Click for next page ( 146


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 145
APPENDIX C A PROBABILISTIC CRITICAL GROUP Although the components of a probabilistic computational approach have considerable precedent in repository performance, we are not aware that they have previously been combined to analyze risks to critical groups. We have therefore outlined in this appenciix a fairly explicit example of how this approach might be implemented for the case of exposure through contaminated ground water. The main purposes of this example are to show that the approach is feasible and to illustrate the steps necessary to perform such a calculation. The example uses a Monte Cario method for mocleling exposure consistent with that employed in the hydrologic modeling of radionuclide transport. In presenting this appendix, we do not intend it as a detaileci recommendation, but an exploration of at least the more important issues that are likely to arise in an actual compliance calculation. The additional detail in this appendix is warrantee} because the technique has not been applier! to this problem in the past, as far as we are aware. The following outline of steps is designed to provide an illustrative example of the types of calculations that could] be employee] in an exposure scenario analysis. The specific process described here is only one of a variety of alternatives that EPA might consider during its rulemaking. It is based on a number of choices and general considerations, some of which are reviewed below prior to a description of the steps themselves. a. Technical feasibility of the calculations requires specification of one or more exposure scenarios. As described In (chapter 3, a scenario includes parameter values or distributions that provide quantitative descriptions that include where people live, what they eat and drink, and what their sources of water and foot! are. A given scenario might inclu(ie the lifestyle and activities of only farmers or a mix of economic lifestyles and activities of farmers, miners, defense workers, and casino operators, for example. It might be based on actual current activities in the area of interest, on current activities in some adjacent area, or potentially on any . .. . . ~ . 145

OCR for page 145
146 YUCCA MOUNTAIN STANDARDS number of hypothetical future activities. The only technical consideration in the selection of an exposure scenario is whether the specified scenario provides sufficiently well definer} parameters or parameter distributions to make calculations feasible. The selection of the exposure scenario, along with its associated parameter values, is fundamentally a policy choice and therefore an appropriate responsibility of rukemakers. Broacl participation in this policy decision by the various affected interested parties ant! acceptance of the scenario as a reasonable basis for performance assessment are likely to be essential to acceptance of any results of the analysis (NRC, 19931. Even for a narrowly specified set of parameters, it is possible that the calculation procedure can be manipulated to obtain results closer to those desired by the analyst. It might not be possible to eliminate all opportunities for this type of manipulation. However, careful consideration of these possibilities during the rulemaking process might help to develop guidelines for calculations to address some of the potential pitfalls. For example, we were particularly concerned with avoiding strategies that would reward uncertainty in the temporal or spatial distribution of radionuclides in ground water. A procedure in which larger uncertainty in transport parameters leads to a reduction in calculated risk, relative to the risk that would be calculated were transport parameters less uncertain, would provide a strong disincentive to reduce uncertainty through site- characterization activities. A second issue is how to quantify properly the risk in areas of low-population density, because the probability of an incliviclual receiving a close in these areas is dependent on whether any individual is present in the area at the time when raclionuclicles are present in the underlying ground water. A critical feature of this model, therefore, is that a method must be incorporated for calculating the probability that people are present over the contaminated plume of ground water. The method illustrated in this appendix employs a fully probabilistic treatment of all aspects of the exposure b. C.

OCR for page 145
APPENDIX C - A PROBABILISTIC CRITICAL GROUP 147 scenario. This results in a computationally intensive procedure. It might be possible to recluse the computational requirements by treating parts of the calculation deterministically or analytically. d. The illustrative example focuses on exposures ant! risks associates] with grounci-water use. The fact that gaseous releases are not included in this example should not be interpreted as a judgment that such releases can be excluded from performance assessment and compliance evaluation. A separate exposure scenario, with a different critical group, would be required for evaluation of the gaseous exposure pathway. In the end, however, one pathway will result in the maximum risk ant! define the critical group whose protection would be the primary metric for setting the stanciar~i. Example Steps Required for Implementation of a Monte Carlo Analysis Step I: Identify general lifestyle characteristics of the larger population that includes the critical group. The first step is to identify the type of people who would be likely to receive the highest doses and therefore be at greatest risk. These people make up a group that might be considerably larger than the critical group, but of which the critical group will be a subset. As noted earlier, this step involves subjective choices that should be part of the rulemaking process. For purposes of illustration, this example assumes a farming community scenario, based on present-day conditions in the Amargosa Valley. Step 2: Quantify important characteristics, distributions of characteristics, and geographic location of the chosen population. The second step addresses two aspects of the exposure analysis. First, any analysis of exposure will require specific information on the living patterns' activities and other characteristics of potential members of the exposed population that can be used as input to deterministic or

