal.

Costs of cleanup by, for instance, a pump-treat remedy depend, to a great extent, on the area or volume of contaminated aquifer. For example, wide plumes will likely require more interceptor wells to capture the contamination, and long plumes require either long system operation and maintenance times to allow COCs to migrate to a single well field, or they require additional well fields within the plume. Wide plumes in permeable aquifers also require high pumping rates for containment. The costs of constructing and operating a water treatment system often are controlled primarily by the rate of water flow through the system.

The area-of-contribution method of allocation attributes multiple source area contributions to a single chemical plume. Based on plume area, the method allows analysis of overlapping contributions from several PRPs. The following equation is used to determine the percent relative contribution of a source to the contaminated aquifer:

[IMAGE CANNOT BE RENDERED]

where:

RC_a

percent relative contribution of source a to the area of contaminated aquifer

A_i

subarea of aquifer, where source a contributed contaminants

L_i

number of contributing sources

m

number of subareas contaminated by source a

A_T

total area of contamination

The application of this method is shown in Figures 2 and 3. Figure 2 shows a regional plume consisting of the components of three smaller overlapping plumes. The area-of-contribution method can be applied to any of the three smaller plumes (sources) to determine their relative areas of contribution to the regional plume. Referring to Figure 2, assume that source a has a total plume area of 100 square feet (sq. ft.), source b is 30 sq. ft., and source c is 20 sq. ft. Also, assume that total area of the regional plume is 125 sq. ft. Therefore, the total square footage of the regional plume is reduced by the overlapping sections of the smaller plumes.

Assume that source b is selected to determine its contribution to the regional plume. Figure 3 shows source b and its overlapping sections and areas. Sections A1, A2, A3, and A4 have areas of 22, 4, 3, and 1 sq. ft., respectively. These areas make up Ai in the equation.

In the equation, Li consists of the number of contributing sources per subsection of source b. As indicated on Figures 2 and 3, area A1 consists of two contributing sources, created by the overlapping of sources a and b. Area A2 is impacted only by source b, while A3 consists of contamination from all three sources, a, b, and c. Finally, area A4 shows two contributing sources, b and c.

Figure 2.
Sample Calculation of Relative Area

[IMAGE CANNOT BE RENDERED]

Figure 3.
Sample Calculation of Relative Area (cont'd)

[IMAGE CANNOT BE RENDERED]

Completion of the equation with these data results in an area of contribution for source b of 0.13, or 13 percent of the area of the regional plume:

[IMAGE CANNOT BE RENDERED]

Although the area-of-contribution method provides a systematic, scientific approach, it doesn't always provide an equitable assessment of a PRP's impact to an aquifer. For instance, PRPs with small, continuous releases over long time periods may have a contaminant plume of large area, but insignificant concentrations. A large, low-concentration plume may pose lower risk and require a less aggressive, less costly remedy.

In addition, the area-of-contribution approach is limited to a two-dimensional assessment of the contamination. Therefore, this method can account for the depth of the plume within the aquifer if all monitor-well data are projected to the same horizontal plane.

Volume-of-Contribution Method

The volume-of-contribution method is identical to the area-of-contribution method, except that the depth of the contamination is considered. This method is useful for sites with thick aquifers, where contamination from PRPs has migrated to various depths.

Mass-of-Contribution Method

The mass-of-contribution method of allocation is straightforward and simply considers the percentage of total plume mass that was contributed by any one facility. The contribution percentage of a particular facility is calculated by dividing the amount of mass released to groundwater from that facility by the total amount of mass in the groundwater plume, multiplied by 100. For example, the allocation for facility a, if two facilities, a and b, exist, is given by:

[IMAGE CANNOT BE RENDERED]

where Aa is the allocation percentage for facility a, Ma is the mass attributable to facility a, and Mb is the mass attributable to facility b. The amount of mass released from a facility can be estimated from plume concentration contours and by modeling. Chemical purchase records and waste inventory records may be used as surrogates for groundwater impacts in some cases.

