ML20202C815

Draft Generic Position on Groundwater Travel Time

ML20202C815
Person / Time
Issue date:	06/30/1986
From:	NRC OFFICE OF NUCLEAR MATERIAL SAFETY & SAFEGUARDS (NMSS)
To:
Shared Package
ML20202C779	List: ML20202C775 ML20202C803 ML20202C815
References
REF-WM-1 NUDOCS 8607110411
Download: ML20202C815 (34)

Text

' I DRAFT GENERIC POSITION Cf. GF0Lt.DWATER TRA'/EL TIME e

f Division of kaste t'anagement Office of f.t. clear Fr.terial Safety and Safeguards U.f. N6 clear Regulatory Corraission Jure 20, 1986 8607110411 860701 PDR WASTE WM-1 PDR

GWTT/ JUNE 86 DRAFT GENERIC TECHNICAL POSITION ON GROUNDWATER TRAVEL TIME (GWTT)

Prepared by Richard Codell, Hydrology Section, Geotechnical Branch, DWM, NMSS June 30, 1986 TABLE OF CONTENTS

1. Introduction 1.1 What is Groundwater Travel Time

2. Interpretation of pre-waste-emplacement groundwater travel time 2.1 Pre-waste-emplacement environment 2.2 Identification of fastest path of likely radionuclide travel 2.3 Groundwater Travel Time along the Fastest Path 2.4 Rationale for the choice of the percentile of the cumulative distribution function of GWTT 2.5 Special Considerations for U.nsaturated Media

.3 . Summary and Statement of Regulatory Position. '

3.1 Summary ,

3.2 Statement of Regulatory Position 6

S i

4 I .

GWTT/ JUNE 86 Appendix A Calculation of GWTT A.1 Travel time distributions A.2 Mathematical representation of the repository and its environment A.2.1 Deterministic models '

A.2.2 Stochastic models ~

A.3 Site characterization from field data A.3.1 Treatment of uncertainty in site characterization A.3.2 Determination of input for the model A.4 Estimating GWTT from deterministic models A.4.1 Treatment of spatial variability A.5 Simplified analysis Appendix B - Definition of Paths B.0 Introduction '

B.1 Repositories in saturated media -

. B.2 Repositories in unsaturated media References Figures

1. Disturbed Zone, Path and Accessible Environment

2. Three-compartment Model

3. Groundwater travel time distribution

4. Breakthrough curves

5. Variance of groundwater travel time Distribution Bl. Paths of likely radionuclide transport - saturated media B2. Paths of likely radionuclide transport - unsaturated media

GWTT/ JUNE 86 DWM TECHNICAL POSITION ON GROUNDWATER TRAVEL TIME 1.0 Introduction One of the NRC performance objectives for High Level Waste repositories, commonly referred to as the " groundwater travel time (GWTT) objective", is stated in 10 CFR 60.113 (a)(2) as:

"The geologic repository shall be located so that pre-waste-emplacement groundwater travel time along the fastest path of likely radionuclide~

travel from the disturbed zone to the accessible environment shall be at least 1000 years or such other time as may be approved or specified by the ,

Commission."

The " disturbed' zone" is defined.in 10 CFR 60.2 as:

"...t' hat portion of the con' trolled are'a the physical or chemical ,

I properties of which have changed as a result of underground facility construction or as a pesult of heat generated by the emplaced radioactive wastes such that the resultant change of properties may have a significant effect on the performance of the geologic repository."

The " accessible environment" is defined in 10 CFR 60.2 as the atmosphere, land surface, surface water, oceans and the portion of the lithosphere that is outside of the controlled area. For purposes of this GTP, the " controlled area" is defined (consistent with the Final EPA high level waste rule 40 CFR 191) as extending no more than 5 kilometers from the original emplacement of the waste in the disposal system, with a traximum surface area of no more than 100 square kilometers. The disturbed zone, path and accessible envi.ronment are illustrated in Fig. 1.

The Disturbed Zone definition and groundwater travel time (GWTT) objective were established as part of a multiple barrier approach to high level waste i sol a. tion. The Disturbed Zone criterion is intended to prevent t.he reliance on only the zone dire ?!y adjacent to the engineered facility for the major portion of the gec.;gic barrier protection, and to avoid the complication.of consideration of coupled processes close to the emplaced High Level Waste whe.n demonstrating complianc'e with the GWTT performance objective. The Disturbed Zon'e is'being addressed by the NRC staff's Generic Technical Position which is presently under review. As the. Commission stated'when it proposed its technical crite.'3.for licensing. activities at' geologic repositories, the GWTT

GWTT/ JUNE 86 objective should be viewed as a conceptually simple measure of the overall quality of the geologic setting (46 FR 35280, July 8,1981).

It is generally agreed that groundwater is the most likely means by which significant quantities of radionuclides could escape a High Level Waste (HLW) repository. Transport of radionuclides to the biosphere then depend on factors which are directly related to the travel time of groundwater from the.

engineered facility to the environment. The 1000 year GWTT objective helps to assure that groundwater conditions are favorable, since a repository in compliance with the GWTT performance objective will be influenced by regional hydrogeologic processes (which are characterized by long travel times), rather tt -... any local, relatively fast-moving groundwater.

The apparent conflict between the terms " pre-waste-emplacement" and " path of likely radionuclide travel" is recognized. The staff intended that the concept of pre-waste-emplacement groundwater travel time and " path of likely radionuclide travel" be interpreted to mean the paths which radionuclides would -

b'e likely to take if they wefe released from the disturbed zone under pre-waste-emplacement cortditions.

Releases of radionuclides t'hrough groundwater pathways is limited by the three primary barriers:

(1) the integrity of the waste package and overpack; (2) the ability of the groundwater to transport radionuclides, irrespective of geochemical effects; and (3) the geochemical interaction of the radionuclide with the rock along the path of groundwater movement.

The present Position deals only with the second barrier.

1.1 What is Groundwater Travel Time?

Groundwater travel time was envisioned to be the time that it would take inert tracer particles released at the disturbed zone to reach the accessible environment under pre-waste-emplacement conditions. This travel time is often considered to,be synonymous with the travel time T calculated by the average i seepage velocity along the path s:

A.E. U T=f - ds (1) l 0.Z. "e -

where A.E. = accessible environment, D.Z. = disturbed zone, l

l

. U = tra Darc velocity along the path, and,

~ ' '

n.e = the effective porosity t

I

GWTT/ JUNE 86 is generally known as the seepage velocity; it is the apparent The speed term U/n' water in the open spaces in the rock. The travel time expressed by of the Eq.1 however, may not be the same as the travel time based on the transport of an inert tracer, particularly in a fractured medium. The bases for the differences are described below.

Transport of a non-decaying dissolved tracer in the groundwater can be expressed by a three-compartment model as shown in Fig. 2. The three compartments are; a) the mobile liquid phase (e.g., flow through connected pores and fractures), b) the immobile liquid phase (e.g., dead end pores) and c) the solid phase associated with the rock. This model can be succinctly represented by a material balance for the case of a dissolved tracer (Codell 1982):

n C + (n - ne ) * (1 ~ ") * "e div ( 0 grad C - U_ C) (2) e t

- +n e AC.+ (n - ne ) A0 * (1 - ") A0 '

where C = the concentration in the mobile' liquid phase, Q = the concentrltion in the solid phase, G = the concentration in the immobile liquid phase, n = the total porosity of the rock, the porosity of the rock in which the mobile water flows, n'==

D the dispersivity tensor representing molecular diffusion and mechanical dispersion in the liquid phase, U = the seepage velocity vector A = the decay rate Diffusion in the liquid phase is caused by the random motion of the water and solute molecules. Dispersion in the liquid phase is caused by deviations from the mean velocity vector U. Diffusion and dispersion alone cause a spread in the travel time predicted from Eq. 1.

