Christine Daily, Frank Cardile
US Nuclear Regulatory Commission

The U.S. Nuclear Regulatory Commission published the final rule on radiological criteria for license termination on July 21, 1997. This final rule amends 10 CFR 20, Subpart E, and establishes criteria for the remediation of contaminated sites or facilities that will allow their release for future use with and without restrictions. As part of the work associated with implementing the final rule, a decision methodology has been developed to support implementation of the dose assessment requirements in the new Subpart E. A logical, consistent decision process is viewed as a useful tool that will support licensee planning of decommissioning activities and NRC review of license termination requests.

The U.S. Nuclear Regulatory Commission published the final rule on radiological criteria for license termination on July 21, 1997. As part of the work associated with implementing the final rule, a decision methodology has been developed to support implementation of the dose assessment requirements in the new 10 CFR 20, Subpart E (see Figure 1). Subpart E establishes criteria for the remediation of contaminated sites or facilities that will allow their release for future use with or without restrictions. The decision process supports assessment of the entire range of dose modeling options from which a licensee may choose, from changing a single parameter to changing multiple parameters and modifying pathways or models. A logical, consistent decision process is viewed as a useful tool that will support licensee planning of decommissioning activities and NRC review of license termination requests.

Generic exposure scenarios and pathways have been defined based on the NUREG/CR-5512 methodology and can be used without further analysis or justification by licensees who are applying the default scenarios and parameters using the DandD software. The default screening scenarios and pathways provide the licensee with a simple method to demonstrate compliance using little or no site-specific information. The generic models and default parameters are intended to estimate the upper range of the dose that the average member of the critical group could receive. The default parameters were developed probabilistically to control the regulatory risk associated with releasing a site based on source term data alone.

For licensees with more complex decommissioning situations, the decision process supports the modification of model parameters to allow site specific factors to be taken into account while still using the default models. This allows a licensee to use site-specific values in place of some or all of the default parameters. Thus, the dose estimates are more realistic, but should still be conservative for a particular site based on the use of the default models. The site specific data are used to support modifying or eliminating a particular scenario or pathway, or to demonstrate that a parameter or group of parameters can be better represented by site specific values. Alternative exposure scenarios may be appropriate based on site-specific factors that affect the likelihood and extent of potential future exposure to residual radioactivity.

The purpose of the framework is to provide a logical structure for regulatory decision making within the context of the requirements of the rule on radiological criteria for decommissioning. In the following description of the framework, a licensee with a relatively simple decommissioning situation would follow steps 1 through 7. For licensees who do not meet the 25 mrem/y criterion the first time through step 5, steps 8 through 13 are described and an example application is provided below.

Step 1: The first step of the decision process involves gathering and evaluating existing data and information. Licensees should check their records to determine the types and amounts of radioactive material they possessed on their site. They should also gather information about any surveys and leak tests that had been performed, as well as any records that would support their ability to "Certify the disposition of all licensed material, including accumulated wastes, by submitting a completed NRC Form 314 or equivalent information" [10 CFR 30.36(j)(1)].

Step 2: The second step in the decision process involves defining the scenarios and pathways that are important for the site dose assessment. For a generic application, this step has already been completed by the NRC, based on the generic scenarios and pathways for screening that have been defined and described in NUREG/CR-5512, Volume 1.

Step 3: This step involves system conceptualization, which includes conceptual and mathematical model development and assessment of parameter uncertainty. For a generic application, this step has already been completed by NRC, using the models described in NUREG/CR-5512, Volume 1, and implemented in the DandD software.

Step 4: involves the dose assessment for the site, which means running DandD or equivalent software with the appropriate site specific source term.

Step 5: is the first major decision point in the methodology, and involves answering the question of whether the dose assessment results from the model are less than the dose criterion of 25 mrem/yr in 10 CFR 20, Subpart E.

Step 6: If the result in Step 5 is that the 25 mrem/y criterion has been met, the licensee can proceed to satisfy any remaining ALARA requirements.

Step 7: In this step the final paperwork requirements are completed, including documenting any survey results used to calculate the source term and the results of the dose calculations. The documentation is sent to the NRC to support the request for license termination.

