ML20101D375

Initial Demonstration of the U.S. NRCs Capability to Conduct a Performance Assessment for a HIGH-LEVEL Waste Repository

ML20101D375
Person / Time
Issue date:	05/31/1992
From:	Codell R, Eisenberg N, Fehringer D, William Ford, Margulies T, Timothy Mccartin, James Park, Randall J NRC OFFICE OF NUCLEAR MATERIAL SAFETY & SAFEGUARDS (NMSS), NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES)
To:
References
NUREG-1327, NUDOCS 9206150272
Download: ML20101D375 (167)

Text

{{#Wiki_filter:.

p 6'y

^ g h NUREG-1327 3 -S Initial Demonstration of the LXRC's Capability tox Conduct a Performance Assessment for a ~ CHigi-Level Waste Repository j t 9

_l i.

)UiSkNhcle'ariRsgulatory Commission: Office of Nucl$arlM.at rial Safety and Safegua' ds r 1 . OtYice of Nuclear Regulatory Research.

iRf Codell, N. Eisenbergl DJ Feliririger, W. Ford,

.T!Margulies,' T. McCartinlJ. Park, J. Randall g(Isho g.,

o x,x.;., 615e 92c$3; 1327 R PDR

w .y .m s a i - AVAILABILITY NOTICE

Availability of Reference Materials Cited in NRC Publications f

f-J Most. documents cited (n'NRC publications will be available from one' of-the following. sources: : 1, = The NRC Public Document Room, 2120 L Street, NW., Lower Level, Washington, DC .-20555 ,2." ; The Superintendent of Documents, U.S. Government Printing Office, P.O. Box 37082, 'i 1 Washington, DC120013 7082: a '3. The National Technical Information Service, Springfield, VA 22101 iAlthough the listing that tvuows represents the majority of documents cited in NRC publica-tions, it is not intended to be exhaustive, n . Referenced. documents avs able for inspection and copying for a. fee from the NRC Public' 1 L Document Room include NRC correspondence and intamal NRC memorandaLNRC buNetins.

circulars, information notices, inspection and investigation notices; licensee event reports; vendor reports and correspondence; Commission papers; and applicant and licensee docu-ments and. correspondence, 4

LThe following documents in the NUREG series are available for purchase from the GPO Sales ~ Programt iformal NRC staff and contractor; reports, NRC-sponsored conference proceed-Lings, international' agreement reports, grant publications, and NRC booklets and brochures, i Also available are regulatory guides, NRC regulations in tne Coce of Federal Regulations, Iand Nucles: Regulatory Commission Issuances. Documents available from the Na'tional Technical Information Service include NUREGs,eries " reports and. technical reoorts prepared by other Federal agencies and reports prepared by. _ the Atomic Energy Commissioniforerunner agency to the Nuclear Regulatory Commission. o Documents. available' from public and special technical libraries-include all open literature. l items, such as books; joumal articles, and transactions.. Federal Register _noticos Federal f arid Statelegislatione and congressional' reports can usually be obtained from these ~ litwar'ese H i Documents such as theses.: dissertations, foreign reports and translations, and non-NRC - 1 = conference proceedings are available for purchase from the organization sponsoring the. .l Lpublication cited; Single copies of NRC draft reports are available free, to the extent of supply, upon written j request to.the Office of Administration,. Distribution and Mall Services Section,-U.S. Nuclear < Regulatory Commissioni Washington, DC 20555, ~ (Copies of industry codes and standards used in a substantive manner in the NRC regulatory ? process are maintained at the NRC Library, 7920 Norfolk Avenue, Bethesda, Maryland, for use by the public._ rCodes and standards are usually copyrighted and may be purchased i from the originating. organization or, if they are.American National Standards, from the ] t American National Standards institute,1430 Broadway, New York,- NY 10018.

@hhhGfffm 7 g y@ g$ $ g& m u MR 3 a gg op, u $4mw&T $$;i ( }, &,. $ f, (,h:$ r p ~_ M. mm ga %.MMM fp.mwpw$pkph F i * '" g W yQf gg p.gn% p%nwyym~ ;..g} 9 x j uu . 9m). $g3,. ;.';A '4 c..,WM$b$Y k., c. .@ gi % ;,s@N& v 4;AM,ty 'g' I- ! + itg.9 %' i.,e Ne 9 '... i ri.. L Tf QN[y v.M$,A.: A 9 ,,y' > 9@g NC-Ys., y$.w '.3 $;j 9 c. n. x. x.:. ro p me,. g ; s. p.w 2. m,:- y s gj, a

. E s c :( ! ;j :s ir- ?! % Y L d 3!. M

"/$.(R N:1.[: E l c: 8.# f;,4 ?% 7 f 2 h - md m 3 e % g w w,L.m$ y a@ _? @ -@.l _% a m@a,Q. %_.e- @ gy% Q uyj .m m.; ..v : e; a w y.* g.3 3 v x ? m :. m w w w % w%,>%d h [P y N p;py % gd. ph@f N k' lN " y Cli MM.QQ+:.Af%p dg!.kf$e.bM. h Mghp M Q' E h% $E 0 74 g Z vw.N b hp w7 p P xws w h.,. 3 r e p s f gyga.jggQMygQQJ:#Qf jg,V e - pM g. M lll 4

w..s4 4 spy.