Another, potentially more important~ mechanism which could lead to the spread in l

predicted travel time is the partitioning of the tracer between the three I compartments. The - ' leliquidphaseoccupiesa'portionoftherock,n$t affective porosity" (although this term is somewh

-usually known as t ambi The immobile phase occupies'the fraction (n - n ) of the rock.

The'guous). solid phase occupies'the fraction (1 - nl of the roc'k. Therelationship between'the'three compartments is key.to undtirstanding the movement of the tracer through the groundwater. An otten-used simplification of the equa. tion

.is t.o assume that the ccr. centration of~the' constituent in all three e

i

GWTT/ JUNE 86 compartments is in equilibrium, and that the solid phase concentration is related to the liquid phase concentration by a constant, K:

Q = KC (3) 1-n where K=- n - n* + Kd R3 (4) 1-n e 1-n e K = the distribution coefficient, ml/gm, and Rd = the real specific gravity, gm/ml.

s In this case, a commonly used form of the transport equation can be derived (5)

R d t = Div ( 0 Grad C - UC) + R AC d Wnere R is .the reta.rdation coefficient, wirich expresses the velocity at which d

the tracer is being transported relativa to the average seepage velocity: .

• (0)

Rd" d For a non-sorbing dissolved tracer, K d would be zero. Note that the retardation coefficient for this case is not equal to unity, as is often considered to be the case; it is always equal to or greater than unity. This deduction reflects the assumption stated above that the mobile and immobile phases are in equilibrium. Actually, the concentration of an inert tracer in the immobile phase can only approach and not reach' equilibrium with the mobile water. This equilibrium is limited by the rate of transport between the phases, and depends on a number of factors, including the conditions under which the tracer is moving and the diffusion coefficient of the tracer in the water. In other words, if GWTT is supposed to represent the travel time of inert tracer molecules from their points of release along the disturbed zone to the accessible environment, then the diffusive properties'of the tracer are important. Processes that control the transport of tracer between the mobile

.ind immobile water phases havA been called " matrix diffusion" (81encoe and Grisak, 1984).

If transport between the mobile and immobile phases is insignificant, R would approach unity and the GWTT could be based on the average seepage vel ocTty of

.groun:Nater along the catri, T, as determined. from Eq.1. C.onversely, if transport between the immobile and mobile phases is relatively fast, then the e

GWTT/ JUNE 86 retardation factor R w uld approach n/n , eand the GWTT would be greater than d

T.

Tracer particles considerably larger than molecules will not exhibit the same diffusive behavior as molecular tracers, and will be transported at a speed more typical of the average groundwater seepage velocity. The difference between the apparent velocity of a diffusive tracer.and the apparent velocity of a non-diffusive tracer can be dramatic. For example, Cathles (1974) described a dual tracer experiment in a fractured granitic rock, where the GWTT for a non-diffusive tracer (0.5 micron silica spheres) was several orders of magnitude greater than that for a non-reactive diffusive tracer (salt water).

The effect of matrix diffusion is probably most significant in media with high matrix permeability, especially where groundwater movement is very slow (Blencoe and Grisak, 1984).

It should be noted that the tracer does not cause matrix diffusion. The process proceeds because of Brownian motion of the molecules. Both the tracer and water molecules are diffusing. The magnitude.of the diffus_ive. flux in the.

matrix is proportional to the diffusion coefficient of~the molecule in water raised to a power less 9 an 1. The diffusion coefficient is an intrinsic property of the solute' idlecule in the solvent 5 (*2g., water). The self-diffusion of water is estimated to be 2.7x10 cm /sec. The diffusion coefficient of nearly any molecular or ionic solute is well within ar. order of magnitude of this value. Many common electrolytes such as chloride are within a factor of 2. Therefore, the effect of the diffusion coefficientThis on water is an must be fairly close to that of most common dissolved tracers.

important point, because in a situation where matrix diffusion is an important factor in the transport of a tracer, it would also be important in the transport of the water. In other words, the GWTT based on Eq. I does not necessarily ' account for the fact that the water arriving at the acc'essible environment is not all the same water leaving the disturbed zone, but may contain an amount of water exchanged with the immobile water along the pathway.

This fact tends to support the notion that diffusion can rightfully be included-into the concept of GWTT, even though the inclusion of diffusive effects may preclude the. isolation of the absolute " fastest path" groundwater travel time.

Molecular diffusion decreases with increasing size of the tracer particle. In the _ case of the 0.5 micron silica spheres in the tracer experiment-8 I Cathles 2

(1974), diffusion coefficient can be. estimated to be roughly 1.0x10 cm /sec (CRC 1986), which.is three orders of magnitude less than the diffusion coefficient of molecular tracers. Therefore, it is not suprising that this tracer.was not affected by diffusion into ,the matrix. -

GWTT/ JUNE 86 It can be argued that the effect of diffusion into the matrix is taken into account in Eq. 1 through the effective porosity term, since n is usually determined by means of a tracer experiment. Measurementsc'Theeffective porosity are difficult, however, and dependent on the experimental procedures and tracers used. For example, the effective porosity determined by a tracer test in a well will be sensitive to the rate at which the groundwater is moving. If the test is being conducted under conditions where the velocity has been maximized, such as in a two well test, the tracer might not diffuse into the matrix to the same extent that it would under unpumped conditions, whereupon the effective porosity would be underestimated. Although usually represented as a scalar quantity, effective porosity appears to be a tensor (i.e., directed) quantity in fractured geologic media (Endo and Long, 1984).

The radically different behavior of diffusing and non-diffusing tracers in some media may lead to different interpretations of the GWTT position. While ,

the staff intended the definition of GWTT to be the travel time for non-reactive tracers, th'e effect of tracer diffusion was not widely recognized.

- ~

The consensus in the hydrogeologic community is that GWTT should be based on

~

the average seepage velocity and should not consider matrix diffusion. There are several factors which tend to support this point of view:

The apparent retardation caused by matrix diffusion is conservatively neglected if it is assumed that the tracer particles travel with the mean seepage velocity, except to the extent that tracer diffusion was a factor in the determination of the effective porosity.

Transport of particulate or colloidal radionuclides would not be affected significantly by matrix diffusion. These larger particles would travel at a velocity close to the average seepage velocity, if not affected by mechanisms such as sorption. In addition, phenomena such as anion exclusion can reduce the ability of certain dissolved species to diffuse into small pore spaces, thereby reducing the importance of matrix diffusion.

.The mechanisms of matrix di.ffusion are difficult to evaluate in the field.

Without direc* easurements of the appropriate coefficients," estimates of the effect of atrix diffusion would have to be based on mathematical models which are largely untested, using parameters which are difficult or impossible to subst.antiate.

On th easis of the above points, and'in keeping with the Commission's s'tated posit:on that GWTT should be a simple measure of the overall quality of the

GWTT/ JUNE 86

_g.

repository, the staff will proceed with the understanding that GWTT is based on the travel time of non-diffusive, inert tracer particles which move with the average seepage velocity, and encourages the applicant to follow this approach.

Groundwater travel time also could be interpreted to consider the exchange of flowing and immobile water by diffusion. The staff would entertain arguments for travel times based on inert, diffusing tracers if ample justification is provided. Alternatively, such arguments for matrix diffusion might be used to support the satisfaction of the GWTT performance objective in the case where the GWTT based on the average seepage velocity is calculated to be less than 1000 years.

