Yucca Mountain - Federation of American Scientistsbrought in by rail, canisters would be placed on support cradles by remotely controlled cranes. (b) Multiple Containment Barriers

464 Los Alamos ScienceNumber 26 2000

Yucca MountainLooking ten thousand years into the future

by Roger C Eckhardt for David L. Bish, Gilles Y. Bussod, June T. Fabryka-Martin, Schön S. Levy, Paul W. Reimus,

Bruce A. Robinson, Wolfgang H. Runde, Inés Triay, and David T. Vaniman

Number 26 2000 Los Alamos Science 465

I t’s a dry, brown, nondescript ridge near theAmargosa Desert, just north of Death Valleyand west of the Nevada Test Site. The

landscape is a drab mix of desert grasses, cacti,shrubs—like needleleaf rabbitbrush and Coopergoldenbush—and, of course, the occasionalYucca plant. Nothing moves but scrawny black-tailed jackrabbits and desert collared lizards. No man’s land.

Yet for the past twenty years, the ridge—calledYucca Mountain but looking more like a geologicspeed bump—has been scrambled over by geolo-gists, hydrologists, volcanologists, engineers, and environmental scientists. It has been photographedfrom every angle and cored at hundreds of drill sitesto characterize its geology and hydrology. Its plant and animal life has been cataloged, its topography charted, and its underlying volcanic tuff captured inthree-dimensional computer grids.

Yucca Mountain

Yucca Mountain


Potential repositoryand engineeredbarrier system

Infiltration

Flow of groundwater

Unsaturatedrock (~600 m)

Saturatedrock

Figure 1. Potential YuccaMountain Repository Site Yucca Mountain’s remote desert

location and deep water table make it

attractive as a potential burial site for

high-level nuclear waste. The nearest

community is in Amargosa Valley,

30 km away. (Top) The water table is

among the deepest in the world,

which allows building a repository

that is both deep underground and

yet well above the water table.

(Bottom) The proposed repository

with its rows of storage tunnels is

bordered on one side by an ex-

ploratory tunnel drilled into the

mountain for research experiments.

A cross-drift tunnel has also been

drilled above the repository and

through several faults.

Subsurface layout

Exploratory tunnel

Repository

Cross drift

Storage tunnels

Tunnel’s north portal Tunnel’s south portal

N

Such intense scrutiny has resultedfrom the mountain’s selection as a potential burial site for high-level radioactive waste. Much of this wastecomes from the nation’s nuclear powerplants: spent fuel rods laden with high-ly radioactive fission products, unfis-sioned uranium, and plutonium. Thereare about a hundred commercial reac-tors in the United States, many operat-ing since the sixties and seventies, andtheir spent fuel—some 39,000 tonnes—has been accumulating in cooling poolsand dry casks with nowhere to go. By2035, this tonnage could more thandouble if all power plants completetheir full licensing cycles. Waste fromresearch reactors and the Navy’s nuclearfleet plus plutonium from dismantlednuclear weapons will add another 2500 tonnes. Because Congress hasbanned reprocessing spent fuel, all thiswaste must be safely stored—for eons.

A critical storage issue is the linger-ing radioactivity of plutonium, neptunium, and other actinides in thespent fuel. The half-lives of these ele-ments are so long that the waste mustbe stored for more than 10,000 yearswithout significant leakage to the envi-ronment. To meet Environmental Protection Agency (EPA) standards, radiation doses to the public that resultfrom leakage must remain below 20 millirems per year (mrem/yr). These are daunting requirements for ageologic repository (see Figure 1). No matter how clever we are in engineering containment barriers—designing storage canisters and tunnelsto isolate the waste—eventually, waterwill seep through the repository, corrode the canisters, dissolve waste radionuclides, and carry them off.When that happens, nature itself—thenatural geologic barriers—will have tolend a hand in containing the waste.Studying and modeling the effective-ness of nature’s barriers have been thefocus of work at Los Alamos NationalLaboratory since the early 1980s.

In our studies, we have sought answers to a number of questions:Which radionuclides are most apt to dis-

solve and be carried off? How fast willthey move through the rock? What radiation doses might citizens living in Amargosa Valley, the nearest commu-nity, receive if radionuclides reach thegroundwater that feeds their wells? Willthese doses stay below 20 mrem/yr, lessthan 10 percent of normal backgroundradiation, for the next 10,000 years or more?

Reliable answers require amassingscientific data about the site’s geochem-istry and hydrology—for example, itsgroundwater chemistry, the sorptioncharacteristics of site minerals, and po-tential groundwater flow paths—as wellas about the radionuclides themselves.These data must then be combined toassess the performance of a complexphysicochemical system over many millennia. Because no experiment cancome close to analyzing radionuclidemigration for 10,000 years or more, wehave developed computer models tosimulate the migration. These simula-tions are now being checked againstdata from extensive field tests.

Early in our work, we determinedthat neptunium was a leading “worst-case” radionuclide in terms of both thelikelihood and consequences of its mi-gration. As a result, much of our workhas gone into characterizing neptunium’sspeciation, solubility, and sorption andmodeling its transport. For the latter, wehave developed a detailed picture ofhow water percolates through the moun-tain—in particular, what fraction willflow through fractures and what fractionwill diffuse into the rock matrix. Morerecently, we identified colloidal transportas another pathway that must be incor-porated in our modeling.

So far, our simulations show that itwill take much longer than 10,000 yearsfor neptunium to escape from the repos-itory. In fact, it will take at least100,000 years before there is anychance that radiation doses from escap-ing radionuclides could reach the EPA’slimit of 20 mrem/yr. And we feel thatthese predictions are conservative. Keyuncertainties in our transport parametershave been evaluated through sensitivity

analyses, and our simulations have assumed worst-case scenarios for radionuclide release and transport.

Given the challenge of predictingcomplex geochemical processes on geo-logic time scales, however, it’s not sur-prising that scientific debate over therepository’s viability is intense. Thereare still many uncertainties associatedwith our predictions, and we are just beginning to get transport data from fieldexperiments near Yucca Mountain thatwill help us validate those predictions.We are also weighing other factors, suchas the likelihood of climate changes thatcould increase rainfall in the area and themountain’s vulnerability to volcanic activity (see the box on page 492). Whatfollows, therefore, is a summary of workin progress at Los Alamos—of how weare developing and using computer sim-ulations to look 10,000 years into the future, and far beyond.

Yucca Mountain Repository

To date, the Department of Energy(DOE) has spent $6 billion studying thefeasibility of storing spent fuel deepwithin Yucca Mountain. The project in-volves scientists and engineers from theU.S. Geological Survey, several nation-al laboratories, and a slew of privatecompanies and government agencies.The task is to design a repository, char-acterize its site geology, and predict itsreliability. By 2001, the Secretary ofEnergy must decide whether to recom-mend to the President that a repositorybe developed at Yucca Mountain. If thesite is recommended, and if the Presi-dent and Congress concur with that rec-ommendation, then DOE will apply tothe Nuclear Regulatory Commission(NRC) for a license to build the reposi-tory. As part of that application, DOEwill have to present a credible case thatthe repository will perform as predicted.

As currently envisioned, the reposi-tory is a labyrinth of tunnels some300 meters beneath the mountain’s surface (Figures 1 and 2). Waste wouldbe placed in steel canisters 2 meters in

Yucca Mountain


diameter and up to 6 meters long, withthe canisters nestled end to end alongreinforced storage tunnels. Once therepository is filled with waste (at least70,000 tonnes), remote sensors wouldmonitor the canisters, tunnels, and surrounding rock to make sure theyfunction as predicted. During the first100 years, waste could be retrievedshould problems arise. Eventually, all shafts, access ramps, and tunnelswould be sealed.

Two key reasons for studying YuccaMountain as a burial site for nuclearwaste are its dry climate and deepwater table. The first minimizes waterthat could seep through the repository,corroding waste canisters and carryingoff radionuclides. The mountain aver-ages only 15 centimeters of rain a year.Of this, about 95 percent evaporatesquickly, and most of the rest is takenup by plants and lost via transpiration.Only 1 or 2 percent actually soaks intothe ground and percolates downward.

The mountain’s low water table en-ables building a repository that is deepunderground (300 meters) yet still inthe unsaturated zone, well above thewater table (another 240 to 300 meterslower). For waste radionuclides to posea danger to the public, they will haveto reach the water table and, through it,infiltrate the wells supplying water toAmargosa Valley. Several hundred me-ters of unsaturated rock beneath therepository pose a formidable barrier tothis pathway. In addition, groundwaterin the region is trapped in a closeddesert basin and does not flow into anyrivers that reach the ocean.

Still, rock at the depth of the poten-tial repository contains some water.When nuclear waste is first stored inthe mountain, heat from its fissionproducts will dry out the rock, pushingmoisture away from the tunnels. Afterabout a thousand years, however, theshort-lived fission products will havedecayed, the temperature will drop, and

water will rewet the surrounding rock.If some of the canisters have failed bythen—if they have pinholes from man-ufacturing defects such as weakwelds—dripping water could penetratethem and start dissolving radionuclides.There are many unknowns about thebehavior of manmade materials subject-ed to heat and radiation for so long; thefailure rate could accelerate, releasingmore and more radionuclides into thesurrounding rock.

The presence of minerals that couldsop up the radionuclides is a third rea-son for studying Yucca Mountain as a storage site. The mountain compriseslayers of volcanic tuff that were deposited millions of years ago. Manyof these tuffs contain zeolites, hydrousaluminosilicate minerals that have acagelike structure.

Zeolites are commonly used to soft-en hard water. Calcium and magnesiumions in water flowing through the zeo-lites exchange with sodium ions in the

Yucca Mountain


��

@@@@@@@@@

��

ÀÀÀÀÀÀÀÀÀ

Backfill

Ceramic coating

Drip shield

Figure 2. Engineered Barriers at Yucca Mountain (a) Underground Storage Tunnels. As

currently envisioned, the repository

consists of about 160 km of under-

ground tunnels with access ramps

and ventilation shafts. The steel- and

concrete-reinforced storage tunnels

and the waste canisters they would

house constitute the main engineered

barriers for waste isolation. After being

brought in by rail, canisters would be

placed on support cradles by remotely

controlled cranes.

(b) Multiple Containment Barriers.

The original repository design

included the tunnel, floor, and sup-

port cradle. Modifications now call

for enhancing containment by

adding an outer ceramic coating to

the waste canisters, placing an

umbrella-like drip shield over them,

and backfilling the tunnels.