OCR for page 145
148 YUCCA MOUNTAIN STANDARDS probabilistic simulations. Second, if identification of the characteristics of currently occupied land and technologies (such as soil type, slope, depth to ground water, well depth, etc.) provides a technical basis for limiting the simulation area for exposure analysis, significant reduction in the computational effort required for the calculations would result. In a Monte Cario simulation, each of the pertinent parameters is represented by a distribution of values, from which one value for each is randomly selected for each of many calculations. For the purpose of this example, we assume that each of these factors could be quantified using surveys anti studies of the existing population in the region. Correlations between factors would need to be identified, such as relationships between farm density and soil type or depth to ground water. Analysis of these tiara would provide a basis for a mode! of the farming economy that can be user} to identify geographic areas in the basin that have the potential for farming and grounci-water use. It is important to note that these areas wouicl not necessarily correspond to the current areas of highest population density or water use, since there might be areas of arable land that have not been clevelope~i due to restricted access (anywhere in the Nevada Test Site, for example). There might be areas where higher rates of water use could be easily sustained but have not been implemented by some farmers, or for a variety of other reasons. Step 3: Simulation of radionuclide transport and identification of potential exposure areas The third step is to identify the potential intersections of potentially farmable areas and areas beneath which radionuciide-contaminated ground water occurs. Delimiting the intersections of these areas can further reduce the computational effort. The physical location and chemistry of the plume of contamination can be identified by performing a series of Monte Cario simulations of the release and transport of the wastes-through the unsaturated zone to the water table and in the saturated zone. Each simulation will generate a plume path (direction, wicIth, depth below the water table, thickness) and its surface footprint. This footprint can be overlaid on the map of potential farm density or water use to determine a potential exposure area. If the mode} employs an appropriate sampling of the input parameters controlling

OCR for page 145
APPENDIX C - A PROBABILISTIC CRITICAL GROUP 149 radionuclide release ant! transport, each of the many plume realizations can be considered an equally likely outcome of radioactive waste disposal at Yucca Mountain. If the number of plume simulations is sufficiently large, the series of calculations defines the statistical characteristics of the problem. Step 4: For each plume realization, identify critical "snapshots" of radionuclide distribution at timers) for which the plume underlies exposure areats) identified in step 3. Even if the plume evolution were perfectly predictable, ant} hence the potential exposure area perfectly constrained, not all inhabitants of this exposure area would be at risk. There will be a long perioc! of plume history (that floes not even begin until radionuclides reach the saturated zone) during which radionuclide contaminates! ground water will not have reached the aquifer beneath a potential exposure area. Inhabitants of a potential exposure area living there during these periods are at no risk. Once the plume reaches the aquifer beneath an exposure area, the risk to inhabitants will vary with time as the areal extent of the plume ant} radionuclicle concentrations in the contaminated ground water change cluring plume migration. If the critical group comprises a set of individuals who have the greatest average risk, then the temporal as well as spatial distribution of risk must be considered in identifying the group. The purpose of this step is to account for the temporal variation in risk by identifying a) the time at which inhabitants of a potential exposure area will be at maximum risk and b) the corresponding radionuclide distribution in ground water at that time. The subsequent exposure analysis can then be conductecI employing the ractionuclide distribution for this critical time. Each of the simulations produces a realization of plume evolution in space and time. The spatial distribution of radionuclide concentrations in ground water at an instant in time constitutes a plume snapshot. If rates of plume evolution are slow, as would be expected from performance assessment calculations conducted to date for Yucca Mountain, a snapshot for an instant in time is also likely to be representative of the plume distribution over the course of a human lifetime, or even over many generations. Examining a series of snapshots generated by a simulation, one can identify the period of time, for each simulation, during which