The staff has endeavored to present in this Technical Position a workable definition of the pre-waste-emplacement groundwater travel time objective to be used for HLW repository licensing. The definition will assist the staff in evaluating compliance of a specific site with the performance objectives of 10 CFR 60. However, this Technical Position is however intended to constitute guidance only. It reflects the Staff's interpretation of the GWTT objective, -

but does not prevent the Applicant or others from advancing alternative interpretations. This GTP is not intended to be a prescriptive guide to conducting field tests. Such guidance is beyond the scope of this document.

It is instead a guide to de' fining the GWTT objective, and presenting the results in a defensible manner.

2.0 Interpretation of GWTT Objective Compliance with the GWTT objective in 10 CFR 60.113 (a) (2) requires carrying out the following steps:

Properly identifying and considering the pre-waste-emplacement environment and its' potential spatial and short-term temporal variabilities';

Identifying the' fastest path of likely radionuclide travel; and Calculating the appropriate travel time along this path.

2.1 Pre-Waste-Emplacement Environment

(

l Pre-waste-emplacement pertains to conditions which exist prior to significant disturbance of the geologic or hydrologic setting by construction o.r major testing activities capable of seriously disturbing the geologic setting.

Restriction of the GWTT requirement to pre-disturbance conditions is in accord with the original intent.of 10 CFR 60 to establish a straightforward cr.iterion which is easily. defined and determined.. The present position does not deny the importance of post.-waste-emplacement effect's. ' Evaluation of groundwater and i

I l

GWTT/ JUNE 86 radionuclide movement under post-waste-emplacement conditions will be required as part of the demonstration of overall compliance of the repository with the EPA standards (40 CFR 191) as implemented by NRC.

The site must be characterized and understood to the extent that the fastest

- path of radionuclide travel (S2.2) can be identified and the ground water travel time (S2.3) can be determined. The determination of.GWTT will be for present day environmental conditions only. Short-term changes to the environment, (e.g., tens of years) which can be reasonably inferred from records in the vicinity of the site, such as cycles of wet and dry years, local flooding, changes in groundwater and surface water use and irrigation practices, and any other factors that may alter hydraulic heads should be factored into the conceptual model for determining GWTT. Groundwater systems which have been demonstrated to exhibit significant transient behavior for the period of record may have to be modeled in a time-dependent rather than a .

steady-state manner to demonstrate compliance with the GWTT requirement. The determinations do not have to take into account the long-term projections *

. ' (e.g. , thousands of years) ' of changes to the physical setting of the repository, such as earthquakes, changes t.o global climate, major changes to surface morphology or u;e'o,f groundwater and land. If present-day conditions have varied markedly over the period of record, the investigator must question whether inappropriate credit is being taken for excessive groundwater travel times caused by these variations. For example, if a cone of depression has formed as a result of large groundwater withdrawals, it could reduce or even reverse the direction of an unfavorable hydraulic gradient. In this case, it would be prucent to consider the effects of an otherwise-likely hydraulic gradient corrected for the effect of the cone of depression. The rationale behind this philosophy is to avoid the appearance of taking credit for

~

processes for which there could be no assurances of long-term reliability, e.g., continued groundwater withdrawals maintaining the favorable gradient.

2.2 Identification of fastest path of likely'radionuclide travel The paths from the disturbed zone to.the accessible environment are to be described in'a macroscopic sense; e.g., aquifers. In crystalline rocks, paths probably will consist of fractured, weathered or brecciated zones.. In porous media, paths gener will consist of layers of permeable rock. Paths may

'also consist of fr. . red zones in consolidated non-crystalline rocks. Several examples of paths for generic repositorie's are covered it) Appendix B.

Several alternative conceptual models for a iepository site may be proposed, each of which.might determine a different path for radionuclide ' transport. For example, borehole inforr.ation in a' saturated zone might indicate the presence of permeable zones, but the investigator may be unable to determine whether or

GWTT/ JUNE 86 not these zones were connected in such a way that they constitute a path. Such information could only be gathered by multiple-well tests on the length scale of the order of the dimensions of concern (e.g., hundreds to thousands of meters). The analysis of GWTT therefore should consider all paths for radionuclide transport defined by alternative conceptual models, unless they can clearly be demonstrated to be unlikely, preferably through direct measurements of hydrogeologic properties at the site. Data collection must be focused on identifying and quantifying paths so that a high degree of confidence is provided that potentially faster paths have not been overlooked.

Selection of the proper drill and test program for the conceptual model is a key element in this process.

2.3 Groundwater Travel Time along the Fastest Path Groundwater travel time as' envisioned in this Position paper is a distributed variable rather than a fixed quantity. It can be quantified as a cumulative frequency distribution of the times of. travel for inert tracer particles released fro.m the disturbed zone reaching tTie accessible environment.along:the fastest macroscopic path.- Several reasons for the distributed nature of the groundwater travel time are: s Uncertainty - Measurement error or sparseness of data necessary to characterize the site adds uncertainty to the travel time estimates for the tracer particles. Site data must always be collected and interpreted in terms of a conceptual model. An invalid conceptual model will lead to an incorrect interpretation of the data. Drilling a well tc, an improper depth relative to a valid conceptual model or performing'an inappropriate test are typical of common errors in measuring and interpreting field data. Such errors enhance uncertainty in groundwater travel time. <

Distributed source - The disturbed zone and t(cessible environment are defined as surfaces rather than points. 7&ier particles released at different points along the disturbed : n m 'd reach the accessible environment at different times.

. Spatial variability of the. properties of the medium (e.g. , non-constant thickness of the hydrostratigraphic units, inhomogeneous hydraulic conductivity and porosity).

o Temporal variability - Hydrologic and hydrogeologic data within recorded history of the site might indicate that the. groundwater velocities are fluctuating. Temporal variations over the time period of concern are not'

~

egected to be'an. imp'ortant consideration on the regional scale for saturated flows. These variations ~might be important' at sites built in

GWTT/ JUNE 86 unsaturated media, however. For example, it is conceivable that a period of unusually heavy precipitation for several years (unrelated to a gicbal climatic change) could increase hydraulic heads, decreasing travel times along a normally slow pathway. A transient GWTT should be weighted according to its frequency and duration. In addition, a path which changes direction or length over time as a result of variable fluxes of groundwater should be considered to be a single path for the purposes of GWTT calculations. This allows the low probability, fast GWTT's to be fairly weighed with the high probability, slow GWTT's.

The estimation of GWTT must accommodate spatial variability, temporal variability and uncertainty. GWTT can be presented as a distribution for each of the paths in terms of a Cumulative Distribution Function (C0F), an example of which is shown in Fig. 3. This CDF will combine all spatial variability, temporal variability and uncertainty of the GWTT into a single curve for each of the paths. The C0F itself however is assumed to contain no uncertainty. It is important to note that the CDF does not deny the existence of uncertainty; -

all uncertainty is incorporate'd into the CDF. Spatial and temporal variability

- and uncertainty can theofetically be treated separately, but grouping them both into a single C0F has t;> advantage of simplicity. Compliance with the 1000 year objective would be Jemonstrated if it could be shown that any tracer particle leaving the disturbed zone has a (100-X)% or greater probability of arriving at the accessible environment in a time greater than 1000 years, where X is a small number. The basis for choosing of X% is presented in 62.4.

The 15th percentile is shown in the figures for illustrative purposes only.

Overall, the identification of likely paths and reliable estimation of GWTT is strongly dependent on the adequate characterization of the hydrogeologic conditions between the disturbed zone and the accessible environment.