A useful way to describe framework steps eight through thirteen is to use an example application. The example application involves a complex case where a decision must be made between unrestricted and restricted release. For the purposes of this example, the licensee is interested in terminating the license for an outdoor location that is believed to have areas of soil contamination from leaks in a waste tank. Although this licensee has a more complex situation and would not expect to be able to perform a simple demonstration of compliance using default parameter values, they would still follow the same steps described above, at least for the first iteration. As before in step one, they would gather as much information as possible about their site, including radionuclides and processes used, quantities and forms of material that might still remain on site, and anything else that would be useful for performing a site dose assessment.

For the scenario definition and pathway identification in step two, the licensee in this example decides to begin the decision process by using the pre-defined scenarios and pathways in the residential scenario (soil contamination) described in NUREG/CR-5512, Volume 1. In step three, they also accept the default parameters and use the DandD software. For the step four dose assessment, they run DandD using a source term developed from the information gathered in step one, and which is the maximum reasonable value they believe they can defend.

Based on the results of step four, in step five it is clear that the site does not meet the Subpart E dose criterion of 25 mrem/yr. The licensee would therefore proceed to step eight and begin defining their options. Note that there are basically three options that the licensee could apply either alone or in combination: Option 1 - Activities that reduce uncertainty (information/data collection), Option 2 - Activities that reduce contamination (remediation), and Option 3 - Activities that reduce exposure (land-use restrictions). Table 2.2.1 lists some of the options that a license could consider, including two related to reduction of uncertainty, one related to reducing contamination, and one related to reducing exposure.

Only a limited number of sites will need to perform complex dose assessment and options analyses, with most sites performing an options analysis that is relatively simple and straightforward. For example, a site with a small, contained source of contamination that is obviously simple to remove would not perform extensive analyses on large suites of alternative data collection and remediation options. The same may be true for certain complex sites, where the configuration of the contamination, site conditions, or regulatory requirements cause the options for proceeding forward to be relatively limited. The sites which will benefit the most from this options analysis are those with complex contamination situations where this process can be used to analyze a variety of simple and complex options and define the most effective and cost-efficient decontamination and decommissioning strategy.

For the first option, activities that reduce uncertainty, it is useful to begin by looking at the default parameter values in the NUREG/CR-5512 model and what they represent. The default parameter values for the NUREG/CR-5512 modeling (that have been implemented in DandD) were developed based on probability distributions representing the expected variability across all NRC sites in the country. A probabilistic parameter analysis was performed to select a set of default parameters that meet the NRC's requirements to control the regulatory risk associated with releasing a site based only on source term information. The regulatory risk is defined as the risk that a site will be released when it exceeds the dose criterion. The risk is controlled by selecting screening parameters that, as a set and within the context of the specified model, provide a specified level of confidence in the dose estimate and control the amount by which the dose could exceed the criterion. The parameter analysis will be published as Volume 3 of NUREG/CR-5512 and will list the parameter distributions used to develop the default parameters as well as providing information regarding valid ranges for site specific parameter changes that a license could propose without an additional site specific uncertainty analysis. As a consequence, the licensee needs little supporting information to defend changes to the parameter values that are within the limits specified in the parameter analysis. This is important in evaluating the relative worth of collecting additional data on these parameters under Step 9 of the decision framework.

For example, the probability distribution used in defining the default values for radionuclide sorption in soils for the NUREG/CR-5512 residential scenario models is based on the variability across all possible soil types at NRC licensed sites. To provide the NRC with an acceptable level of regulatory risk in terminating the license for a site based only on residual contamination data, the default value for sorption coefficient defined in the preliminary parameter analysis is representative of a very clean sand. Therefore, in this step of the options analysis the licensee might be able to propose that the sorption be increased using site specific data within the limits for this parameter provided by the NRC. The associated cost for this activity could be the cost of obtaining a soil conservation map for the site area or obtaining data on the site soil type (covered under Step 9). If the licensee wanted to apply a parameter value outside the specified range, they could do so if they performed a site specific uncertainty analysis. This graduated approach of moving away from the "prudently conservative" values used in the NUREG/CR-5512 modeling based on site-specific information could be used by all sites until the point that further reduction in simulated dose would require model changes. At that point, probability distributions for the new model parameters would have to be developed and defended by the licensee.