Mass/Volume Contribution Method

The mass contribution of a facility may not be the complete deciding factor in developing an allocation percentage because it fails to account for sites where the plumes are widely dispersed. Plume mass drives treatment costs, such as carbon canister usage and air stripping tower height, while volume drives pumping costs. (A large plume takes a long time to pump out, which translates to high pumping costs.) Plume volume is associated with the cost of pumping, including such items as capital costs to contain and capture plume volume (number of wells, well spacing, distance to treatment system, operation and maintenance, power consumption, piping). A concentrated but compact plume will be inexpensive to pump but expensive to treat. A large but low-concentration plume usually will be relatively inexpensive to treat but expensive to pump. Both the mass and volume of a plume should be considered and compared against the capital costs of pumping and distribution facilities, as well as treatment operating costs. Typically, both are considered, and some sort of weighting scheme is developed.

The mass/volume contribution is based on the following formula:

[IMAGE CANNOT BE RENDERED]

where:

P

fraction of remedial costs associated with pumping

T

fraction of remedial costs associated with treatment

V_A

volume fraction contributed to plume A

M_A

mass fraction contributed to plume A

Obviously, the proportion of volume- and mass-related costs depends on the remedy. Generally, four classes of groundwater remediation are available for a site: (1) active, such as pump and treat; (2) passive, such as a funnel and gate system; (3) limited action, such as in situ natural attenuation; and (4) no action, monitoring only.

Both the mass and volume of a plume should
be considered and compared against the capital
costs of pumping and distribution facilities,
as well as treatment operating costs.

For the allocation formula to be used, each type of remediation must be evaluated, and a consensus among the parties or allocation panel members regarding the assignment of the remedy costs to either a mass- or volume-related category is necessary. As an example of this approach, refer to Figure 2 on the previous page. Assuming that the three depicted plumes all have uniform depths, then the volume fraction contributed by source b would continue to be 13 percent, as calculated with the area-of-contribution method. Also assume that, by modeling or contouring contaminant concentrations for the three sources, assigning a total mass to each plume is possible and, consequently, the mass fraction for source b is 65 percent.

The remedy is a pump-and-treat system that requires 70 percent of the total costs for pumping (volume) and 30 percent for treatment (mass). Using the formula, the resulting allocation percentage for source b is as follows:

[IMAGE CANNOT BE RENDERED]

Application of this approach may present challenges at some CERCLA sites. Other factors may need to be considered, including (1) the proximity of the PRP to the remedy and other PRPs; (2) the contention that a PRP's contaminated groundwater will not enter the remedy during the treatment period; and (3) the potential risk and harm associated with the PRP's contamination. As an example of how to address some of these issues, a PRP group might decide that the allocation is not tied to any specific remedy design, since this would specify well locations and treatment rates that are inherently unfair to certain PRPs, simply due to their proximity to the remedy.

DEALING WITH ORPHAN SHARES

An orphan share consists of contamination unattributable to a PRP. One of the ways to deal with orphan shares is to increase each known PRP's share to cover the total costs of the orphan shares. The increase is proportional to each PRP's known relative contribution.

The IPA recognizes the orphan shares, since the allocation is a percentage of PRP plumes only. The IPA is useful in determining and allocating orphan shares, since the mass and volume from PRPs, based on the IPA, can be subtracted from the total plume mass and volume to estimate the orphan shares. The orphan shares can then be reallocated to the PRPs, based on each PRP's allocation percentage.₍₃₄₎ For example, in a mass allocation method, assume that 100 pounds of TCE occurs in an aquifer. Further assume that the mass contribution for Site A is 50 pounds and Site B is 30 pounds. Therefore, the total mass attributable to the PRPs is 80 pounds and the orphan share is 20 pounds.

Site A's new share is: [IMAGE CANNOT BE RENDERED]

Site B's new share is: [IMAGE CANNOT BE RENDERED]

CONSIDERATIONS FOR SUCCESSFUL ALLOCATION

Many different scientific methods are available for CERCLA allocation processes. As shown above, significantly different results are likely depending on the method selected. For example, a mass-only allocation may unfairly penalize PRPs with small, concentrated plumes. Conversely, a volume-only plume might be biased against PRPs with low-concentration, widely dispersed plumes. An equitable allocation typically uses some combination of approaches designed to be representative of site-specific conditions for the PRPs.