Inner wall lidOuter wall lid

Fuel assembly

Corrosion-resistant inner wall

Structurally strong outer wall

(c) Waste Canisters. Nuclear waste would be stored in double-walled, cylindrical

canisters 2 m in diameter and up to 6 meters long. The initial design called for

the canisters’ inner wall to be made of a corrosion-resistant, high-nickel alloy

almost 2 cm thick. For structural strength, the outer wall was to be carbon steel

almost 10 cm thick. Recent modifications reverse the two layers.

zeolite structure and remain trapped:hard water flows in, softened waterflows out. The zeolites could similarlytrap radionuclides that have dissolvedin groundwater. Some of the moun-tain’s layers of zeolitic tuff also con-tain clay, another sorbing material. The zeolites and clay could effectivelyremove strontium and cesium and, to a lesser but still significant extent, ura-nium, plutonium, neptunium, and othertransuranics that have leached intogroundwater seeping through therepository and toward the water table.

Modeling the Mountain

Work at Los Alamos in developinga radionuclide transport model forYucca Mountain began by incorporat-ing the mountain’s stratigraphy intocomputational grids. Using techniquesthat ranged from x-ray diffraction and fluorescence to microautoradiographyand potassium/argon dating, we ana-lyzed hundreds of borehole samplestaken from the mountain. The picturethat emerged from these and otheranalyses is that the mountain is com-posed of alternating layers of weldedand nonwelded tuff—volcanic ash—that are tilted, fractured, faulted, andlocally altered to zeolites and clay minerals (Figure 3).

The welded tuff consists of dense,nonporous deposits that have been hard-ened by heat and pressure and are typi-cally highly fractured. By contrast, thenonwelded tuffs are soft, porous depositsthat contain few fractures. Three signifi-cant zeolitic layers underlie the reposito-ry site: one in the unsaturated zone, onein the saturated zone, and one that slopesdownward through both zones. A thinlayer containing clay and zeolites alsostretches beneath the repository.

One important finding from our rockanalyses is that zeolites and clays poseeffective natural barriers to radionu-clide migration. Sorptive zeolites(clinoptilolite and mordenite) offer themost massive and mappable barrier.Whether their interactions are strong,

as with strontium and cesium, or weak,as with neptunium, their sheer abun-dance makes them a formidable obsta-cle to radionuclide transport. Smectiteclays are not so abundant, but theirwidespread occurrence ensures that alltransport pathways will eventually en-counter them. The clays’ strong affinityfor plutonium and moderate affinity forneptunium make them an effective bar-rier. One other mineral group, man-ganese oxides, should also function asa barrier. Although these oxides are not

so abundant or widely distributed asthe clays (and consequently not “map-pable”), they commonly coat fracturesurfaces in both the saturated and un-saturated zones. Their strong interac-tion with neptunium should impede the latter’s transport through fractures.

Our transport model incorporatesthese various stratigraphic layers, including their mineral compositions,porosities, fault locations, and fracturedensities. It also incorporates hydrolog-ical data from the U.S. Geological

Yucca Mountain


��

��

��

Sat

urat

ed z

one

Uns

atur

ated

zon

e

Welded Tiva Canyon Tuff

West

East

Welded Topopah Spring Tuff

NonweldedCalico Hills Formation

Prow Pass Tuff

Bullfrog Tuff

Tram Tuff

Layer containing clay and zeolites beneath potential repository (smectites plus calcic clinoptilolites and heulandites)

Principal zeolitic layers (abundant clinoptilolite)

Water table

Potential repository

NonweldedPaintbrush tuffs

��Figure 3. Yucca Mountain Stratigraphy This east-west cross section of Yucca Mountain shows the potential repository rela-

tive to the water table and principal rock strata. The main upper layers of volcanic

ash alternate between welded tuff (highly fractured, dense volcanic deposits) and

nonwelded tuff (highly porous deposits with few fractures). The clay-altered layer of

nonwelded Paintbrush tuffs above the repository is expected to impede the percola-

tion of water from the mountain’s surface toward the repository. The zeolitic layers

beneath the repository could filter out or slow down radionuclides that escape the

repository’s engineered containment barriers and move toward the saturated zone.

(Note: the vertical scale in this cross section is exaggerated.)

Yucca Mountain


Figure 4. FEHM Models of Yucca MountainThe top graphic superimposes the computational

grid for FEHM’s 2-D finite element model on 3-D

stratigraphic framework models, all for east-west

cross sections through Yucca Mountain. The right-

hand graphic shows slices through the 3-D steady-

state flow model used to study chlorine-36 migration;

contours correspond to the water’s age as based on

its chlorine-36 concentration. The bottom graphic is

the 3-D grid used for most of our site-scale FEHM

calculations (at 3 times vertical exaggeration). The

grid describes 21 stratigraphic zones including

the repository and the zeolitic zones of the Calico

Hills Formation. To depict the mountain on 125-m

centers, the grid uses more than 51,000 nodes and

350,000 tetrahedra.

Apparent age (years)

N

0 106

Survey and from Lawrence BerkeleyNational Laboratory. These data include rock matrix and fracture permeability, residual water content inthe tuff, and a spatial map of precipita-tion infiltration along the mountain’ssurface. Finally, the model incorporatesradionuclide transport properties derived from extensive laboratory experiments, such as radionuclide solu-bilities in Yucca Mountain groundwa-ter, sorption coefficients for the varioustuffs, and diffusion coefficients for themovement of dissolved radionuclides in the rock matrix.

The nexus for all these data isFEHM—a finite element heat- andmass-transport code. FEHM solves theequations of heat and mass transport inporous and fractured media in two orthree dimensions. The code also offers acomprehensive set of models for simu-lating the transport of dissolved speciesin either the gas or liquid phase. It

combines the capability of simulatingtransport using either finite element orparticle-tracking solutions with a dual-permeability capability that captures theeffect of fractures on flow and transport.

The code is linked to software thatproduces finite element computionalgrids that not only preserve the site’shydrostratigraphic structure but alsohonor the numerical conditions neededto perform accurate simulations. Thegrids, or meshes, are made of trianglesfor 2-D models and tetrahedra for 3-Dmodels (Figure 4). Using the physicaland chemical properties we have tied toeach node in the calculation grids,FEHM calculates, in discrete spatialand temporal steps, the transport of in-dividual radionuclides through themountain. (See the box on the follow-ing two pages for an explanation ofmesh generation.)

Figure 5 shows the geochemical fac-tors for radionuclide transport that we

have focused on in assessing the effec-tiveness of Yucca Mountain’s geologicbarriers. Percolating groundwater is theprimary carrier for the radionuclides,which are all soluble in water to someextent. After dissolving, they must travel through the repository’s backfilland heat-altered tunnel walls, throughthe unsaturated rock beneath the reposi-tory, and finally through the saturatedzone’s groundwater flow system. In addition to the hydrologic processesthat are important to radionuclide migration, transport processes such assorption onto minerals and diffusioninto dead-end rock pores will controlthe ultimate movement of radionu-clides. Findings at the Nevada Test Siteof colloid-facilitated transport for actinides have added yet anotherprocess of concern, which is discussedin the box on page 490). To be credi-ble, our transport models must accountfor all these processes.

Yucca Mountain


Am

Am

NpU

Cs

Pu

��Colloid

transport

Precipitate

Desorptionfrom colloids

Sorption to differentcolloids

NpPu

UAm

TcCs

NpNp

F R A C T U R E

R O C K M A T R I XR O C K M A T R I X

R O C K M A T R I XR O C K M A T R I X

Solubility and speciation

Sorption to fracture minerals

Different species

Sorption to tuff

Dissolvingsolid

Diffusion into solid matrix

Dissolvingwaste

Mineral coatingAm

Figure 5. Geochemical Factors for Radionuclide Transport This cartoonlike depiction of a breached waste canister shows how dissolved radionuclides could escape from the repository

through water-filled rock fractures and how the surrounding rock matrix could impede their escape. To assess the effectiveness o f

Yucca Mountain’s geologic barriers in slowing radionuclide transport, we have conducted laboratory-scale experiments to learn h ow

soluble the radionuclides, particularly neptunium-237, would be in mountain groundwater, to what extent they would be removed

from the water by sorption onto mountain tuffs and fracture minerals or by diffusion into the rock matrix, and how colloids

(nanometer-sized particles) would affect their transport through rock fractures. Data from these studies are important in defini ng the

chemical and physical parameters for our modeling.

Yucca Mountain


Mesh Generation for Yucca MountainCarl W. Gable

To model the transport of wasteradionuclides through YuccaMountain, we must generate a

grid, or mesh, on which our FEHM cal-culations can be run. Our primary toolfor generating, optimizing, and main-taining computational meshes is LaGriT(for Los Alamos Grid Toolbox), a general-purpose software package.LaGriT is a spinoff of X3D, which wasdeveloped in the 1980s by HaroldTrease.

Developed in the 1990s, LaGriT is a collaborative product of the AppliedPhysics, Theoretical, Earth and Envi-ronmental Science, and Computing, Information, and Communications Divisions at Los Alamos. It has beenused to model such varied phenomenaas shock physics, combustion, semi-conductor devices and processes, bio-mechanics, the evolution of metallic microstructure, porous flow, and seismology.1

A mesh consists of nodes (points) atspecific locations in space that are connected to form elements. These ele-ments can be triangles or quadrilateralsin 2-D models and tetrahedra, hexahe-dra, prisms, or pyramids in 3-D mod-els. The elements fit together like thepieces of a puzzle to represent physicalsystems such as the rock layers inYucca Mountain, a human knee joint,or a semiconductor chip. Physicalquantities such as pressure, tempera-ture, or density, which are continuous

in real materials, are usually represent-ed by discrete values at the nodes orwithin the elements.

Mesh generation draws on both cre-ativity and advanced mathematical algo-rithms. As Thompson et al. (1999) note,grid generation is “still something of anart, as well as a science. Mathematicsprovides the essential foundation formoving the grid generation processfrom a user-intensive craft to an auto-mated system. But there is both art andscience in the design of the mathematicsfor . . . grid generation systems, sincethere are no inherent laws (equations)of grid generation to be discovered. Thegrid generation process is not unique;rather it must be designed.”2

Mesh generation can be automatedbut never becomes automatic. Althoughsoftware like LaGriT helps automatecomplex “meshing” operations, generat-ing successful meshes still rests on aseries of judgment calls by an expert,who must weigh many tradeoffs. Forexample, imagine a calculation done on

a 10 × 10 × 10 grid of hexahedral ele-ments. If we double the resolution to a20 × 20 × 20 grid, the total number ofelements goes from 1000 to 8000. Inaddition, if the calculation involvesmodeling the change over time of aquantity such as saturation, doublingthe resolution may require cutting thetime steps in half, which doubles thecomputer time needed. Overall, then,doubling the calculation’s resolutionwill increase calculation time by a fac-tor of 16. Although high resolution bestrepresents complex geometry and pro-duces the most accurate physics solu-tion, it requires more elements, whichresult in calculations that require morecomputer memory and cycles.