As stated above, the options that have been identified in this iteration include two related to reduction of uncertainty, one related to reduction of contamination, and one related to reduction of exposure. The first option would reduce uncertainty in the source term, while the second would replace the default kd with a more site specific value based on the site soil type. The third option listed in Table I would result in an actual reduction of the quantity of residual radioactivity remaining on the site. If the final option, reduction of exposure, were pursued, the licensee would be required by 10 CFR 20, Subpart E, to demonstrate that unrestricted release was not ALARA. This would require additional site specific modeling to ensure that the decision has a sufficient basis.

The licensee now moves to step 9, analysis of options in terms of cost and the likelihood of success. To evaluate the likelihood of success, an analysis of the potential outcome (consequence analysis) will need to be performed for each of the options. Depending on the option, this consequence analysis could be very simple (e.g., the option is complete remediation and the consequence is effectively restoring the system by removing residual contamination) to as complicated as refining and expanding the dose assessment. The cost and time required to complete each option must also be estimated. The consequence analysis should also address the uncertainty associated with each potential outcome. The desired endpoint is a determination of the likelihood or probability that employing a given option will result in meeting the criteria of 10 CFR 20, Subpart E.

The result of the activities performed under Step 9 is a logically organized list of options, and the corresponding cost, likelihood of site release (probability of success), and other important considerations given that the option is pursued. Table II contains examples of how the options could be organized. In some cases, the decision regarding the preferred option will be obvious; for example, a low cost of success and failure, high probability of success option will always be selected over a high cost, low probability of success option, given that the regulatory requirements are satisfied. However, the preferred option will not always be obvious, and additional analysis may be required for sites attempting to balance complex issues.

To analyze the potential outcome of the selected options, the licensee can use the DandD software to perform some low cost "what-if" calculations. For example, they can review the existing information about their source term and try to estimate how it would change based on additional characterization. Based on the quality of the existing information, they may be able to modify the source term and obtain a less bounding value. This modified source term would then be input into the model and a revised dose estimate calculated.

In the same way, the licensee could review site specific or regional data to determine the predominant soil type at their site. If the soil type is not well characterized by a clean sand, as was used to define the default soil parameters, the licensee could investigate the impact of changing parameters associated with soil type, such as kd. The NRC will provide guidance regarding what parameters can be changed and the valid range for modified parameters before additional uncertainty analysis is required. This information is useful for estimating the analysis costs associated with modifying default parameters. This process of evaluating the effect of changing parameter values can be continued for other model parameters that the licensee believes could be changed based on site-specific information. This is similar to performing an informal sensitivity analysis, and will help focus attention to those parameters likely to have the most impact on the calculation of dose. The licensee can then direct resources to reducing the uncertainty in those parameters, or can determine that a different approach is necessary before any higher cost activities, such as soil removal or site surveys, are begun.

For this example case, it is assumed that a preliminary evaluation of the remediation option indicates that it is not cost effective to remove the contaminated soil and transport it off site. However, the preliminary analysis is based on the default dose screening and initial bounding estimate of the source term, both of which impact the estimated soil volume requiring remediation, and the cost of remediation. These estimates will change as more site-specific data is obtained, which may make remediation a more reasonable option at another point in the decision process. At this point in the decision process, the idea is not to permanently eliminate options from further consideration, but rather to select the optimum approach based on the current state of knowledge.

As noted above, use of the option of setting land use restrictions requires a demonstration to the NRC that further reduction in dose levels to unrestricted use is not ALARA. Thus, the absence of cost and probability of success values for the restricted release option in Table II is used here to illustrate two important points. First, given the NRC's preference for unrestricted release, licensees are expected to fully evaluate unrestricted release for the first iteration through the decision process. Second, any information gathered to support other options can be used in a later iteration to support restricted release if necessary. It should be noted that the dose modeling must include as much site-specific information as necessary to provide a reasonable evaluation of future impacts, both with and without institutional controls in effect, to show compliance with restricted release criteria. The regulatory activities that will need to be completed prior to NRC granting a license termination under 20.1403 include (1) development of a safety evaluation report, (2) an environmental assessment (and, possibly, an environmental impacts statement), and (3) possibly, additional requests for information to allow staff review of the license amendment. In addition to the safety analysis report for the license amendment, licensees requesting license termination under restricted release will need to submit an environmental report, which will include the cost-benefit analysis for the ALARA determination. Under 20.1403, licensees would also need to seek advice from affected parties in the community regarding the restrictions on use.