A clear understanding of the scientific assumptions can help to minimize the potential bias that may be introduced by a proposed allocation approach. This includes an understanding of how site data have been interpreted for the construction of groundwater plumes, the significance of modeling input parameters, and the interpretation of the fate and transport of contaminants in soil and groundwater. A willingness to be flexible, but diligent, in promoting an equitable and representative approach is one of the keys to a successful allocation outcome.

ABOUT THE AUTHORS

Barbara Graves is a senior scientist at Daniel B. Stephens & Associates, Inc., where she serves as the Midwest operations manager. She currently provides project management for environmental forensics applied to litigation and alternative dispute resolution.

David Jordon, P.E., is a senior hydrogeologist with Daniel B. Stephens & Associates Inc. He primarily manages and provides technical expertise for environmental litigation and allocation cases.

Dominique Cartron, J.D., is a water resources specialist with Daniel B. Stephens & Associates, Inc. Ms. Cartron brings a multidisciplinary approach to water resources projects with an emphasis on regional groundwater planning, environmental regulation, and water rights.

Daniel B. Stephens, Ph.D., is principal hydrologist and president of Daniel B. Stephens & Associates, Inc., which provides hydrogeology and engineering consulting services. Dr. Stephens has been extensively involved in soil and groundwater litigation support projects, including site investigations and remediation efforts throughout the United States.

Michael Francis, is a partner at Demetriou, Del Guercio, Springer, & Moyer, LLP, in Los Angeles, CA, where he practices environmental law. Mr. Francis has actively participated in the development of ADR allocation agreements and the allocation of responsibility in the San Fernando Valley and the San Gabriel Valley Superfund sites. Mr. Francis was formerly with Rockwell International Corporation.

ENDNOTES

(1)

42 U.S.C.§ 9601 (1994).

(2)

Because other parties are also responsible for the CERCLA response costs, a liable party will seek to limit the amount it is ultimately required to pay during the allocation process, whether part of a settlement process or after adjudication of liability. Although CERCLA identifies four classes of persons who can be held liable, it does not provide any meaningful guidance on how CERCLA response costs should be allocated among multiple liable parties. 42 U.S.C. § 9607(a). Therefore, the allocation process becomes critical.

(3)

H.R. Rep. No. 96-1016, at 17-18 (1980), reprinted in 1980 U.S.C.C.A.N. 6120.

(4)

42 U.S.C. § 6901 (1994).

(5)

Bulk Distribution Centers, Inc. v. Monsanto Co., 589 F. Supp. 1437, 1441 (S.D. Fla. 1984) (CERCLA necessary because RCRA not equipped to deal with the problems of abandoned hazardous waste sites).

(6)

As explained by many other commentators, the legislative history of CERCLA is complex. CERCLA was enacted in a lame duck session of Congress. Basically, four predecessor hazardous substance site cleanup bills were introduced in the 96th Congress. H.R. 7020 was passed by the House after due considerations for the Senate's deliberation. S. 1480 was vigorously opposed, and its sponsors made several substantial last-minute changes to appease the bill's opponents, including reducing the appropriation for the Superfund program, eliminating provisions of liability for personal injury and economic loss, and striking certain provisions dealing with joint and several liability. The Senate passed this amended version of S. 1480. Because revenue measures must come from the House, H.R. 7020 was replaced in its entirety, with the amended S. 1480, and presented to the House as an amendment to the earlier legislation. By this time, too little time remained, and the House was faced with the choice of adopting the bill word for word or letting the bill die. The bill was then passed by the House under a suspension of the rules, which prohibited the introduction of amendments and permitted only limited debate. United States v. Reilly Tar & Chem. Corp., 546 F. Supp. 1100, 1111 (D. Minn. 1982).