Conceptualizations of paths will likely be simple during the early reconnaissance phase of site characterization. Continued characterization activities will produce more detailed and realistic conceptualizations of hydrostratigraphy and geologic structure, which will lead to improved estimates of GWTT. One of the goals of field experiments is to narrow the GWTT distribution-by eliminating as much of the uncertainty as possible; i.e.,

increase the steepness of the CDF. The criterion discussed in S2.4 is sensitive to the steepness of the GWTT distribution, thereby providing an incentive to reduce uncertainty. Further discussion of the concept of GWTT and procedures for its calculation are presented ire Appendix A.

GWTT/ JUNE 86 2.4 Rationale for Choice of the Percentile of the Cumulative Distribution Function (CDF)

In applying 10 CFR 60.113(a)(2), the staff recognizes that groundwater travel time along the paths defined for each conceptual model can be represented by the Cumulative Distribution Function (CDF) rather than a single value, because of uncertainty in understanding the hydrogeology of the site, measurement errors, temporal variations in flow, multiple particle trajectories and a spatially-distributed source. (A single-valued GWTT determined from conservative models and coefficients would also be acceptable to demonstrate compliance with the GWTT objective). Uncertainties in estimating these phenomena are expected to cause the GWTT distribution to span as much as several orders of magnitude. Phenomena leading to the distributed nature ~of the predicted GWTT are elaborated in Appendix A.

It is difficult to deal directly in terms of a distribution when stating performance criteria. I't is often useful instead to specify a scalar norm of the distribution; e.g., the mean, median, or s.ome percentile of the CDF. TM

" fir'st particle" approacif is a norm based on the zerceth percentile of the CDF.

This approach has a certain, appeal because the travel time of the first particle is obviously the "' fastest". There are some serious shortcomings to this approach, however. Consider for example the two curves shown in Fig. 4 which represent the cumulative distribution of GWTT for two sites. In this example, consider further that both sites are perfectly characterized, and that any variations in travel time are due to sp.atial variability of the medium or the distributed nature of the accessible environment and disturbed zone. These curves could represent breakthrough curves from tracer experiments at the sites. In Case 1, the distribution indicates a single groundwater travel time, t'. In case 2, there is a distribution of travel times with a minimum of t'.

A zero percentile criterion would treat both cases as equals, whereas case 2 is obviously superior in terms of post-emplacement repository performance, if all other things are considered equal. The choice of a higher percentile would distinguish between Cases 1 and 2 and give credit to case 2.

A choice for the percentile which is too high, say the median, would be undes.irable because it.may be in. sensitive to the variance of'the GWTT distribution. Thir.:cncept can be demonstrated for the hypothetical example

' depicted inrFig. 5. The two CDF. curves of GWTT in this figure have the same median of 1000 years, but different variances. They may, for example, represent different sites. The curves may a]so represent a single site for which thE data have no experimental bias, but at different points in the site characterization process. Under the median GWTT criterion, sites which exhibit a wide variance of the travel tims dis'trib'ution for reasons such as great.

spatian variability, an inadequate conceptual model, inadequate drill and. test

GWTT/ JUNE 86

_ 14 -

plan, or measurement uncertainty, would be treated as equals. A smaller l percentile justifiably favors the site which has the smaller variance in the GWTT distribution. If the wider variance is due to quantifiable uncertainty (e.g., lack of c ta), the smaller percentile would serve as an incentive to further characterize the site. A smaller percentile favors the site which has l the smaller variance in the GWTT distribution.

l i The percentile for the CDF as the criterion for GWTT is unspecified presently, but the rationale from the above two examples suggests that a value greater than a few percent and considerably less than 50% would be desirable. The determination of the percentile for the GWTT criterion also should be based on considerations of " reasonable assurance". Licensing considerations to be made in connection with GWTT involve substantial uncertainties, some of which are unquantifiable (e.g., those pertaining to the correctness of the models used to evaluate GWTT). Such uncertainties can be accommodated within the licensing process only if a qualitative test such as reasonable assurance is applied for the level of confidence that the numerical performance objective is expected to '

achieve. Both the quantifiabls uncertainties incorporated in the GWTT distribution and the ungdantifiable uncertainties which'are not included must be considered together in r,eaching a finding of reasonable assurance. It might, for example, be proper to select a different percentile criterion for a relatively well-understood, easily modeled site, whera unquantifiable uncertainties are small, than would be appropriate for a site with larger unquantifiable uncertainties. Stated another way, selection of the percentile criterion is a qualitative judgement that is part of a larger set of judgements necessary to reach a finding of reasonable assurance that the performance objectives will be achieved.

The applicant is not required to generate a detailed CDF of the GWTT distribution. A simplified approach would be acceptable, provided that achievement of the 1000 year GWTT objective could be demonstrated with reasonable assurance. Such a simplified approach could for example, define a

~

conservatively short path along which the travel time of 'a single particle

- could be calculated using Darcy's law in one dimension, with conservatively chosen coefficients of hydraulic conduct n :ty, gradient and effective porosity.

2.5 Special Considerations for Unsaturated Media Groundwater movement through unsaturated media for pre-was'te-emplacement conditions differs from that of saturated media in a number of important ways:

1. In a medium unaffected by boundaries, the gradient'and th'erefore the

- dirsc-tion of unsaturated flow is pred.ominantly vertical, although confining features such as aquicludes, faults and dikes can .

'~ --

GWTT/ JUNE 86 complicate this general picture. Saturated flow is primarily i horizontal, except in areas of rccharge or discharge.

2. Unsaturated flow tends to be more responsive to episodes of recharge than does saturated flow.

3. Unsaturated flow parameters are highly nonlinear, and depend on the degree of saturation of the medium. This nonlinear dependence could also lead to changes in the flow trajectories for differing levels of saturation, e.g., saturation of fractures or creation of a perched water table.

The transient nature of flow in the unsaturated zone causes a certain difficulty in defining groundwater travel time. A conceptually important distinction must be made between an episodic recharge event in an unsaturated .

medium and nearly-steady groundwater flow in a saturated medium. Even though little downward, flow through an unsaturated medium may occur normally, it is

' conceivable that unusually heavy precipitation over short periods could lead -

to small travel times dufing those periods., at least through the unsaturated portion of the medium. ,The, definition of GWTT as a cumulative distribution function allows the low probability, short travel time events to be weighted fairly with more-typical travel times.

Travel times would be weighted according to the intensity, frequency and duration of the event. The travel time distribution could be estimated, for example, from a transient groundwater flow analysis, coupled with the transport of hypothetical tracer particles released at constant time intervals at points along the disturned zone. The cumulative frequency distribution of predicted GWTT in this case would incorporate time variability of recharge, as well as the spatial and temporal variability in path lengths. It should be noted however, that the measurement of parameters for unsaturated systems is considerably more difficult than for saturated flow, which may impose increased conservatism on the uncertainty analysis.

3.0 Summary 'and Statement of Regulat'ory Position 3.1 Summary ,

Groundwater travel time is a measure of t'he merit of the, geologic setting of a.

high level waste'reposit'ry. o The Staff recocnizes that alternative conceptual models of the site may be defensible Mcause'of the inability to characterize

~

it completely.with the available data. This inability may lead to a numb.er of potential paths forclikely radi-onu'clide travel. The groundwater travel time .

along the paths will be a distributed quantity because of spatial variability,

a G A Surface ~

- - - - - - {, - - ter Table Wa i

Unlikoty Path Aquifer A Fracture Ukely Path Aquifer B Q

/ N Ukely Path Aquifer C / ',

-A

~

g

/

Underground Facility \  %

v: / w ///// w //////////a Disturbed Zone

\ / -

Aquifer D \ /  :

\  % # Ukely Path l

r

(

Figure B1 - Paths of Likely Radionuclide Travel - Saturated Media l

l

s Ground Surface Y/N//$//W//$/l4N//S//Wll W//N//W//$//,

Rock Unit A Unlikely Path Possible Perched Water -

Disturbed Zone Fracture Rock Unit B

['" ~ ~ %]

\ /

Possible Perched Water N

%.__s' Ukely Path Rock Unit C Possible Perched Water r 1, . 2 Rock Unit D U

vn.