The accompanying graphics illustratevarious computational grids that couldbe generated for modeling YuccaMountain. Our starting point is a geological model that represents themountain as a sequence of sloping rocklayers offset by two vertical faults (a).Focusing on the boxed area around theleft-hand fault, we first create a simplemesh of square elements (b). The ele-ments’ colors correspond to those ofthe cross-section layers they represent

1More information on LaGriT is available from the software’s Web site:http://www.t12.lanl.gov/~lagrit/.

2Handbook of Grid Generation. 1999. J. F.Thompson, B. K. Soni, and N. P. Weatherill,Eds. New York: CRC Press, p. iii.

(b) Low-Resolution Regular Mesh

(a) Geological Model of a Cross Section at Yucca Mountain

(c) High-Resolution Regular Mesh

6400 elements400 elements

and thus also to the layers’ varying material properties (such as density and porosity).

The low-resolution squares in (b) doa poor job of representing the geology.The vertical fault is lost and individualrock layers are nearly lost because themesh is so coarse. We can improve the mesh by increasing its resolution: thesquares in (c) are one-quarter the size ofthose in (b). Now the geology is betterrepresented, although the interfaces be-tween rock layers are still represented byjagged stair steps. Also, only the thickestlayers are represented by contiguous ele-ments; thin layers and small features arestill lost in the grid’s coarseness.

Another approach is to use variablemesh spacing, as shown in (d). Variablespacing allows us to “zoom in” withhigh resolution on some areas andmaintain low resolution in others. How-ever, variable spacing generally workswell only for simple geometries inwhich the phenomena being modeledtake place in a small portion of the entire computational domain and thusonly a few areas require high resolution.

A more flexible approach is to adaptmesh resolution to the geometry of inter-est, as done in the quad-tree meshesshown in (e) and (f). These meshes allow cascading refinements: each element issubdivided into four elements, each ofwhich is then subdivided into four stillsmaller elements, and so on.

In (e) and (f), we start with the meshof (b) and then refine only selected ele-ments. In (e), we refine along all materialinterfaces by a factor of 16 but leave regions far from these interfaces at lowresolution. In (f), we refine only thinrock layers by a factor of 32, increasingtheir resolution while maintaining lowerresolution in the thick layers. In bothcases, however, the number of elementsis much greater than in the previousmeshes, which will slow down our calculations.

Our mesh examples so far are allstructured: that is, they are made ofquadrilateral elements whose positionsare readily defined in terms of rows and columns and whose connectivity is

logical. Our last mesh examples aremade of unstructured triangular elements whose connectivity is morearbitrary: for example, the nodes havevarying numbers of triangles attachedto them. This unstructured approach,however, allows us to create meshesthat actually conform to the mountain’svaried material interfaces.

Both low-resolution (g) and high-res-olution (h) meshes do well in represent-ing the geologic interfaces. The high-resolution mesh, however, will do abetter job solving the physics of radionuclide transport because its small-er elements can more accurately repre-sent variations that occur over short distances. In modeling Yucca Mountain,we must also contend with phenomenathat lack symmetry, have a wide rangeof length scales, and involve very thin

layers that must be preserved as contin-uous. As a result, triangles in 2-D mod-eling and tetrahedra in 3-D modelinghave been our meshes of choice.

In addition to tradeoffs in how wellthey represent the mountain’s geometry,meshes also pose tradeoffs in their suit-ability for different physics codes.Some codes can solve problems onlyon regular grids like those shown in(b)–(d). Others can use quad-tree mesh-es like those shown in (e) and (f) butnot the unstructured meshes of (g) and(h). Thus the meshing approach mustbe compatible with the physics codethat will be used.

Developing flow and transport mod-els for Yucca Mountain has pushed thelimits of mesh generation technology.

The models’ size requires us to keepthe number of elements as low as pos-sible, their complex physics requiresus to accurately represent the geologyof the repository site, and the need fortimely results requires us to automatemesh generation whenever possible.These often conflicting demands havebeen met by a collaborative effort inenhancing mesh generation capabilitiesto meet the challenges of modelingYucca Mountain. ■

Yucca Mountain


(e) Quad-Tree Mesh

(g) Low-Resolution Triangular Mesh

(d) Variably Spaced Regular Mesh

(f) Quad-Tree Mesh

(h) High-Resolution Triangular Mesh

3844 elements

31,428 elements 11,673 elements

7880 elements 31,520 elements

Neptunium Chemistry

Of the several hundred radionuclidespresent in spent fuel, only six are long-lived, soluble, mobile, copious, andhazardous enough to contribute signifi-cantly to calculated radiation exposuresshould the nuclides reach well water inAmargosa Valley. Four of them (tech-netium-99, iodine-129, uranium-234,and neptunium-237) could be transport-ed by groundwater because of theirhigh solubility and weak adsorption tominerals. The other two (plutonium-239and plutonium-242) tend to adsorb tominerals (because of their IV oxidationstate) but could be transported on or ascolloids.

Although technetium-99 and iodine-129, both abundant in spent fuel, wouldbe the dominant radionuclides to reachthe valley in the first 10,000 years (Figure 6), radiation doses from themare not expected to exceed EPA limits.After 10,000 years, neptunium-237starts to become the radionuclide ofconcern. While neptunium concentra-tions in spent fuel are small (only 0.03

percent), they will increase over timethrough the decay of americium-241,which has a half-life of 432.7 years.Because of its radiotoxicity, long half-life (2.14 × 106 years), high solubility,and relatively low sorption on YuccaMountain tuffs, neptunium has been theradionuclide of prime concern in ourtransport calculations. Thus we haveconducted extensive laboratory tests todetermine its solubility, speciation,sorption, and transport. Understandingneptunium’s chemical behavior is fun-damental to modeling its transport.

Solubility. As in most naturalaquifers, Yucca Mountain groundwatercontains a number of dissolved speciesthat can interact with the radionuclidesand alter the latter’s solubility. Radionuclides such as neptunium, tech-netium, and plutonium are also verysensitive to changes in the water’sredox potential and pH. Such changesaffect the stability of the radionuclides’oxidation state, which is the main para-meter that controls the extent of theirgeochemical reactions.

Table I characterizes groundwaterdrawn from two saturated-zone wellsand pore water found in the unsaturatedzone at Yucca Mountain. Because car-bonate is the predominant ligand forneptunium complexation, and becausethe two saturated-zone waters roughlybracket the bicarbonate concentrationsfound at Yucca Mountain, we haveused their chemistries in our transportcalculations.

Under oxidizing conditions (i.e., inwater with a redox potential of >200millivolts), neptunium will be stable inthe V oxidation state, which generallyhas the highest solubility among ac-tinide oxidation states. However, reduc-ing conditions from oxygen depletionmay exist in saturated-zone water atYucca Mountain given the presence oflow-valent iron. These conditionswould stabilize neptunium in the IV oxidation state and reduce its solubilityby several orders of magnitude. Whileresearch is still ongoing, current datastrongly support the existence of somereducing groundwaters in the saturatedzone at Yucca Mountain.

Carbonate concentrations, which atYucca Mountain are in the millimolarrange, will have an important effect onneptunium transport. In typical ground-waters, NpO2

+ and the monocarbonatecomplex, NpO2CO3

–, are the predomi-nant neptunium solution species. Withdecreasing carbonate concentration,NpO2

+ and, to a lesser extent, the first hydrolysis product, NpO2OH(aq),will dominate the solution speciation.Changes in species composition andcharge will change neptunium’s sorp-tion, which in turn will alter its transportcharacteristics. Speciation changes maybe expected in different site-specificgroundwaters and when going from unsaturated- to saturated-zone water.

Under different physicochemicalconditions, neptunium can either accumulate and form a precipitate afteroversaturation or be transported bygroundwater as dissolved or particulatespecies. Over time, the precipitatesformed initially may transform to morethermodynamically stable and thus less

Yucca Mountain


Total99Tc129I237Np239Pu242Pu

1010

108

106

104

102

102 103

Time (yr)

Act

ivity

(C

i)

104 105 106

Figure 6. Radionuclide Decay Rates for Nuclear Waste The waste’s initial activity (first 1000 years) is primarily a result of short-lived fission

products. Having fallen about three orders of magnitude, the remaining activity is then

a result of technetium-99, iodine-129, and three long-lived actinides: neptunium-237,

plutonium-239, and plutonium-242.

soluble solid phases. Despite the impor-tance of carbonate for neptunium speci-ation in solution, however, solid Np(V)carbonates with the general formulaMNpO2CO3•nH2O, where M is any alkali, are not likely to be stable because of the low concentrations of alkali metal cations in Yucca Mountaingroundwaters. Thus, solid Np(V) oxides and/or hydroxides determineneptunium’s solubility in mountaingroundwaters. At redox potentials lowerthan about 300 millivolts, Np(IV)oxide/hydroxide is the solubility-controlling neptunium solid, with Np(V)predominating in solution.

In transport and dose calculations,our current solubility limit for neptuni-um is about 10–4 molar, with minimumand maximum values of about 10–6 and10–3 molar, respectively. By compari-son, the solubility limit for plutonium isaround 10–8 to 10–7 molar, three ordersof magnitude lower than that of neptu-nium. Plutonium’s lower solubility re-sults from the existence of Pu(IV) inthe solid state. Neptunium’s muchgreater solubility drives the search tofind reducing conditions as an addition-

al barrier to neptunium transport atYucca Mountain.

Sorption. Once groundwater dis-solves the radionuclides and begins tocarry them away from the repository,their sorption on mineral surfaces is themain geochemical mechanism for limiting their migration. As describedearlier, Yucca Mountain is composed of a thick (>1.5 kilometers) layered sequence of volcanic tuffs and lavas(see Figure 3).

To determine the mineral composi-tion of these layers, we analyzed thou-sands of core samples from the manywells that penetrate the mountain andsurrounding area. These analyses showthat the tuffs can be classified as vitric,devitrified, and zeolitic. Vitric tuff iscomposed primarily of volcanic glassfragments. Devitrified tuff, present inmore than half the layers, is composedof glass fragments that have crystallizedinto an assemblage of feldspars and sil-ica minerals, forming densely weldedstratigraphic units. Zeolitic tuff is com-posed mostly of volcanic glass that hasbeen altered to zeolites.