OCR for page 145
150 YUCCA MOUNTAIN STANDARDS peak radionuclicle concentrations or high total (volume integrated) activities are present beneath the areaLs) clelimited in step 3. These periods should correspond to the times at which the population in the exposure area would be at significant risk. Determining the time of greatest risk might not be straightforward, however, because times of peak concentration (possibly over a very limited area) might not coincide with times of greater plume extent, that would have somewhat lower concentrations but greater total activity. Step 5: Generate exposure realizations Having identified the time period of maximum potential exposure for each plume realization, it is also necessary to determine the spatial distribution of potential doses and health effects to identify the critical group and to calculate the risk to an average individual in that group. The next step, then, is to use the plume snapshots in the Monte CarIo series of exposure simulations. For each of the plume snapshots selected in step 4, a large number of Monte Cario simulations would be performed. For each exposure simulation, statistical distributions of population characteristics as determined in step 2 would be sampled to generate a distribution of farms with associated inhabitants, wells, crops, livestock, and support services within and surrounding the exposure area (as determiner} in step 3~. Well depth and screened interval, rates of water use, food sources and consumption rates, etc. would also be determined by sampling from the parameter distributions. The number of exposure simulations must be large enough to produce an adequate sampling of exposure parameter distributions. Each simulation should cover a large enough region outside the exposure area to allow adequate definition of dose variations between the exposure area and the surrounding region. Exposures outside the area overlying the plume could result from local export of water or food from the exposure area, factors that must be inclucled in the exposure analysis. Some exposures might also occur to inhabitants living over the plume but outside areas of intense farming or water use.

OCR for page 145
APPENDIX C - A PROBABILISTIC CRITICAL GROUP Step 6: Calculation of dose distributions for exposure realizations 151 The spatial relations between plume boundaries and well locations in the exposure realizations will determine which wells have the potential, constrained by well depth and screened interval, to produce water leading to human exposures. For a known concentration, rates of water use for drinking en c! irrigation will determine the activity extracted from the ground, anti the subsequent distribution of that activity to humans, crops, livestock, etc., ant! the resulting close to each inhabitant represented] in the exposure realization. Step 7: Interpretation of exposure simulation results to identify critical subgroups For each of the plume realizations, the results of the exposure simulations can be combined to yield a spatial distribution of expected close, which can then be used to identify tile geographic area inhabited by the critical subgroup for a given plume realization. For example, the individual closes of the combiner! plume and exposure simulations could be clivicied into subsets based on geographic location of the inhabitants. The sizes of the subareas should be adjusted to provide adequate resolution of the spatial variation in individual dose and to account for the variations in the scenario-specific population density over the simulation region. This could result in a highly variable grid size. A sufficient number of individuals must be simulated in each subarea to allow computation of a meaningful average dose. For each subarea, an average individual dose court! be computed as the arithmetic mean of the indiviclual doses in that subarea generated by the exposure simulations. The product of this average close ant! the factor relating doses to health effects (5 x 10-2 fatal cancers/Sv) would be the average lifetime risk for an individual in the subarea. The procedure for identifying the critical subgroup for one of the plume realizations would begin by delineating the subarea of the simulation region with maximum average risk plus additional subareas in which the risk is greater than or equal to one-tenth the risk in the subarea with maximum risk. These subareas constitute a trial area for a critical subgroup that is homogeneous with respect to risk. The average risk in this

OCR for page 145
152 YUCCA MOUNTAIN STANDARDS trial area is calculated as the arithmetic mean of the subarea risks. A critical sub-group can be considered! homogeneous if it satisfies the criteria detailed in Chapter 2. Step S: Calculation of average risk to members of the critical group The procedure outliner! in step 7 will generate a risk for the critical subgroup corresponding to each of the plume realizations. The arithmetic average of these critical subgroup risks over all plume realizations is the technically appropriate representation for the critical-group risk. The variability in risks between critical subgroups is related primarily to the variability in potential plume concentrations and locations resulting from the probabiliistic simulations of release and transport mechanisms. Using the average critical subgroup risk provides an estimate of the risk to the critical group exposed to the average plume. Additional insight might be obtained by examining the cumulative distributions of the critical subgroup risks.