Regional Water Table Ukely Path l

~

Figure B2 - Paths of Likely Radionuclide Travel - Unsaturated Media

GWTT/ JUNE 86 temporal variability, the distributed nature of the disturbed zone and accessible environment, and model or data uncertainty. Groundwater travel time should therefore be represented as a cumulative frequency distribution, i although a single-valued GWTT would be acceptable if it were derived from appropriately conservative models and coefficients. The " pre-waste-emplacement groundwater travel time along the fastest path of likely radionuclide travel" should be represented as a percentile of all travel times contained in the Cumulative Distribution Function for each of the potential paths identified.

Pre-waste-emplacement pertains to conditions at the site prior to any significant disturbance of the hydrological or geological setting such as construction activities or the effects of radioactive waste, and whose spatial and temporal variability can be reasonably inferred from historical records at or near the site. Testing activities capable of altering the pre waste-emplacement environment should be taken into consideration. The analysis must take into account any information pertaining to preferential points of release from the Disturbed Zone, and consider reasonably likely gonceptual models which might lead to transport through other paths.

3.2 Statement of Positiod It is the staff's position'that an acceptable means of demonstrating compliance with groundwater travel time performance objective of 10 CFR 60.113 is as follows:

1. Determine the paths of likely radionuclide travel for the site as described in 62.2 and Appendix B.

2. For each of the paths, determine the pre emplacement groundwater travel time as described in 62.3 and Appendix A.

3. Select the fastest such travel time so determined.

4

GWTT/ JUNE 86 Appendix A - Calculation of the Groundwater Travel Time (GWTT)

A.0 Introduction This section provides guidance on how to calculate the GWTT distributions for each of the identified macroscopic paths defined by conceptual models considered. Section A.1 describes the utility of hypothetical tracer particles and uses the concept to illustrate why there would be a distribution of travel times rather than a single value.

Section A.2 describes several mathematical modeling schemes which could be used to calculate the GWTT distribution. Section A.3 discusses the various methods for estimating parameters, quantifying their uncertainties, and choosing the input for the mathematical models on the basis of the available data. Section A.4 discusses a particular approach to calculating the GWTT distribution by applying a Monte Carlo sampling schame to a deterministic mathematical model.

• Finally, Section A.5 describes how simplified analyses may be used in some

- cases to satisfy the GWTT' performance objective without having to resort to complicated analyses. , ,

A.1 Travel Time Distributions It is useful in subsequent discussions to think of released tracers as consisting of discreet particles, although it should be recognized that these are figurative rather than real. A single " particle" leaving the disturbed zone would generally follow the path traced by the moving groundwater, except for phenomena such as molecular diffusion and chemical interaction. Molecular diffusion would cause random motion to be added to the trajectory of the particle, allowing it to move into areas such as pores with little or no net flow. Chemical interaction with the surrounding rock would cause the radionuclide particle to leave the groundwater'and become fixed temporarily or permanently in or on the surface of the rock. Because of a consensus in the hydrogeologic community that the GWTT should be calculated using the average pore velocity, we restrict all subse'quent discussion to transport of the tracer by this velocity, without considerations of molecular diffusion or geochemical effects. Geochemical effects are covered in another regulatory position (Bradbury, 1986).

Along any "riath" as defined in 62.2 and Appendix B, natural spatial variability'in the properties of the medium will exist; e.g. porosity, hydraulic. conductivity. The tracer particles moving in the groundwater will follow trajectories governed by.the hydraulic. properties of the medium and.the ,

drivi~ng forces. The more uniform the medium, the more' parallel will be the

GWTT/ JUNE 86 trajectory of the tscer particles. Conversely, the tracer particles in a heterogeneous medium may diverge from their neighbors for certain types of heterogeneity, following trajectories of least resistance which are not necessarily the shortest trajectories. Travel time distribution caused by non-uniformity of the medium is generally known as mechanical dispersion.

Unsaturated media are somewhat more complicated than saturated media. Not only the speed but the trajectory of tracer particles could change with time as a result of a change in boundary conditions or flow parameters in the unsaturated case. For example, in a fractured porous medium, conditions of high infiltration could cause certain fractures to fill with water and establish paths not present during periods of lower infiltration.

A.2 Mathematical Representation of the Repository and its Environment Analysis of the GWTT for any real repository must depend on observations of hydrogeologic data at the site. These' data must be collected with the hppropriate' drill and test program, based on a valid conceptual model for the site. Artificial tracers are useful in some cases (e.g., determining the effective porosity and thickness), but the time and distance scales are too great for direct characterization of the GWTT by such methods. Naturally occurring isotopes and those produced from atmospheric weapons testing and 4

nuclear reactors can be useful for groundwater dating to support estimates of travel time distributions for real sites. Such techniques should be used whenever possible. Investigations must usually resort to mathematical models of the repository for predictions of performance.

Once conceptual models and drill and test plan programs have produced the appropriate data, values of travel time from the disturbed zone to the accessible environment are usually obtairied from mathematical models consisting of the equations governing the hydraulic potential, flow of groundwater, and transport of a tracer. There are many models for groundwater flow in various media which are based on the equations at steady state or transient conditions in one, two or three dimensions.

Dete?ministic models consist of equations whose solution is based on the assumption,that hydrogeologic properties, initial conditions and system geometries'are known. Uncertainty and variability'of the data are sometimes' taken into account ~by obtaining many deterministic solutions,. each one based on a different realization of the hydrologic properties.

~

Such techniques are generally known as "Mo'nte Carlo" simulations. The results obtai'ed n by applying many andom realizations (but~ chosen from a data base collected in,a well' anceived 'and carefully ~ executed experimental program) of the ' parameter sets to the mathematical model can then be analysed statistically in order to

'v-, r.m., . - - -w - - - - . - -- , -- -

GWTT/ JUNE 86 estimate the travel time distribution (Smith and Schwartz, 1980, Smith and Schwartz,1981, Cli f ton,1984). Alternatively, the model may be used with conservative values of the input parameters in order to obtain conservative estimates of the GWTT.

Models that are exclusively stochastic deal with the variability and uncertainty of the data in a more direct way. The coefficients and/or variables of the equations are treated as random processes rather than deterministic quantities. The partial differential equations (PDE's) are solved indirectly in terms of the moments of the dependent variables (e.g.,

mean and variance). This technique has the advantage of requiring only one solution rather than the numerous Monte-Carlo solutions required for the deterministic approach. Stochastic approaches to modeling are at a much less developed state than Monte Carlo techniques, although it is an area of rapid development. The stochastic approaches have been used to estimate means and variances of fields such as head (Mizell et.al.,1982) and concentration (Gelhar and Axness, 1983). They apparently have not yet been used to calculate '

directly such spatially integrdted properties as GWTT.

A.3 SiteCharacterizationIromFieldData Four levels of parameter quantification for site characterization are: (0NWI, 1983):

Bounding value estimates - the range of possible values of the parameter.

These are usually extreme values of parameters which do not take into account the correlation between different parameters.