To characterize the transport behav-ior of radionuclides within these layers,we measured radionuclide sorption ontothe tuffs and, for comparison, ontopure, well-characterized minerals.Crushed tuff and mineral samples wereequilibrated with groundwater typical ofYucca Mountain before the radionu-clides were added. From the measuredamounts of radionuclides remaining insolution (C) and sorbed onto the sub-strate (F), we then calculated the batchsorption coefficient, Kd = F/C.

The Kd value (in milliliters pergram) is the ratio of moles of radionu-clide per gram of solid phase to molesof radionuclide per milliliter of solu-tion. A large Kd value indicates highsorption on the mineral or tuff.

Table II gives values for neptuniumand plutonium sorption onto samples ofvitric, devitrified, and zeolitic tuff andonto several minerals. As expected, thehighest sorption coefficients for neptu-nium are obtained for zeolitic tuff andclay-bearing vitric tuff. The same sorp-tion behavior would be expected forneptunium and plutonium ions in thesame oxidation state. The fact that plu-tonium has a much higher retention viasorption than neptunium most likely reflects the nuclides’ presence in differ-ent oxidation states: Np(V) and Pu(IV).The low sorption of neptunium is dueto the small charge-to-radius ratio, the large size, and the low tendency forcomplexation reactions of the neptunylion (NpO2

+). Because of steric effects,the sorption mechanism for neptuniumonto a zeolite may be a surface reactionrather than cation exchange within the zeolite’s cagelike structure.

We found that neptunium sorptionfor multiple tuff samples is actuallyrather variable. Some zeolitic, vitric,and devitrified tuff samples have almostno affinity for neptunium, whereasother samples with similar mineralcompositions show sorption coefficientsin the range of 5 to 10 milliliters pergram (mL/g). This variation suggeststhat the favorable sites for sorption areassociated with some minor mineralphases, such as iron oxides or clays,

Yucca Mountain


Table I. Chemistry of Yucca Mountain Groundwater

Concentrations of the major dissolved species in water from two saturated-zone wellsand in pore water from the unsaturated zone. In our calculations, we have used the twosaturated-zone waters to bracket the water chemistries expected at the potentialrepository.

Saturated-Zone Well Water (mg/L)Unsaturated-ZonePore Water (mg/L)

Species

Sodium 45 171 26–70Bicarbonate 143 698 20–400Calcium 12 89 27–127Potassium 5 13 5–16Magnesium 2 32 5–21Sulfate 18 129 39–174Nitrate 10 <0.1 0–40Chloride 6 37 34–106Fluoride 2 4 –Silicon 30 30 72–100

pH 6.9 6.7 6.5–7.5Eh (mV)a 340 360 400–600b

aRedox potentialbMay be lower locally

whose abundance varies in the tuff. In-deed, the largest values for neptuniumsorption in Table II are for the pureminerals hematite (an iron oxide) andcalcite (calcium carbonate) and forsmectite (a clay). Experiments withpure clinoptilolite, a common zeolite inYucca Mountain tuffs, and tuff samplescontaining significant amounts ofclinoptilolite show that neptunium sorption increases with decreasing pH,contrary to the sorption of other radionuclides onto zeolites. Given therelative abundance of clinoptilolite atYucca Mountain, we have found that areasonable approach for predicting thesorption of neptunium onto zeolitictuffs is to assume that clinoptilolite isthe only sorptive mineral present.

Because one of the main compo-nents of Yucca Mountain groundwateris bicarbonate, we also examined howneptunium sorption is influenced by thegroundwater’s carbonate content andionic strength. Although calcite showedan increase in neptunium sorption ascarbonate content and ionic strength in-creased, zeolitic tuff showed a decreasein sorption. Such behavior is indicativeof different sorption mechanisms forthe two substrates.

Sorption Mechanisms. If we assume that clinoptilolite is the onlysorbing phase for zeolites and calculatevalues of Kd divided by the solid-phasesurface area, we find that neptuniumsorption onto a series of tuff samples iscorrelated with surface area (Figure 7).This correlation again suggests thatsorption is a surface reaction rather thancation exchange throughout the volumeof the zeolite. Based on such data, wefeel that neptunium sorption onto YuccaMountain minerals is governed by surface complexation on a variety ofoxide phases and by coprecipitation andsurface adsorption involving carbonateminerals, such as calcite.

The surface-complexation mechanism appears to be relatively insensitive to variations in ionicstrength, groundwater composition, and pH values between 6.5 and 8.5.

This mechanism is likely responsiblefor the 0.5- to 5.0-mL/g range in neptu-nium sorption coefficients measured inmany different rock samples. The highend of this range may reflect secondarymechanisms, such as the reduction ofNp(V) to Np(IV) on mineral surfaces

containing ferrous iron, making the actinide more strongly sorbing. Becauseof these uncertainties concerning sorp-tion mechanisms, we have used a probability distribution for neptuniumsorption coefficients in our transportcalculations for Yucca Mountain.

Yucca Mountain


Table II. Comparison of Sorption Coefficients for Neptunium and Plutonium

Kd (mL/g)a

Neptunium Plutonium

Samples pH = 7 pH = 8.5 pH = 7

TuffsZeolitic tuff 0.30 1.5 300—500Devitrified tuff 0.007 —0.04 40—100Vitric tuff 0.2 0.3 600—2,000Vitric tuff with 10% clay —8 — —

MineralsQuartz —0.1 —0.2 <10Albite —0.08 —0.1 3—10Calcite — 50 200—1,000Hematite 70 600 >10,000Clay

Smectite 80 — —

ZeoliteClinoptilolite 2.6 1.4 600—3,000

aNegative values are due to uncertainties in the data (±0.5).

30

25

20

15

10

5

0

30

25

20

15

10

5

0

Sur

face

are

a (m

2 /g)

73

2

14

05

19

36

76

7

26

8

27

0

27

2

15

05

15

06

15

10

15

29

16

25

17

72

20

77

25

70

74

7

13

94

14

05

14

07

15

55

23

25

Tuff samples

Surface area

Sorptioncoefficients

Sor

ptio

n co

effic

ient

, Kd

(mL/

g)

Figure 7. Correlation of Neptunium Sorption and Tuff Surface Area Comparison of the coefficients for neptunium sorption onto various tuff samples and

the samples’ surface area shows the two values to be correlated. The sorption coeffi-

cients are for neptunium solutions in typical Yucca Mountain groundwater under

atmospheric conditions and at initial neptunium concentrations of about 10 –7 M.

Dynamic Measurements.In con-trast to the batch sorption equilibriummeasurements discussed above, the mi-gration of radionuclides through tuff isa dynamic, nonequilibrium process.Thus, we needed to confirm our equi-librium measurements by carrying outdynamic transport experiments. We didthis by eluting radionuclide solutionsthrough columns of washed, crushed,and saturated Yucca Mountain tuff,using the same tuff as in our batchsorption experiments. (Migrationthrough a solid rock column under un-saturated conditions would require moretime than was feasible for our work.)

The arrival time for neptunium at thebottom of the column was consistentwith the sorption coefficients measured in our batch sorption experi-ments. Consequently, we can predictthe retardation of neptunium transportsimply by using the equilibrium sorp-tion coefficient. However, while theelution time was accurately predicted,the amount of eluted radionuclide dif-fered from modeling predictions. Theradionuclide did not travel uniformlythrough the column but arrived dis-persed, probably because of variablepath lengths through the rock.

A key implication of these dynamicmeasurements is that the use of batchsorption values in site calculations willyield conservative predictions. In otherwords, the predicted radionuclidetransport times to the repository’s siteboundary (defined as 5 kilometers dis-tant) will be shorter than actual traveltimes. Thus, if the simulations showthat the natural rock barriers effective-ly impede neptunium transport, thenuncertainties associated with neptuni-um sorption mechanisms will only im-prove the barriers’ effectiveness overour predictions for them.

Matrix vs Fracture Flow

Another key issue for radionuclidetransport is groundwater flow—howquickly will water move through themountain and what pathways will it fol-

low? Early theories ranged from a “tinroof” scenario in which the upper thinlayer of nonwelded tuffs acts as a rela-tively impervious barrier, divertingwater laterally and drastically reducingpercolation through the repository, to ascenario of rapid flow along majorfaults and fractures that allows water toinfiltrate the repository within only afew decades. Our studies have shownthe actual flow mechanisms to be a bitmore complex than either of these twoextremes.

To assess the movement of ground-water through Yucca Mountain, webegan by looking at how water entersits upper surface. Scientists from theU.S. Geological Survey have measuredinfiltration rates, conducting neutron as-says of soil moisture to a depth of75 meters. By balancing such parame-ters as precipitation, runoff, evapora-tion, soil thickness, plant transpiration,and hydraulic rock properties, they de-veloped a spatial map of infiltrationacross the surface of the mountain. Ac-cording to this map, infiltration is high-est along the crest, where rain cloudsaccumulate and soil cover is almostnonexistent, and much lower wherethick alluvial deposits and relativelyabundant plant life lead to high rates ofevapotranspiration.

Next, we augmented the survey databy examining the water content in rocksamples taken from deeper within themountain. As wells were bored, we ana-lyzed pore water trapped in core sam-ples at various depths to determine itsage and infiltration rate. Also, when anexploratory tunnel was drilled into themountain, we collected hundreds ofrock samples at regular intervals alongits length. This 8-kilometer-long, U-shaped tunnel descends to the level ofthe potential repository and parallels therepository’s eastern boundary along itsown curved base (see Figure 1). Thetunnel intersects major fractures in themountain that are due to faulting, tilting,and block rotation in the distant past.

We analyzed the pore water in theserock samples for its chlorine concentra-tion. Because chloride salts are highly

soluble, they dissolve in rainwater andmove into the mountain with the infil-trating water. Once water has traveledthrough the upper soil zone, where itcan evaporate or be taken up by plants,its chloride concentration remains rela-tively constant. Comparing the chlorideconcentration in pore water below thiszone with that in rainwater yields ameasure of how much water has beenlost to evaporation and, thus, allows usto determine net infiltration rates for themountain.

In addition, chlorine has a radioac-tive isotope, chlorine-36, that is pro-duced naturally in the atmosphere by the interaction of cosmic rays with argon. Because the amount ofchlorine-36 so generated has changedover the ages and because the isotope’shalf-life is 300,000 years, the ratio ofchlorine-36 to stable chlorine (36Cl/Cl)in groundwater serves as a measure ofthe water’s age.