(7)

Because of the way CERCLA was passed, no committee or conference reports address the legislative version that became law. Thus, courts, lawyers, and interested persons cannot find significant guidance when interpreting CERCLA's often ambiguous provisions. For example, in United States v. Price, 577 F. Supp. 1103 (D. N.J. 1983), the court noted that "[e]ven the legislative history must be read with caution since last-minute changes in the bill were inserted with little or no explanation." Id. at 1109.

(8)

42 U.S.C. § 9607(a).

(9)

Pub. L. No. 99-499, 100 Stat. 1613 (1986) (codified in scattered sections of 42 U.S.C.).

(10)

42 U.S.C. § 9613(f).

(11)

42 U.S.C. § 9622.

(12)

42 U.S.C. § 9621.

(13)

42 U.S.C. § 9617(a).

(14)

42 U.S.C. § 9607(a)(1)-(4).

(15)

42 U.S.C. §§ 9601, 9607(a); see United States v. Alcan Aluminum Corp., 964 F. 2d 252, 254 (3d Cir. 1998); United States v. R.W. Meyer Inc., 889 F. 2d 1497 (6th Cir. 1989); United States v. Monsanto Co., 858 F. 2d 160, 167 (4th Cir. 1988).

(16)

See United States v. Chem-Dyne Corp., 572 F. Supp. 802 (S.D. Ohio 1983); United States v. Northeastern Pharmaceutical & Chem. Co., Inc., 810 F. 2d 726 (8th Cir. 1986); Price v. U.S. Navy, 39 F. 3d 1011, 1018 (9th Cir. 1994); B.F. Goodrich v. Betkoski, 99 F. 3d 505, 510 (2d Cir. 1996).

(17)

42 U.S.C. § 9607(a)(A)-(D).

(18)

42 U.S.C. § 9607(b).

(19)

42 U.S.C. § 9601(350).

(20)

42 U.S.C. § 9613(h); see United States v. City and County of Denver, 100 F.3d 1509, 1513-14 (10th Cir. 1996).

(21)

42 U.S.C. § 9622(e)(3).

(22)

Interim Guidelines for Preparing Nonbinding Preliminary Allocations of Responsibility, OSWER Directive #9839.1, 52 Fed. Reg. 19,919 (1987) [hereinafter Guidelines].

(23)

Id. at 19,920.

(24)

Id.

(25)

Id.

(26)

Id. at 19,919. One court used a similar two-step approach where allocation was made first in terms of volume and then adjusted relative to "equitable factors." U.S. v. Davis, 31 F. Supp. 2d 45, 65 (D. R.I. 1998).

(27)

Guidelines, supra note 22, at 19,920.

(28)

Id.

(29)

Review of the Hazardous Substance Superfund: Hearings before the Subcomm. on Oversight of the Comm. of Ways and Means, 102d Cong. 122 (1992).

(30)

See Superfund: Further EPA Management Action Is Needed to Reduce Legal Expenses, GAO/RCED-94-90 (1994).

(31)

H.R. 7020, 98th Cong. (1980).

(32)

42 U.S.C. § 9613(f)(1).

(33)

See Gould Inc. v. A & M Battery & Tire Service, 987 F. Supp. 352 (M.D. Pa. 1997); Boeing Co. v. Cascade Corp., 920 F. Supp. 1121 (D. Or. 1996); U.S. v. Colorado & Eastern R.R. Co., 50 F.3d 1530 (10th Cir. 1995); Control Data Corp. v. S.C.S.C. Corp., 53 F.3d 930 (8th Cir. 1995); Kerr McGee Chemical v. Lefton Iron & Metal, 14 F.3d 321 (7th Cir. 1994); United States v. R.W. Meyer, Inc., 932 F.2d 568 (6th Cir. 1991).

(34)

Browning-Ferris Industries of Ill. v. Ter Maat, 13 F. Supp. 2d 756 (N.D. Ill. 1998) (court allocated orphan share on a pro rata basis to solvent PRPs).

©Copyright 2001 - All Rights Reserved

DO NOT REPRODUCE WITHOUT WRITTEN PERMISSION BY AUTHORS.