Best estimate values - a single value of the parameter which is based on field measurenients, laws of physics, expert opinion, or combinat' ions of the above.

Interval estimates a bounding estimate which has been tempered by field data, laws of physics, expert opinion or combination of

~

Correlations among the parameters may'be taken into account, the above.

e.g., relationships between porosity and hydraulic conductivity.

Probability density functions (PDF's). A function in which the probability that the parameter exceeds _a certain value is known.

GWTT/ JUNE 86 The PDF is of course the most informative quantification of the parameter, but requires the most knowledge of the site. In those cases where the data are too sparse for direct inference, rough estimates of the variability of parameters in the field may be inferred during early phases of the site characterization from expert opinion and observations of the distributions of the parameter in similar rock masses. For example, parameters such as hydraulic conductivity are often observed to follow a log-normal distribution, and conform to certain models of the spatial covarience function (Neuman, 1982). Expert opinion is not a substitute for field data, however.

Both data gathering and modeling depend on the establishment of a valid conceptual model for the site. The conventional quantification of aquifer hydrogeologic parameters (i.e., transmissivity, storativity, hydraulic conductivity, effective thickness) is based on a framework of established assumptions. Significant departures from the conceptual model will yield .

nonrepresentative values of the quantities sought.

Errors may be ihtroduced because the collected data are misinterpreted. For-example, water levels determined by a steel tape may be interpreted incorrectly because of temperature or salinity differences in the wells (ONWI, 1983).

Another example might be trie misinterpretation of transmissivities (or hydraulic conductivities) from a drawdown test caused by phenomena such as leakage from another aquifer or a boundary of low permeability (e.g., fault or dike) within the cone of depression. In these cases, the principal cause of error is once again the inadequacy of the conceptual model.

The scale over which the data are collected is an important factor. For example, hydrogeologic properties determined from point tests (e.g., slug test) might be valid for characterizing the medium only a few meters from the borehole. It may be difficult or impossible to substantiate hydraulic continuity between the boreholes which determine the paths of groundwater flow without multiple-well tests on a scale similar to the dimensions of interest.

Lacking multiple-well tests, conservative assumptions about the connectedness of units between boreholes must be used. The fallibility of multiple-well tests must be recognized, however. The gradient in the medium may be highly distorted by pumpina in order to complete the measurements in a reasonable length of time. F, ermore, multiple-well tests do not appear to be a a viable procedure tuations such as unsaturated flow. Therefore analyses must be based on point measurements.

Oveiall discussions of parameter estia.3 tion Saould include the reasonableness, within the known hydrogeologic regime, of.all key assumptions. The likel.y effects.of erroneous assmptions en parameter estimation and GWTT calcu1'ations

i GWTT/ JUNE 86 should be stated. The staff recognizes the importance of expert opinion in providing defensible interpretations of all types of aquifer field testing.

A.3.1 Treatment of Uncertainties in Site Characterization Many possible sources of uncertainty in the characterization of the site for determining GWTT can be identified. Among the most likely sources are:

Measurement Errors in Data. These errors may be procedural (e.g. human) errors or systematic errors caused by faulty or improperly calibrated instruments. The staff recommends that these types of errors be minimized and quantified by standard techniques such as calibration, redundancy, and by using several independent ways of obtaining the same data.

Validity of Analytical Assumptions for Site Simulation. The simulation of flow and transport may not be representative of the physical system because of a poor. understanding of the basic-physical phenomena or oversimplification because of computational expediency. For example,..the equivalent porous media (EPM) approach is often used to represent a fractured medium as a ' porous medium. The EPM approximation may be useful only for large scale transport, and not valid at scales in which the effects of individual fractures are important (Long et.al., 1982). Some investigators question the validity of the EPM approximation for properly modeling transport along the direction of fracture orientation regardless of scale (Endo,et.al., 1984). The validity of the conceptual model for simulating the site is closely coupled to the conceptual model used to interpret site data.

The staff recommends that alternative conceptual models be proposed and tested in order to determine the sensitivity of the results to the choice of the conceptual models which can reasonably be constructed from the available data.

Interpretation of Sparse Data. The temporal and spatial distribution of hydrogeol'ogic field data are always less' dense than desired. In the case of point measurements, c.onditions between points must be inferred, either

. b'y fitting a surface through the data points, or by using a physically realistic interpolation model. Sophisticated interpolation. methods such as Kriging (e.g,, Matheron, 1971) yield an estimate of the variance a well as the mean of spatially varying data. Mathematical models may be adjusted manually in, order to produce a best fit to the'available data

~

(e.g., Fogg, 1978, Mercer and Faus.t, 1980). In some cases, the fitted parameter may be determined automatically Mithout the need for manual

GWTT/ JUNE 86 adjustment. Statistical inverse methods are available for fitting the hydraulic conductivity to head data in saturated media, and also for calculating the variance of the hydraulic conductivity (e.g., Neuman and Yakowitz, 1979, Hoeksema and Kitanidis, 1984).

Computational Errors. Since computer codes must be used extensively,

, errors may be introduced because of mathematical approximations (e.g.,

element size, step size) and intrinsic errors such as those produced by round-off and truncation. Computer codes should be verified with analytical solutions, validated with real field data, and compared or benchmarked with other similar computer codes (Silling, 1983). The sensitivity of the results to node size, time steps, grid orientation, or other parameters and assumptions should be tested by computational experiments.

A.3.2 Determination of the Input to the Model,.

Once the conceptual model has.been code'd into a, computer program, the -

computations must be performed with parame,ters inferred from the available data in order to generate t The types and quality of data available will determi( nowGWTT distribution.

r the computations are to be performed. For example, if only a few data points are available for a particular parameter, a conservative estimate of that parameter may have to be made and carried through the calculations. With more data, a mean and variance of the parameters can be calculated and used with a simple sampling approach (ONWI, 1983). If the site is characterized correctly with sufficient data, spatially varying properties of the parameters can be generated, thereby permitting conditional simulations or stochastic models to be applied.

The GWTT computed using this general guidance will be sensitive to the degree of characterization of the site. That is, investigators of poorly characterized sites will be forced to use conservative or at least overly wide estimates to represent the distribution of the input parameters. Sites that have been tested with valid drill and test programs based on defensible conceptual models will facilitate the development of a more defensible GWTT distribution function. The GWTT distribution with smaller variance is preferable for the reasons stated in S2.4 of the Position.

A.4 Estimating GWTT from Deterministic Models with Randomly-Generated Input The GWTT' distribution can be calculated from multiple runs of deterministic models, with each run made for a realization of the data which can be inferred for t.he site. In the steady-state'5aturated flow examp,le, each realization of .

the data requires the solution of the hydraulic head and velocity field. This

GWTT/ JUNE 86 solution generally is accomplished by solving the partial differential equations (PDE's) using techniques such as finite differences or finite elements. Once the velocity field is known, travel time distributions can be calculated by simulating the release of tracer particles from single or multiple locations along the Disturbed Zone, then counting their arrival times as they reach the accessible environment.

A.4.1 Treatment of Spatial Variability A large part of the variability of GWTT is caused by spatial inhomogeniety of the. parameters which determine groundwater movement, particularly hydraulic conductivity and effective porosity. The motion of hypothetical tracer particles leaving the disturbed zone will be determined by the gradient, the hydraulic conductivity and the effective porosity encountered along the path.

This variability alone will cause the paths of the particle leaving different parts of the disturbed z.one to diverge. The incompleteness of the data which determine flow paths within the hydrogeologic regime and the uncertainty due to seasurement. errors in field data add to the source of variability.