Looking back tens of thousands ofyears, we find that the 36Cl/Cl ratio hasvaried because of variations in chloridedeposition rates (perhaps due to climatechanges) and variations in chlorine-36production (due to changes in theearth’s geomagnetic shielding). Abroad-brush picture shows a bimodaldistribution: a fairly constant 36Cl/Clratio of about 5 × 10–13 over the last10,000 years (the Holocene signal) andan elevated ratio of about 10 × 10–13

before that period (the Pleistocene signal). We confirmed this bimodal distribution by analyzing fossilizedpack-rat middens (replete with crystallized urine) for chlorine-36 andplotting the measurements against theirradiocarbon ages.

During the 1950s and early 1960s, a chlorine-36 spike occurred when atmospheric nuclear tests in the SouthPacific significantly increased the iso-tope’s production rate. The resulting36Cl/Cl ratio rose above 15 × 10–13.This “bomb pulse” can be used to testfor recent groundwater, such as wouldbe expected if water infiltrated YuccaMountain by way of fast transport pathsthrough fractures and faults.

Yucca Mountain


When the pore-water data from thetunnel rock samples were examined sta-tistically, we found four distinct sets of36Cl/Cl ratios. The two main ratioswere the expected Holocene and Pleis-tocene signals. A small set of lower ratios in which chlorine-36 had notice-ably decayed may represent water thatis 300,000 years or more old. Finally, a small set of elevated 36Cl/Cl ratios (12 × 10–13 and greater) represents veryyoung bomb-pulse water.

We found a strong correlation be-tween the distribution of bomb-pulsewater and major faults (Figure 8). Forexample, 58 percent of the bomb-pulsesamples were within 10 meters of afault, and 75 percent were within30 meters. We also found a strong cor-relation between the age of the waterand the thickness of the nonwelded layerof Paintbrush tuffs that lies between the

mountain top and both the potentialrepository and the exploratory tunnel.Pore water at the tunnel level wasyounger Holocene water where the layeris thinner and older Pleistocene waterwhere it is thicker. Thus the nonweldedtuffs, which resemble spongy, hard-packed sand, apparently impede thedownward flow of water. Only where amajor fault intersected this tuff layer did any young bomb-pulse water reachthe tunnel.

A simulation that assumes a constantwater infiltration rate of 0.5 millimeterper year (mm/yr) shows general agree-ment with the chlorine-36 data (Figure9). Bomb-pulse water moves rapidlythrough the upper layer of welded andhighly fractured Tiva Canyon Tuff, butthe nonwelded Paintbrush tuffs keepmost water from percolating down intothe Topopah Spring Tuff, which in-

cludes the tunnel and the repositorysite. In the northern half of the tunnel,most water in the Topopah Spring Tuffis at least 10,000 years old. In thesouthern half of the tunnel, where thePaintbrush tuffs are thinner, Holocenewater as young as 1000 years old isfound. Exceptions to this age patternoccur at major faults, where a smallfraction (1 or 2 percent) of the water infault zones at the repository level isbomb-pulse water that moved quicklyto that depth.

Despite the general agreement be-tween the simulation and chlorine data,however, there were some discrepan-cies in the details. In particular, the predicted bomb-pulse signal at thesouthern end of the tunnel did not appear in the pore-water samples. Nordoes the simulation agree with the U.S.Geological Survey data, which indicatea spatially varying water infiltrationrate. In fact, when we use the survey’sinfiltration rate in our simulations, predictions and experimental data areeven more at odds. There is no flow ofbomb-pulse water along fault fractures,and the distribution of the water’s ageis relatively uniform throughout thetunnel rather than changing as thethickness of the Paintbrush tuff layerchanges.

However, our measurements of totalchloride concentrations in pore wateralso differ from the survey results—forinstance, infiltration rates toward thesouthern end of the tunnel appear to belower than those deduced from the surface measurements. Lower infiltra-tion rates would account for the factthat no bomb-pulse signal has been detected in faults in the southern halfof the exploratory tunnel.

Thus the mere presence of majorfaults may not, by itself, result in fastwater flow. Solutes can also travel bymolecular diffusion into the surround-ing rock matrix, where flow rates areapt to be orders of magnitude lower. Awide body of evidence, both theoreticaland experimental, suggests that matrixdiffusion is an important transportmechanism in fractured media. If the

Yucca Mountain


1 × 10–13

10 × 10–13

100 × 10–13

0 1 2 3 4 5 6 7 8

Distance along the exploratory tunnel (km) North end South end

Mea

sure

d 36

Cl/C

l rat

io

Bomb pulse

Approximate bomb pulse

Background

Fault

Figure 8. Distribution of 36Cl/Cl Ratios in Pore Water from ExploratoryTunnel Rock SamplesMost bomb-pulse water is associated with faults (vertical lines). At the southern end

of the tunnel, however, no bomb-pulse water was found even though faults are

present. At both ends of the tunnel, but especially at the south end, the 36Cl/Cl ratio

drops off, indicating that these regions contain predominately Holocene water

(<10,000 years old). By contrast, the north-central part of the tunnel has predominately

Pleistocene water (>10,000 years old). These variations likely result from the changing

thickness of the Paintbrush tuffs above the tunnel: where they are thicker, the travel

time for water is longer.

infiltration rate is not high enough tosupport fracture flow, water will diffusefrom fractures into the rock matrix, anddownward flow may revert to slowerpercolation through the matrix.

It turns out that establishing the trueinfiltration rate, which ultimately gov-erns the water flux through the unsatu-rated zone at Yucca Mountain, is muchmore important than showing the exis-tence of fast-flow pathways. A neptuni-um atom cannot distinguish betweenwater that is 50 years old and waterthat is 10,000 or 300,000 years old. Itstransport toward the saturated zone willdepend on the average net flux of water

through the repository. As the flux ofinfiltrating water increases, rock satura-tion increases. As rock saturation increases, the capillary suction of the rock matrix decreases, reducing thematrix’s ability to pull water out offractures. Thus the potential for waterflow in fractures increases.

Through a process that weighs thevarious surface, chloride, and chlorine-36 data and that reexamines the hydro-logic properties of the various strata(i.e., their porosity, matrix permeability,and fracture density), we are movingcloser to establishing a valid flux ratefor assessing repository performance.

At this point, we have found that usinga spatially varying rate that averages 4 mm/yr yields reliable but conservativecalculations. Ongoing field experimentsat Busted Butte, discussed later, arehelping us benchmark this parameter.

Laboratory transport tests indicatethat fracture coatings also affect flowrates. These tests involved columns ofYucca Mountain tuff containing bothnatural and induced fractures. The natural fractures were coated with min-erals that had been deposited over theeons; the induced fractures had no min-eral coatings. By using a variety oftracers, we were able to sort out the

North end of tunnel

South end of tunnel

Tiva Canyon Tuff, welded and fractured

Level of exploratory tunnel

Younger Holocenewater (<10,000 yr)

Older Pleistocenewater (>10,000 yr)

Bomb-pulsewater in fault

Bomb-pulse water in upper units

Topopah Spring Tuff,welded and fractured

4 x 10–13 12 x 10–1336Cl/CI ratio

Paintbrush tuffs, nonwelded and unfractured

Yucca Mountain


Figure 9. Simulation of 36Cl/Cl Ratios in Yucca Mountain GroundwaterThis simulation, which contains only one fault, generally agrees with analyses of the pore water in rock samples from the explo rato-

ry tunnel. Water moves rapidly through the upper Tiva Canyon layer of welded and highly fractured tuff. The next layer of nonwe ld-

ed, unfractured, and porous Paintbrush tuffs, however, significantly impedes water flow. Thus, most of the young bomb-pulse water

(red) remains in the Tiva Canyon Tuff. In the northern end, by the time water seeps through to the Topopah Spring Tuff—another

welded, highly fractured layer—it is at least 10,000 years old (orange and yellow). Where the Paintbrush layer is thinner, the water is

predominantly Holocene (blue and green). The exception occurs at a major fault, where a small amount of bomb-pulse water reache s

the lower rock strata (red region on the right). Other simulations also predict bomb-pulse water at fault zones in the northern end of

the tunnel.

relative effects of flow through frac-tures, of matrix diffusion into rock micropores, and of sorption by fractureminerals. We found that because ofsorption, neptunium arrived at the bot-tom of a column long after a nonsorb-ing tracer. Thus, minerals that appear tocontribute insignificantly to sorptionwhen they are present in trace quanti-ties in the bulk rock may quite effec-tively retard transport when they areconcentrated on fracture surfaces. Sen-sitivity studies are under way to assessthe impact such sorptive minerals couldhave on radionuclide transport.

Unsaturated-Zone Transport

Our modeling studies of how effec-tively the natural barriers at YuccaMountain will contain migrating wasteradionuclides began with modelingtravel times for neptunium through rockbarriers in the unsaturated zone. Tohelp us benchmark our modeling para-meters, we are now conducting large-scale field experiments to measure theactual in situ transport properties offractured rock. The experiments areunder way at Busted Butte, about 8 kilometers southeast of the potentialYucca Mountain repository.

Unsaturated-Zone Simulations. Wesimulated neptunium transport from therepository to the water table with a site-scale model that accounts for specia-tion, sorption, diffusion, radioactivedecay, repository heat, and both frac-ture and matrix flow. Figure 10 showsthe results of a 2-D simulation for ourbase case, which used a spatially vari-able infiltration rate averaging 4 mm/yrand chemical data typical of YuccaMountain groundwater: a pH of 8 with150 milligrams per liter (mg/L) of bi-carbonate, 125 mg/L of sodium, and25 mg/L of calcium. The base case alsoassumed a repository design in whichthe density of waste canisters wouldlead to thermal loads high enough toboil water in the adjacent tunnel rock.

The first panel in Figure 10 showsthe concentration of aqueous neptunium2000 years after waste emplacement.By this time we assume that the wastecanisters have begun to cool, that waterhas rewet the edges of the repository,and that some of the canisters havebeen breached by dripping water. Nep-tunium is dissolving and moving rapid-ly through the welded tuff below therepository via fracture flow. However,its migration is slowed significantlywhen it reaches the layer of CalicoHills nonwelded tuff (defined by thelower blue boundary), where matrixflow dominates.

The second panel, at 10,000 years,shows the transport plume after allrepository rock has been rewet and allwaste canisters are releasing neptunium.The third panel, at 50,000 years, showsthe main neptunium plume reaching thewater table. The aqueous neptuniumconcentrations are considerably reducedfrom those in the previous panel, main-ly because of dilution with the percolat-ing groundwater and sorption within thezeolitic tuffs in the Calico Hills layer.The final panel, also at 50,000 years,shows the concentration profile of im-mobile neptunium attached to zeolites.Comparison of the bottom two panelsdemonstrates that even though neptuni-um sorption onto zeolites is relativelysmall, it will still retard transport.