This step in the decision framework should support an evaluation of the cost and time impacts of both success and failure. Generally, low cost / high likelihood of success options, or combinations of options, are preferred. This step should also include ALARA considerations, in terms of cost/benefit calculations as well as qualitative considerations. With regard to costs, the licensee should consider that if the option(s) selected are successful, the license will be released and further costs will be minimized. However, if the selected option(s) are unsuccessful, it may be necessary to perform additional characterization or remediation, or there may need to be an evaluation of restricted use (with its associated costs).

Once the various options have been evaluated, the preferred option can be selected in step 10. Based on the DandD analysis and cost/benefit and ALARA estimates for this example, the licensee decides to perform additional characterization of the source term, with the expectation that this will result in the source term estimate being reduced. The additional characterization will also involve obtaining data on the site soil type to support revision of the default kd. The combination of these two options should have a medium cost and a high likelihood of success. At this stage in the analysis, unrestricted release is preferred, and therefore restricted release not considered further at this time.

Under step 11, the preferred option is implemented. The licensee develops a characterization plan that will support both radiological and soil data requirements, then obtains regional soil maps and performs a radiological site survey. If the licensee has a very high expectation that the additional information will be sufficient to support a revised dose assessment that is less than or equal to 25 mrem, it may be worthwhile to design the site survey so that it can be used as a final site survey. That is how the licensee proceeds in this example, based on the results of their preliminary calculations and estimates of defensible changes to the default parameters. However, it is important to note that the final site survey has more extensive requirements than may be needed if the site requires remediation. The extra cost of a final site survey must be weighed against the need to repeat the survey at a later time.

Once the preferred option has been implemented, the model assumptions, parameter values, and pathways (as appropriate) are revised in step 12 of the decision process. For this example, parameter values associated with soil type (kd) and source term are modified based on the site data. In this case, the licensee decides not to pursue a site specific uncertainty analysis, and therefore modifies the kd only up to the bounding value supplied by NRC. Although the site data would support a somewhat less conservative kd, the effect on the calculated dose was not considered worth the additional effort associated with a site specific uncertainty analysis. To support the future request for license termination, the site survey results, soil maps, and methods used to revise Kd are carefully documented.

The revised source term and parameter values are used in iteration 2 of the dose assessment in step 4. In this example, the licensee decides to leave the original default model assumptions and pathways unchanged, and continues to use the DandD software. [Note that in other more complicated situations a licensee might seek to modify these assumptions and pathways. A detailed submittal discussing such changes would need to be developed]. When the revised parameter values and source term are input into the model, the result is a dose equal to 25 mrem/y.

This brings the licensee back to step 5 and the question regarding whether the site can be released. Since the dose assessment result is equal to 25 mrem/y, and the site survey met the minimum requirements for a final release survey, the licensee can move on to consider any remaining ALARA requirements in step 6. The licensee can document that best practice procedures were applied as part of its operational program. In addition, ALARA was incorporated and documented in the options definition (step 8), analysis of options (step 9), and selection of the preferred option (step 10). Based on the above, the license can be terminated and the site released. The licensee submits all required forms, including NRC Form 314, and documentation of the decision process, and the site is released for unrestricted use.

This decision methodology supports the efficient balancing of options to allow the most effective license termination strategy. The licensee can reduce conservatism in a controlled and cost effective manner, based on the specific conditions at their site. If they can demonstrate compliance without modifying any parameters or model assumptions, the process is straightforward and the NRC has high assurance that the dose criteria has been met. If it is reasonable and defensible to modify parameters or models, that can be done within a known standardized methodology, with only as much site specific information obtained as needed to support the changes from the defaults. For more complex situations, licensees can apply the same uncertainty analysis on a site specific basis that the NRC applied generically, to develop the most reasonable and defensible dose assessment.

BACK