At least one method < conditional (or unconditional) simulation, has been applied to account for the' spatial distribution and uncertainty of field data in the determination of GWTT. This method has been applied to 2-dimensional steady state, saturated flow models for equivalent porous media (e.g.,

Delhomme, 1979, Clifton and Neuman, 1982), but it could be adapted to three dimensions (Mantoglou and Gelhar, 1985). The procedure is outlined below for the 2-dimensional, steady state case (Clifton, 1984):

a. Determine Spatial Variability and Uncertainty of Data Field data for hydraulic conductivity and porosity are collected, and' evaluated by methods of statistical inference in order to determine their spatial covariance and drift, which are measures of the variability of the property in space, and the " nugget effect," which is an indication of the measurement error or uncertainty. Expert judgement based on prior knowledge of the properties of rocks in similar formations may be useful

..in estimating the proper covariance models to apply to these data in this step (Mantoglou and Gelhar, 1985).

b. Generate Realizations of Data Random fields of the model parameters are re generated from the spatial covariances, drift, and uncertainties determined'in Step a, so that the 3tial covariances and auto-covariances of the new field or " realization" are identical to those determined for the original data. It is usually

GWTT/ JUNE 86 necessary to treat the random variable and boundary conditions as "ergodic", for which the principles of first and second order stationarity apply. Cross correlation of the data, e.g., correlation between effective porosity and hydraulic conductivity, may be taken into account in this step. Two widely-used procedures for generating these random fields are the " nearest neighbor" method (Smith and Freeze, 1979) and the " turning band" method (Mantoglou and Gelhar, 1985). The random fields can be forced to comply with the original data by a process known as

" conditioning;" otherwise, the parameter fields are " unconditional".

Conditional simulations reduce the variance considerably, but are generally worthwhile only if there are sufficient high quality data (Clifton, 1984).

c. Run Deterministic Model for Heads The random fields are used with a finite difference or finite element model to generate a steady state head and groundwater flow field under the

• influence of either, fixed'or random' boundary conditions.

d. Calculate Travel Times of Particles The trajectories of tracer particles are tracked from one or multiple locations on the disturbed zone to the plane representing the accessible environment. The travel time of the particles from their starting position to the accessible environment is recorded.

e. Generate Multiple Realizations Steps b through d are repeated numerous times in order to generate a large number of travel times for mul,tiple tracer particles so that their cumulative distribution can be drawn. The probability of each realization is taken to be equal to any other realization for the purpose of constructing the CDF. If more than one particle is released per realization, each particle is given equal weight.

A.5 Simplified Analysis The user is not required to generate a detailed ~C'DF of'the GWTT distribution.

A simplified approach would be. acceptable, provided that the 1000 year GWTT objective could be demonstrated'and the results could be demonstrated to be conservative. Alternatively, it has been shown that in the (cond.itional or unconditional) simulations outlined in SA.4.1, high spatial covarianc~e of hydraulic c5nductivity correlates with wider travel time distributions (Clifton,.1984). If the medium is assumed to be spatially uniform (i.e.,

__ .- -. . - =_ _

I GWTT/ JUNE 86 25 - ,

I infinite spatial covariance), then it must be assumed that all variations of the parameters are caused by measurement error. The cumulative frequency distribution has the greatest variance under these circumstances, which gives a conservative indication of the small percentile criterion for GWTT as discussed in 92.4 of the Position (but not necessarily the median of the distribution).

i J

e e.

, '6

$ 1 i

4 I

4 1 ..

p 8 9

e t

I 4

GWTT/ JUNE 86 Appendix B - Choosing paths of radionuclide travel B.0 Introduction The paths which radionuclides will follow from the disturbed zone to the accessible environment are to be described in a macroscopic sense. In crystalline rocks, paths may consist of fractured, weathered or brecciated zones. In porous media, paths will generally consist of layers of permeable rocks. Paths may also consist of fractured zones in consolidated non-crystalline rocks.

Several alternative conceptual models for the repository may be defensible, each of which might determine a different path for radionuclide transport.

Therefore, the analysis of GWTT should consider all paths for radionuclide transport defined by alternative conceptual models, unless they can clearly be demonstrated to be unlikely. Collection of data at the site must be directed t,oward identifying these paths, establishing the validity of 'the conceptual models for interpreting and simulating the hydrogeology, and making a reasoned determination that potentially faster paths have not been overlooked. -

Examples for several generic types of repository media are given in the sections below.

B.1 Repositories in saturated media High Level Waste underground facilities located in saturated media will usually be emplaced in a rock unit of low permeability. However, more permeable units may underlie, overlie or lie alongside the repository. Figure. B.1 shows several such possible pathways (note, however that it is not likely that both vertically upward and downward flows could occur at the same time). Some of these more permeable hydrostratigraphic units may intersect the disturbed zone.

While little movement of groundwater may occur in the host rock, factors may be present which could cause the movement of radionuclides from the disturbed zone to these more permeable hydrostratigraphic units. Transport between hydrostratigraphic units could occur through fracture or porous media flow

~

under the driving force of natural hydraulic gradients. Therefore the fastest paths should follow the hydrcstratigraphic units which have the highest Jroundwater velocities.

The choice of the' path need not be mechanistic; e.g. , it is not necessary to propose or calculate the mechanisms by which transport from'the hydrostratigraphic units intersected by the disturbed zone to the faster hydrostratigraphic units can occur (unless credit will be taken for the travel-time from the distprbed zone to the hydrostratigraphic units). It may be

GWTT/ JUNE 86 necessary, however, to determine whether such paths are "likely", or can be excluded from consideration. For example, an analysis could determine that the driving force would be inadequate to allow transport to other hydrostratigraphic units above a certain elevation, even if the necessary interconnections exist. Therefore, these hydrostratigraphic units will not be on "likely" paths and could be ignored. Even for "likely" paths, such analyses might facilitate quantification of travel times along the portion of the path between the disturbed zone and the assumed more permeable hydrostratigraphic unit.

B.2 Unsaturated media Definition of paths for repository sites in unsaturated media will differ from those in saturated media. If unsaturated conditions prevail, the direction of flow is likely to be vertically downward until the water table is reached. The path will be defined in terms of the vertical gradient, unless contrasts in hydraulic conductivity that le'ad to perched water exist. The possibility of perched water under reasonably conceivable conditions (e.g. , a period of high "

recharge events which is'not a major clima. tic change, but could occur under present climatic condi'4 1s) should be explored, even if such conditions currently do not exist 2; the site. Paths also should consider the possible connections of perched water to fractures or other structural features of the site which would allow short-circuiting of the unsaturated material in which the repository would be placed. Phenomena peculiar to unsaturated flow such as

" fingering" should also be considered. Examples of such paths are illustrated in Fig. B.2.