We have also carried out other base-case simulations to study the effect ofvarying assumptions for repository heat,water infiltration, and water chemistry.These studies indicate that the heatpulse from the decay of fission productsin the waste does not significantly affectneptunium migration because the timescale of heat-pulse propagation is short-er than the time scales associated withneptunium release and migration. Themajor uncertainty in this conclusion isthe possibility of rock/water interac-tions that permanently alter the rock’sporosity, permeability, or mineralogy.

In regard to water chemistry, ourstudies show that the groundwater’s pHand its calcium and sodium concentra-tions significantly affect how much the

zeolitic and clay-bearing tuffs in theunsaturated zone retard neptuniumtransport. For example, a rise in pH to9 decreases neptunium solubility, whichresults in slower releases from therepository. However, it also decreasesthe sorption of neptunium on zeolites,which leads to more rapid migration.The decrease in sorption would domi-nate, resulting in shorter travel times tothe water table and higher peak concen-trations for the radionuclide.

Our studies also show that small-scale variations in the site’s chemicaland hydrologic properties can affectneptunium transport and that such vari-ations are strongly dependent on miner-al distributions. In particular, zeolitedistributions influence the flow patternsof percolating water and the sorption ofmany radionuclides. Where zeoliticabundance is low (less than 10 percent)and, thus, where neptunium sorptioncoefficients are low (Kd < 1 mL/g), thepermeability is large enough for flow tobe matrix dominated, causing signifi-cant retardation by diffusion despite thelow Kd values. When our simulationsaccount for such correlations, we findthat neptunium transport is significantlyretarded. Its transport is slowed evenmore when we account for the presenceof smectite clays in the rock matrix.

Overall, our studies indicate that ze-olitic sorption of neptunium in the un-saturated zone results in travel timesthat are much longer than 10,000 years,and thus the repository’s primary regu-latory goal can be met for this impor-tant limiting actinide. However, ourmodeling work is based on measure-ments from small-scale laboratory experiments that do not characterize theeffects of larger geologic features suchas faults and stratigraphic boundaries.To characterize the effects of these het-erogeneities in the unsaturated zone, weare conducting large-scale field experi-ments at Busted Butte.

Busted Butte Field Studies. Wechose Busted Butte because two majortuff layers that underlie the potentialrepository—Topopah Spring Tuff and

Yucca Mountain


the Calico Hills Formation—are nearthe butte’s surface and are readily ac-cessible. These layers are actually distalextensions of the formations beneaththe repository. The Busted Butte testsite is 70 meters underground and is divided into several test blocks for a series of experiments (Figure 11). Theidea is to inject aqueous tracers into therock through horizontal boreholes, mea-

sure tracer migration, and then comparethose measurements with our modelingpredictions.

To mimic the behavior of neptuni-um and other radionuclides, we areinjecting a mixture of both nonsorbingand reactive tracers and synthetic mi-crosphere. Nonsorbing, or conserva-tive, tracers (such as bromide) movewith the aqueous phase and thus track

Yucca Mountain


(a)

(b)

(c)

(d)

0 Max

2000 years

10,000 years

50,000 years

50,000 years

Aqueousneptunium

concentrations

Immobileneptunium

concentrations

Figure 10. Simulated NeptuniumTransport at Yucca Mountain Our base-case analysis of neptunium

transport at Yucca Mountain predicts that

it will take the radionuclide 50,000 years

to penetrate the mountain’s rock barriers

and reach the water table. The top three

panels show the concentrations of aque-

ous neptunium as (a) the edges of the

repository are rewet by water and some

waste canisters begin to leak, (b) after all

rock in the repository has been rewet and

canisters throughout the repository are

leaking, and (c) when the aqueous neptu-

nium plume reaches the water table.

Panel (d) shows the concentration of

immobile neptunium sorbed onto zeolites

at the same time that aqueous neptunium

reaches the water table. Comparison of

(c) and (d) demonstrates the effective

role zeolites would play in retarding nep-

tunium transport through sorption.

the movement of water and help deter-mine the rock’s hydrologic response.They capture processes involved inthe migration of nonsorbing radionu-clides such as technetium. Some ofthese tracers are ultraviolet fluorescentdyes (fluorescein and pyridone) thatcan be detected at a concentration ofabout 10 parts per million. Reactive

tracers (such as lithium, manganese,nickel, cobalt, samarium, cerium, andfluorescent Rhodamine WT) diffuseinto the rock matrix or sorb onto minerals and are used as analogs forsorbing actinides, such as neptunium.Finally, fluorescent polystyrene microspheres mimic the transport ofcolloids.

Some of the Busted Butte tests willvalidate our flow and transport modelsand provide new transport data for thehydrologic Calico Hills Formation.Other tests are located in the relativelylow-permeability fractured rock at thebase of the Topopah Spring Tuff to pro-vide data on fracture/matrix interactions.Phase I tests, completed this past winter,involved simple, continuous, single-point tracer injection along horizontalboreholes 2 meters long. Although designed principally to test instrumenta-tion, these tests are also providing valu-able transport data.

Phase II tests, currently under way,are more sophisticated and use multi-point injection systems involving eight8.5-meter-long boreholes, each with 9injection points (Figure 12). The largePhase II test block (7 meters high, 10 meters wide, and 10 meters deep)allows us to study how large-scale het-erogeneities will affect tracer migration.The tracers are dissolved in water thathas the same composition as in situpore water. Injection rates vary from 1 to 50 mL/h, which correspond to infiltration rates of 30 to 1500 mm/yr.The lowest injection rates correspond to the high end of infiltration rates forYucca Mountain; the highest injectionrates are designed to obtain data ongreater travel distances within the experiments’ 2-year time frame.

The Phase II tests are using threegeophysical techniques (based on neu-tron assays, ground-penetrating radar,and electrical resistance measurements)to generate 2- and 3-D images of test-block saturation before and during theexperiments.

Our first results came from the PhaseIb test, which was conducted at thebase of the Topopah Spring Tuff. Thetest used a pair of injection/collectionboreholes located near a vertical frac-ture and an injection rate of 10 mL/h(see Figure 11). The nonsorbing tracersbegan arriving at the collection bore-holes, which are 28 centimeters beneaththe injection points, about a month afterinjection. For the injection rate used,pure fracture flow would have resulted

Yucca Mountain


Figure 11. Busted Butte Field Test SiteTwo of the repository’s key unsaturated-zone tuff layers, the Topopah Spring Tuff and

Calico Hills Formation, are only 70 m beneath the surface at Busted Butte. We are

using these formations to measure tracer-solution transport times through large rock

volumes. The site is divided into several test blocks, but the same general experimen-

tal procedure is being followed at each one: mixtures of tracers are injected through

one set of horizontal boreholes and then detected either at parallel collection bore-

holes or by auger sampling and mineback. Comparing measured travel times for the

tracers with our modeling predictions is helping us validate our transport models.

Main TunnelTo Portal

��

�� Portal

70 m

Calico Hills Formation��

TopopahSpring

Tuff

Phase II TestBlock

Phase Ia

Phase Ib Fault

��

Vitrophyre

Legend ��Injection Boreholes ��Collection Boreholes ��Tomography Boreholes ��

Legend ��Injection Boreholes

Collection Boreholes

Tomography Boreholes

TTestest A

lcove �Alcove �

�

in travel times of hours to days. Al-though the concentration of tracers atthe collection boreholes increased overthe next couple of months, the tracerswere detected at several collection padson both sides of the fracture rather thanjust at the intercepting fracture plane.

These results indicate strong frac-ture/matrix interactions in the tuff,causing the tracers to diffuse from thefracture into the matrix and then moveprimarily by matrix diffusion. None ofthe reactive or colloidal tracers hadreached the collection boreholes evensix months after injection. Sample corestaken from the test block are currentlybeing analyzed to determine how farthese tracers traveled.

Our Phase Ia test was conducted inCalico Hills Formation tuff. In this test,we injected tracers from single pointsin each of four horizontal boreholes attwo rates: 1 and 10 mL/h. After eightmonths of continuous injection, webegan exposing vertical slices of thetest block by mining into the rock facein stages to measure tracer migration(Figure 13). This was a “blind” test inthat we used transport calculations topredict tracer movement before themineback occurred. We are now com-paring our predictions with results forthe nonsorbing fluorescein tracer andare beginning to analyze samples forthe reactive and colloidal tracers.

So far, it appears that strong capil-lary forces exist in the Calico Hills tuff

Yucca Mountain


Injection ports(~1-m spacing)

Matrix flow

Multipoint injectionborehole

Collectionborehole

Collection pads(~1-m spacing)

Fracture flow

~10 m

Figure 12. Multipoint Injection andCollection System for Phase IITests Multipoint injection and collection

systems in the Phase II tests at Busted

Butte are providing data on how large-

scale heterogeneities, such as major

fractures, affect tracer migration. The

gray areas in this figure show hypothe-

tical tracer movement, including rapid

transport along a vertical fracture that

intersects the injection and collection

systems.

Figure 13. Mineback of Phase Ia Test in Calico Hills Tuff In the Phase Ia test, a fluorescing tracer was injected into Calico Hills tuff from a sin-

gle point in each of four horizontal boreholes (see test setup in Figure 11). After eight

months of continuous injection, we exposed vertical slices of the test block by mining

back the tuff in stages. In this photograph, the vertical plane passes through the in-

jection points of the four boreholes. Ultraviolet light reveals the presence of the fluo-

rescing tracer (yellow-green areas) around those points. The two larger fluorescent

areas are a result of a 10-mL/h injection rate; the smaller areas, a 1-mL/h rate. Tracer

movement appears to be mainly by diffusion through capillary flow. The upper-left

borehole shows tracer diffusion in two perpendicular planes due to the presence of

the left wall. The brighter yellow-green strip toward the bottom of the central

fluorescent area reveals tracer accumulation at a boundary between two Calico Hills

sublayers separated by 3 to 5 centimeters of silicified ash. Another boundary, this

one rich in clay, passes horizontally above the upper-left borehole area, impeding

upward capillary flow.

that will modulate fracture flow fromoverlying rock layers, thereby dampingpulses of infiltrating water and provid-ing extensive contact between radionu-clides and the rock matrix. Even wheninjection occurs next to a fracture,

water is imbibed quickly into the sur-rounding matrix, and fracture flow isinsignificant compared with matrix dif-fusion (Figure 14). Such results bodewell for the performance of the reposi-tory’s natural barriers: migration of

water from the fractures into the rockmatrix will lead to increased contactwith sorbing minerals such as zeolitesand clays. The combination of matrixdiffusion and sorption will lengthen radionuclide travel times.

Where no heterogeneities (such asstratigraphic boundaries) are present, ournonsorbing-tracer predictions agree verywell with measurements, capturing theasymmetric migration that results frominjection through a single point on oneside of the borehole. Likewise, our modeling predicts the predominance ofcapillary flow over fracture flow in theCalico Hills vitric tuffs (see Figure 14).However, nothing was included in themodel to account for boundaries be-tween tuff sublayers; these boundariesappear to locally retard tracer migration,whether the boundary lies above orbelow the borehole.

The Busted Butte experiments havealso uncovered two uncertainties inour modeling. In particular, we areconcerned about the adequacy of con-tinuum models to describe transport innonwelded tuff layers with high matrixpermeabilities. Our dual-permeabilitymodels appear to be more accurate inrepresenting the flow and transportprocesses of an unsaturated, fracturedrock mass. Another problem is how toaccurately characterize the transitionfrom fracture flow to matrix flow atboundaries between tuff layers thathave different hydrogeologic proper-ties. When completed, the Phase IItests should help us resolve such uncertainties and bring our modelingeven closer to realistically describingtransport in the unsaturated zone atYucca Mountain.

Saturated-Zone Transport

Although our unsaturated-zonemodeling indicates that neptuniumtravel times to the water table will exceed 10,000 years, radionuclideswill eventually reach it, releasing a radioactive plume into the saturatedzone. The second part of our modeling

Yucca Mountain


Figure 14. Capillary Flow in the Phase Ia Test vs Modeling Prediction The bottom photograph, which focuses on the central fluorescent area in Figure 14,

shows no appreciable tracer flow along a major fracture that passes diagonally just

under the injection point. The fact that the tracer has not migrated significantly along

this fracture indicates that the latter is not a fast-transport pathway but instead behaves

as a large pore within a porous matrix. The photo also shows tracer accumulation

(brighter yellow-green strip) at a layer of ash, which impedes tracer migration. Our mod-

eling prediction for tracer movement (top graphic) agrees well with these test results.

Its asymmetrical shape for tracer migration is accurate; what was missing in our model

was any provision for the stratigraphic boundary formed by the thin layer of ash. We

are now modifying our codes to include the impact of such heterogeneities.

Fluid Saturation

Major fracture

Stratigraphic boundary with ash

Injection borehole

work, therefore, entailed simulatingneptunium transport through the satu-rated zone and again benchmarkingour modeling parameters with fieldexperiments.

As in the unsaturated zone, ground-water will encounter both fracturedand porous media in the saturated zone (Figure 15). In fractured media,groundwater may move relativelyquickly through the fractures, butsome water will move into progres-sively smaller cracks and pores, whereadvection, matrix diffusion, and sorp-tion will help retard radionuclidetransport. In porous media, such as alluvium, groundwater travel timeswill lengthen because the water mustdiffuse through the matrix rather thanflow along fractures. Large-scale dispersion, or dilution, will also lower radionuclide concentrations insuch media.

Saturated-Zone Simulations. Trans-port simulations for the saturated zonewere carried out to examine how long itwould take for neptunium to reach therepository’s site boundary, defined as 5kilometers distant, and then travel an-other 15 kilometers to reach a pointwhere it could be taken up in well waterfor Amargosa Valley. (This 20-kilome-ter point assumes that 10,000 yearsfrom now, the nearest valley populationwill be 10 kilometers closer to themountain than it is today.) Our saturat-ed-zone simulations examined how ad-vective transport, diffusion into the rockmatrix, dilution, and sorption would affect neptunium transport times.

The simulations indicate that the saturated zone could lower peak con-centrations of radionuclides that bypassthe unsaturated zone’s natural barriers.Simply put, whenever a spike in elevat-ed radionuclide flux reaches the watertable, if the spike’s duration is aboutthe same as or less than the transporttime of water through the saturatedzone to the site boundary, then water inthe aquifer will “dilute” the spike, bothlowering its peak and stretching it outover a longer period of time. If, howev-

er, the radionuclide flux arrives at thewater table over a long period of time,dilution will not occur—the flux willact as a constant source of radionu-clides to the saturated zone. Thus, thesaturated zone will provide a hedgeagainst any failures of unsaturated-zonebarriers that result in sharp influxes ofradionuclides to the water table.

Initially, transport through the satu-rated zone will be in porous, fracturedtuff. Our model thus includes both frac-ture flow and matrix diffusion. Inclu-sion of the latter slows neptunium travel times to the hypothetical siteboundary from about one year for purefracture flow to thousands of years. Animportant aspect of matrix diffusion isthat it allows the radionuclides to comein contact with minerals in the sur-rounding rock that may adsorb them.As shown earlier in Figure 10, evensmall amounts of sorption have a largeimpact on radionuclide travel times andpeak concentrations.

Figure 16 presents simulation resultsfor a much more pessimistic waste-leakage scenario than that of our basecase. In this simulation, we assume that the first waste canister fails at 1000 years and that neptunium leakscontinuously over the next 30,000 yearsand then stops. The upper graph plotsthe arrival of neptunium at the watertable; the lower graph plots neptuniumtransport through the saturated zone tothe repository’s site boundary for twocases: no sorption and weak sorption(an average Kd of 2 mL/g). Note thatalthough sorption does not reduce themaximum concentration that arrives atthe boundary, even weak sorption delays the initial arrival of neptuniumby about 10,000 years.

We also examined the transport ofradionuclides to a point 20 kilometersaway from the repository. In these cal-culations, we had to account for flowthat would include movement throughhighly porous alluvium (see Figure 15).

Yucca Mountain


��yy

��

yyyyyyyyy ��yyyyyyyyyyyyyy

��

��

yyyyyyyyyyyyyyyyyy�� yyy

Amargosa Valley

Porous mediumFractured medium

Sorption

Advection and matrix diffusion

10-m scale

Repository

Carbonate aquifer

Volcanic aquifer

100-m scale1-m scale

Micron scale

Yucca MountainWater tableNorth

South

Advection

Dispersion

Matrix diffusion

Radionuclides

��

Mineral grain

1000-m scale

Large-scale dispersion

��

yyyyyy��yyyy��

��yyyy

Adsorbedradionuclides

Water inmatrix or

fracture space

Figure 15. Conceptual Model of Radionuclide Transport in the Saturated ZoneA variety of factors will affect how quickly radionuclides travel through the saturated

zone. As shown on the left, advection, matrix diffusion, and sorption will help retard

radionuclide transport in fractured media. As shown on the right, porous media

will slow advecive transport, and dispersion, or dilution, will lower the concentration

of the radionuclides. Our saturated-zone simulations examine the interplay of all

these factors.

As a result, neptunium travel times areconsiderably longer, with sorption inthe alluvium adding from 10,000 to50,000 years to them. As in our site-boundary calculations, sorption decreases the neptunium concentrationspredicted to reach Amargosa Valley.Our calculations predict that after100,000 years, neptunium concentra-tions in the valley would be no greaterthan 10 parts per trillion, which wouldpose a negligible health hazard. Thus,retardation from matrix diffusion andsorption and from the dilution effectnoted above make the saturated zone an important component in the “defensein depth” provided by the mountain’s natural barriers.

While encouraging, however, thesecalculations are not the final word. Moreexperimental and modeling work must bedone on the possibility of colloidal trans-port. Such transport would occur if ra-dionuclides sorb not to stationary tuff butto particles so minute (1 nanometer to 1 micrometer in size) that they remainsuspended in the groundwater and movewith it. Colloidal transport became a con-cern when it was discovered that in just afew decades, a small amount of plutoni-um from an underground nuclear test hadmigrated more than a kilometer throughthe saturated zone, apparently by sorbingonto colloids. This transport process isdiscussed in the box on page 490.

Saturated-Zone Field Tests.Tobenchmark our modeling of the saturat-ed zone, we conducted a series of fieldtests with tracers at a complex calledthe C-Wells. These wells are locatedabout 2 kilometers southeast of the po-tential repository site and are drilledinto fractured volcanic tuff. In terms of groundwater flow, they are directlydownstream from the southern end ofthe repository.

In the C-Wells tests, we injected avariety of tracers into the saturated zoneat one well and simultaneously pumpedwater out of another well about 30 me-ters away, establishing a recirculationloop between them. By adjusting pack-ers in the injection well that sealed off

selected intervals along its length, wewere able to test transport through dis-tinct stratigraphic layers having differ-ent hydraulic conductivities.

The tracers used were lithium bro-mide (composed of a small cation andsmall anion), pentafluorobenzate(PFBA, a large anion), and polystyrenemicrospheres (simulated colloids with anegative surface charge). The micro-spheres were tagged with a fluorescentdye so that they could be detected withflow cytometry. The bromide andPFBA are nonsorbing solutes with dif-ferent diffusion coefficients, and lithiumis a weakly sorbing solute. We conduct-ed separate laboratory tests to charac-terize the sorption of lithium to C-Wellstuffs and the matrix diffusion coeffi-cients of all tracers.

Figure 17 shows tracer concentra-tions at the second well as a functionof time for one of our field tests. Themost striking feature is the curves’ bimodal shape. This shape is attributedto a relatively small fraction of thetracers moving quickly between thewells through densely welded, frac-tured tuff, while most of the tracerspassed more slowly through partiallywelded, less fractured tuff. These results support the concept of dual-permeability that is incorporated intoour simulation models: flow occurs pri-marily in fractures, but a large amountof near-stagnant water is present in the rock matrix to diffuse the tracers.

Looking more closely at the curvesfor the two nonsorbing solutes, we seethat the PFBA peaks are slightly higherthan the bromide peaks, that the secondbromide peak occurs later than the cor-responding PFBA peak, and that thebromide curve eventually crosses overthe PFBA curve. These features are allconsistent with greater matrix diffusionof the bromide, which is more diffusivethan the PFBA.

The lithium curve is more attenuatedthan the curves for the two nonsorbingtracers, which indicates sorption. Theattenuation in the first lithium peak isalmost exclusively a lowering of theconcentration with little or no delay in

Yucca Mountain


Figure 16. Neptunium Transport inthe Saturated Zone Even weak sorption significantly retards

neptunium transport through the saturat-

ed zone. The two sets of graphs depict

a worst-case scenario for waste leakage:

the first waste canister fails at 1000

years, after which neptunium leaks

continuously over the next 30,000 years.

The upper graph shows the flux of

neptunium-237 that crosses into the

saturated zone. The lower graph shows

resulting neptunium concentrations at

the respository’s site boundary, 5 km

downstream, for two cases: no sorption

(solid line) and weak sorption (dotted

line).