'6 9 e

4

GWTT/ JUNE 86 REFERENCES 81encoe, J.G., and G.E. Grisak, 1984, " Topical Review: Matrix Diffusion of Radionuclides in Rock - Groundwater Systems," ORNL/TM-9155, Oak Ridge National Laboratory, Oak Ridge, TN Bradbury,J.W., 1986, " Determination of Radionuclide Sorption for Assessment of High-level Waste Isolation", Geochemistry Section, Geotechnical Branch, DWM, USNRC Cathles, L.M., H.R. Spedden, and E.E. Malouf,1974, "A Tracer Technique to Measure the Diffusional Accessibility of Matrix Block Mineralization", in Solution Mining Symposium 1974, Editors F.F. Aplan, Society of Mining Engineers, American Institute of Mining, Metallurgical and Petroleum Engineers, New York Clifton, P.M., and S.P. Neuman, 1982, " Effects of Kriging and Inverse s Modeling on Conditional Simulation of'the Avra Valley Aquifer in Southern Arizona," Water Resdurces Research, Vol. 18, no. 4', pp. 1215-1234 Codell, R.B., K.T. Key, and G. Whelan, 1982, "A Collection of Mathematical Models for Dispersion in Surface Water and Groundwater", NUREG-0868, USNRC, Wahington, DC CRC, 1986, Handbook of Chemistry and Physics, Chemical Rubber Corporation Clifton, P.M., 1984, " Groundwater Travel Time Uncertainty Analysis -

Sensitivity of Results to Model Geometry, and. Correlations and Cross Correlations among Input Parameters,* Report no. BWI-TI-256, Rockwell Hanford Operations, Hanford WA Delhomme, J.P., 1979, " Spatial Variability and Uncertainty in Groundwater Flow Parameters: "A Geostatistical Approach," Water Resources Research, Vol. 15, no. 2, pp. 269-280 Endo, H.K., J.C.S. Long, C.R. Wilson, P.A. Witherspoon, 1984, "A Model for Investigating Mechanical Transport 'in Fracture- Networks," Water Resources Research , Vol. 20, no.'10, pp. 1390-1400 Faust, C.R. and JtW. Mercer, 1980, " Groundwater Modeling: ~ Numerical Models", Groundwater , Vol. 18, p. 395

0 GWTT/ JUNE 86 Fogg, G.E., 1978, "A Groundwater Modelling Study in the Tuscon Basin",

M.S. Thesis, University of Arizona, Tuscon Arizona Gelhar, W. and C.L. Axness, 1983, "Three-Dimensional Analysis of Macrodispersion in Aquifers", Water Resources Research, Vol.18, no. 1, pp.

161-180 Gordon, M., N. Tanious, J. Bradbury, L. Kovach, and R. Codell, 1986,

" Draft Generic Technical Position: Interpretation and Identification of the Extent of the Disturbed Zone in the High-Level Waste Rule (10 CFR 60)"

(Draft)

Grisak, G.E.and J.F. Pickens, 1980, " Solute Transport through Fractured Media: 1. The Effect of Matrix Diffusion, " Water Resources Research, Vol. 16, no. 4, pp. 719-730 Hoeksema, R.J. and P.K. Kitanidis, 1984, "An Application of the Geostatistical Inverse Problem in Two-dimensional Groundwater.Modeling," -

Water Resources Research, Vol. 20, no. 7, pp. 1003-1020 Long, J.C.S., J.S. Reder, C.R. Wilson, and P.D. Witherspoon, 1982, " Porous Media Equivalents for Network of Discontinuous Fractures, " Water Resource Research, Vol. 18, no. 3, pp. 645-658 Mantoglou, A. , and L. W. Gelhar,1985, "Large-scale Models of Transient Unsaturated Flow and Contaminant Transport using Stochastic Methods,"

Report no. 287, Ralph M. Parsons Laboratory, Massachusetts Institute of Technology, Cambridge, MA Matheron, G. 1971, The Theory of Regionalized Variables and Its Applications, Ecole des Mines, Fontainbleau, France Mercer, J.W., and C.R. Faust, 1980, " Groundwater Modeling: An Overview",

Groundwater, Vol. 18, p. 108 Mizell, S.A., L.W. Gelhar, and A.L. Gutjahr, 1982, " Stochastic Analysis of Spatial Variability in Two-Dimensional Steady Groundwater Flow Assuming

. Stationary and Nonstationary Heads." Water Resources Research, Vol.18, no. 4, pp. 1053-1068 Nelson, R.W., 1978, " Evaluating the Environmental Consequences. of Ground water Contamination Parts I-IV, " Water Resources Research, Vol. 19, no.3 pp. 409-450 9

/

GWTT/ JUNE 86 30 -

1 Neuman, S.P., and S. Yakowitz, 1979, "A Statistical Approach to the Inverse Problem of Aquifer Hydrology: 1- Theory",

Water Resources Research, Vol.15, no. 4, pp. 845-860.

Neuman, S.P., 1982, " Statistical Characterization of Aquifer Heterogeneities: An Overview," in Recent Trends in Hydrogeology, T.N.

'r Narasimhan, editor, Geological Societies of America Special paper 189, pp.81-102 ONWI, 1983, "A Proposed Approach to Uncertainty Analysis," 0NWI-488, Office of Nuclear Waste Isolation, Battelle Memorial Institute, Columbus, OH Silling, S., 1983, " Final Technical Position on Documentation of Computer Codes for High-level Waste Management", NUREG-0856, U.S. Nuclear .

Regulatory Commission Smith, L. dnd-R.A, Freeze, 1979, " Stochastic Analysis of Steady State -

Groundwater Flow in'a Bounded'00 main - 2: Two-Dimensional Simulations,"

Water Resources Research, Vol. 15, pp. 1543-1559 Smith, L., and F. Schwartz, 1980, " Mass Transport: 1. A Stochastic Analysis of Macroscopic Dispersion", Water Resources Research, Vol. 16, no. 2, pp. 303-313 Smith, L. and F. Schwartz, 1981, " Mass Transport: 2. Analysis of Uncertainty in Prediction", Water Resources Research, Vol. 17, no. 2, pp.

351-369 k

A 6 6 J

1 0

9

- , - ,_ _-._.__.__.___..,.-_,.,,.____m.

,,-,,._.__y ~ - . . _ . _ . . - , , , . _ - _ . , , ..,...,,.,._,,-_.,,.r. m,-_..----~._ _ , . . ,.

Controlled Area ,, ,,

Environment

, wMT W% v6N4'AX W # '*

Disturbed Zone .

- a

, / Groundwater Flow Path l

l A/: N Underground Facility

,*O. g *. . :. '.*., .'. : fCt,[,.,y :.

a:.si:,y,:,it.!a,j.iu -

(Schematic - not to scale)

Figure 1 - DistuFbed lone, Patn and Accessible' Environment I

/ / / /

Immobile Water

/

/ . Rock

' /

> l '// / , i ../

>- Mobile Water >

V\

/// V,/

i .. /

/

/ ////

Figure 2 - Three Compartnent Model of Groundwater

^

I/// /

l l

D d

0 1.0 , g g y

- 0.1 0.9 -

0.2 0.8 -

- 0.3 0.7 .

s b

b

- 0.4 d e d 0.6 ~ s o 4:

e.

a

< O e 5 o -

0.5 w 5 - u w 0.5 E

> a G w 5 -

0.4 0 3

3 04 -

5 8

0.7 0.3

- 0.8 0.2

- ---- 0.85 0.15 ----

- 0.9 0,1- l l

I I 1.0.

I I I '

I I 8 Oi #

8 8 10 4 10' 10 . .

10 2 10 3

10 10 1Q' GROUNOWAT'ER TRAVEL TIME (Years!

I l

Figure 3 - Cumulative Frequency Distribution for Groundwater Travel Tim ,

I

/

< i ___ _.

C e

5

.x e "

8 m e e N e O e o

E E l- O e

.3 i

~

e 3 -

E 3

0 0.15 - - - - - - - -

1 I

'O ,

TIME Figure 4 - Breakthrough Curves 1.0' Narrow Distribution i 31 ii

(

m' Wide Distribution o'

cA y 0.5 P.

' i 5'.

3

• I

'. 3

c' p'

I l

i 1

4 0.is --- ----- -i-----

0

- 1000 years Groundwater Travel Time Figure 5 - Variance of Groundwater Travel Time Distribution

. _ - _ _ - -.