237 N

p co

ncen

trat

ion

(mol

/L)

Time since emplacement (yr)103 104 105 106

Time since emplacement (yr)103 104 105 106

102

237 N

p flu

x at

wat

er ta

ble

(mol

/yr)

101

100

10–1

10–8

10–9

10–10

10–11

Kd = 0

Kd = 2 mL/g

Yucca Mountain


arrival time. This reduction suggeststhat lithium diffuses into the matrix andis then sorbed. The attenuation in thesecond peak involves a clear time delayalong with a dramatic lowering of con-centration. This combination suggestslithium sorption in both the fracturesand the matrix.

The microspheres reach the secondwell slightly earlier than the solutes,but their peaks are significantly attenu-ated. The attenuation implies that therock matrix filters out a fraction of themicrospheres, and the curve’s bimodalshape suggests that two transport path-ways are also available to the spheres.(These data and colloidal transport ofradionuclides in general are discussedin the box on page 490.)

We were able to fit the curves ofFigure 17 using a semianalytical, dual-porosity transport model, demonstrating

that our simulations are consistent withfield data. One important finding wasthat lithium sorption measured at the C-Wells was always equal to or greaterthan the sorption measured in laborato-ry tests. This correlation increases ourconfidence in laboratory measurementsfor radionuclide sorption onto YuccaMountain tuffs.

Will the Repository Work?

Where have our extensive researchand analysis brought us? Can a YuccaMountain repository and its natural bar-riers safely isolate nuclear waste for10,000 years? What radiation dosesmight the residents of Amargosa Valleyreceive 10,000 or 100,000 years fromnow, and when would those dosespeak? The project’s performance as-

sessment team is beginning to answersuch questions by combining the find-ings of the various laboratories study-ing the site. Input to its performance assessment includes calculations ofwaste-canister failure mechanisms andrates, of the effects of a volcanic erup-tion at the repository, of how radionu-clides dissolve and are transported ingroundwater, and of how these radionu-clides would disperse in the biosphereonce they reach Amargosa Valley.

The team’s full report will not bepublished until 2001. However, esti-mates to date for two repository designsare that radiation doses to valley resi-dents will be only 0.1 mrem/yr 10,000years after the repository is closed andonly 30 mrem/yr 100,000 years later.The latter dose is close to the EPA’sregulatory limit of 20 mrem/yr. Radia-tion doses will reach a peak of200 mrem/yr 300,000 years after therepository is closed. This peak, whichis two-thirds of today’s backgrounddose from natural sources such as cos-mic rays, radon, and radionuclides inthe soil, results primarily from weaklysorbing neptunium and, later, from colloid-transported plutonium.

But again, these estimates are notthe final word. For one, data are stillbeing collected, such as from our ongo-ing Busted Butte tests. In addition, theperformance estimates are based onvery conservative assumptions, and ourstudies have shown radionuclide trans-port to be sensitive to a range of para-meters. For instance, Figure 14 showsthe impact strong capillary forces haveon matrix diffusion, and Figure 16shows the impact that even weak sorp-tion in the saturated zone can have ontransport times for neptunium. Morelaboratory and field data should yieldmore realistic assumptions. Finally,there is the challenge of predicting theperformance of natural and engineeredbarriers over geologic time scales.

As the assessment team points out,“whether these calculated annual doseswill actually occur cannot be physicallydemonstrated or scientifically proven.The performance of a repository over

Nor

mal

ized

con

cent

ratio

n

100

10

1

0

PFBA

Br–

Li+

Microspheres

10 100Time (h)

1000 10,000

Figure 17. Saturated-Zone Field Tests at the C-Wells A series of field tests at the C-Wells complex near the repository site helped bench-

mark our saturated-zone modeling parameters. The curves show the normalized con-

centrations of tracers injected at one well that were detected at a second well 30 m

away. The data indicate multiple transport pathways for all tracers and validate matrix

diffusion and sorption as important retardation mechanisms for radionuclide transport

in the saturated zone.

such long periods—longer than record-ed human history—cannot be tested inthe same way that the performance ofan airplane, for example, can be test-ed.” Proof of the repository’s viabilitycannot be given in the ordinary sense ofthe word. However, the project’s itera-tive process of experimental measure-ments, modeling calculations, andanalysis is moving us ever closer to amore accurate portrayal of repositoryperformance and hence to greater confi-dence in our predictions.

Epilogue

We may pause at this point and askourselves what Yucca Mountain willlook like 10,000 years from now. Willblack-tailed jackrabbits and desert col-lared lizards still scurry across a brown,nondescript ridge, hiding behind the occasional Yucca plant? Or will the climate have changed, making themountain greener, perhaps covered with a dense juniper forest? If a solitaryhiker, reaching the ridge top, comesacross a monument inscribed with hieroglyphics, will she be able to decipher the message? Will she realizethat beneath her feet lies the entombedwaste of the 20th century’s nuclear age?

Perhaps in her pack she’ll carry afaded graph handed down over the agesfrom an ancient ancestor, the computermodeler. The graph will have beenpassed on, parent to offspring, for gen-eration upon generation. According tothe graph, by her time in the 121st century, the total dose for all pathwaysand from all radionuclides should stillbe barely 0.1 mrem/yr.

Perhaps in her backpack she’ll alsocarry the instruments needed to checkher ancestor’s prediction. After examin-ing the mountain for signs of erosion,new earthquake faults, or volcanismand after checking the meteorologicalstation for annual rainfall, perhapsshe’ll return to her solar-powered vehi-cle and drive over to Amargosa Valley.If farmers there are still growing crops,what will she find when she pulls out

her mini-mass spectrometer and mea-sures the concentration of neptunium inthe their irrigation water? Will shegrumble at the stupidity of her ances-tors or smile at their wisdom? Accord-ing to the statistical analysis given inthe faded graph she’s carrying, there’s a90 percent chance she’ll smile. ■

Acknowledgments

The research by Los Alamos onYucca Mountain has so many facetsand is so extensive that anyone closeto the project can quickly lose the gen-eral reader in massive amounts of detail. We would like to acknowledgethe difficult and important work ofJudy Prono on this article. She broughtthe right amount of objectivity to thetask, asking various investigators toexplain or amplify on the materialuntil she gradually, paragraph by para-graph, section by section, made the article much more accessible to a wideaudience.

In this same vein, the article repre-sents the work of many individuals atthe Laboratory over several decades.The people listed at the beginning ofthe article or pictured at the end areonly representative of the entire effort.Others whose work also contributed the Yucca Mountain Project includeKay Birdsell, Kathy Bower, Dave Broxton, Katherine Campbell,Dave Clark, Jim Conca, Bruce Crowe, Dave Curtis, John Czarnecki, Zora Dash, Clarence Duffy, NedElkins, Claudia Faunt, Dick Herbst,Larry Hersman, Stephen Kung, EdKwicklis, Ningpin Lu, Arend Meijer,Dave Morris, Mary Neu, HeinoNitsche, Ted Norris, Don Oakley, Jeffrey Roach, Robert Rundberg, Betty Strietelmeier, Kim Thomas, JoeThompson, Bryan Travis, Lynn Trease,Peng-Hsiang Tseng, Greg Valentine,Kurt Wolfsberg, and Laura Wolfsberg.I would especially like to thank ChuckHarrington, Frank Perry, Tom Hirons,Ken Eggert, and Wes Meyers for supporting the work of assembling

this article, critiquing early drafts, or providing helpful material on thehistory of Yucca Mountain.

Further Reading

CRWMS M&O (Civilian Radioactive Waste Management System, Management and Operating) Contractor. 2000. “Analysis of Geochemical Data for the Unsaturated Zone (U0085).” CRWMS M&O report ANL-NBS-HS-000017.

2000. “Calibration of the Site-Scale Saturated Zone Flow Model.” CRWMS M&Oreport MDL-NBS-HS-000011.

2000. “Particle Tracking Model and Abstraction of Transport Processes.” CRWMSM&O report ANL-NBS-HS-000026.

2000. “Saturated Zone Transport Methodology and Transport Component Integration.” CRWMS M&O report ANL-NBS-HS-000036.

2000. “Unsaturated Zone and Saturated Zone Transport Properties (U0100).” CRWMSM&O report ANL-NBS-HS-000019.

New Questions Plague Nuclear Waste Storage Plan. 1999. New York Times. August 10 issue.

Office of Civilian Radioactive Waste Management. 1999. “Viability Assessment of a Repository at Yucca Mountain” (December).U.S. Department of Energy report DOE/RW-0508. (The full report is available on the World Wide Web at http://www.ymp.gov/va.htm)area

Yucca Mountain


Roger Eckhardt has a Ph.D. in physical chem-istry from the University of Washington but hasspent a great deal of hisprofessional life involved with scienceeducation and sciencewriting. Most recently,he has been helping tocompile and write thevoluminous documentsthat describe the scien-tific work being carriedout at Los Alamos forthe Yucca MountainProject.

Yucca Mountain


Andrew V. Wolfsberg Modeling of colloidsand chlorine-36 transport.

Paul R. DixonProject manager.

Gilles Y. Bussod Busted Butte unsatu-rated-zone transport test.

Inés TriayYucca Mountain geo-chemistry, includingsorption, diffusion,and transport in tuffs.Currently, DOE man-ager for WIPP.

James W. CareyMineralogical modeling of YuccaMountain.

June T. Fabryka-MartinYucca Mountainchlorine-36 studies.

Schön S. LevyAlteration history ofvolcanic rocks atYucca Mountain.

Paul W. ReimusC-Wells studies atYucca Mountain.

Wendy E. SollModeling of BustedButte hydrology.

Carleton D. Tait Actinide solubilityand speciation inYucca Mountaingroundwaters.

Jake TurinField hydrology forBusted Butte.

David T. Vaniman Igneous petrology andmineralogical analysisof Yucca Mountaintuffs.

Steve J. ChiperaX-ray diffractionmineralogy.

Mark T. PetersYucca Mountaintesting, including theExploratory StudiesFacility and BustedButte experiments.

David L. Bish mineralogy ofYucca Mountaintuffs.

George A. ZyvoloskiModeling of flow and transport in thesaturated zone.

Wolfgang H. RundeActinide solubilityand speciation atYucca Mountain.

Julie A. CanepaFormer project manager.Currently, programmanager of LANL Environmental Restoration Project.

Bruce A. RobinsonModeling of flow and transport in theunsaturated and saturated zones.

Principal Los Alamos Contributors to the Yucca Mountain Project

Maureen McGrawModeling of colloid-facilitated transport.

Documents

Yucca Mountain - Federation of American Scientistsbrought in by rail, canisters would be placed on support cradles by remotely controlled cranes. (b) Multiple Containment Barriers