,E- % g, y %y, .?.pm).. ,fhf Y ' y%p, f. i. : f.Q ,_ \\ [ f.k f '.'.l;,4 d@):q h ;' lQ k2%y.fj 1 h.. R$$$qfy 7 - ;9 y

w.,

h_f

l?q$h(f%g.h,.d.$.g?c:s;hhh. m -

~ ed

(

y. ..s, p $ u ?.(. e s .. u g i,i f. a,g N l Y $ i &y ". e n gh. Q :,. $,AiW nglh .&. &. 4,: n n:L . > g;w.;s, 4,...p h .A m n W ~ h,_.h 4.s. lY:n. qe f b.'h y m ;u w s 4 ),:;L.6 a a w jv.;- :,y,A $ ;. Q: f ~ b T.., Q:-A f.pg.p... k ... 4.q +,.' .s .(51.[lh Q bkN % ( {. If, "I '4 h l h _,... _. 4 s, -. ' '4...

n.,4.g

.y gg 4 a A. aA Mw; mlsg. NM m ..+e t.

,m

..w. [f;r'h '.hw:~,. N h Y j'h.. ,h.h,4 .4 -,D Q d* Y 'g-4 t.r - :. i...ye; c..... 3 .%w n.m m.. ma f ( 4, .me_ 2 p_ _ p g;gg3gggggg,g7g w maawa"IL 1m" u 'r~ w"l+m ', N %I _ Mhk@Q 9206150272 920531 l@ W p ( Y O q* 7' j g %. g / g UE '= rM, ;4 W %*, ( > Mph PDR NUREC W m..j@ J 2 4 gg 1327 R PDR C .?QNWMREQM@SWy!N$gisfgG $p 9 h 1-i a1$ gg < fWig g d t. #

E R M EElf _ ___ ___ _,,3 y p f wl q <( m e,ee # gen mg , u u s m,a m]s;gggy t j! y7 f 4f 7 m un a u,x w 1 w J p 'tp 'g *g g fq av, ua ep y E~ y og y"g yg }i i: 3 t j 4 ,w g~ n n g,,, 7 g,g ,t p. g e my y , ~~ , n g 7 3 -f ff[ - wwcr 3 .mo - ,A -, ww e m a. a., i (q_ g a. s n '.Mm E h mos n. qN N_,-.h > m [ - x o, m. a. ,u c s. g j . y 7 >g .g a e -[ .$$ ?A ~ j ': .1 s g.y;; i s- < q 1, y ' b W y, ,s. . 'p-jl ? g j y. y p j$ k4 i. yp - k-f' [h

q..

,,E d T ?c -. 2.rynjW 'J+b(K"' 3;^g v l m., f;.gj - Q, p,q y,t., dem $ M ca !! i,, { l w' em% .t y c7 g y y d p . F , u s . b;$ [ 0 r p u l h y j" jk .: { .L t ,.O 8%n u _V d_;

1. 4ggg

\\ k. L. q mo!w:@n Ywa xmf QQ& -~h N 3 v' + .-~ [L e e ~ y, y a.yajn 4 - m J g~g 4, w w -n - - w s 'pf), f.j)jg aggg(j g I wuou wamee gcg_y _ %y_4 7.c mErmw4%pdhpoy. og s. .j s t .r ggj s ee y f K gggA 3 aa

NUREG-1327 Initial Demonstration of the NRC's Capability to Conduct a Performance Assessment for a High-Level Waste Repository hianuscript Completed: March 1992 Date Published: Mey 1992 ' R. Codell, N. Eisenberg, D. Fehringer, W. Ford, T. Margulies*, T. McCartin', J. Park, J. Randall* I Omcc of Nuclear Material Safety and Safeguards Office of Nuclear Regulatory Research l U.S. Nuclear Regulatory Commission Washington, DC 20555 ,p" '%, s +- i l

• Office of Nuclear Regulatory Research

ABSTRACT l In order to better review licensing submittals for a liigh-from a repository via the groundwater and direct release Ixvel Waste Repository, the U.S. Nuclear Regulatory 1.athways provided preliminary estimates of releases to Commission staff has expanded and improved its capabil-the accessible environment for a 10,000 year simulation ity to conduct performance assessments. This report time.12ttin hypercube sampling of input parameters was documents an initial demonstration of this capability. The used to express results as distributions and to investigate i demonstration made use of the limited data from Yucca model sensitivities. This methodology demonstration i Mountain, Nevada to investigate a small set of scenario should not be interpreted as an estimate of performance classes. Models of release and transport of radionuclides of the proposed repository at Yteca Mountain. Nevada. l l l l l-iii NUlt!!G-1327

CONTENTS Page - Abstract............................................,....... iii - Executive S ummary.......................................................................... xiii Ac kn owl ed gm e n t................................................................................ xxi 1 In ti od uction............................................................................... 1 5 2 Purpose and Scope...................... + 3 O rganization and Staffing.................................................................... 9 4' SystemCode.............................................................................. 11 4.1 In t rod u ctio n..................................................................... 11 4.2 - Require,ments for the Development of the System Code...................................... 11 4.3 S urvey of Exist ing Cod es............................................................. 11 4.4 Description of th e System Code......................................................... 12 4.4.1 I n t rod u ct io n..................................................................... 12 4.4.2 Internal vs. External R uns...................................................... 12 4,4 3 Inpu t.......................... 12 J 4.4.4 O pe ration........................................................................ 12 4.4.5. Output......... 13 5 ' So u rce Term............................................................. 19 5.1 In t rod uctio n......................................... 19 5.2 Review of Important Issues for Selecting Source Term Models................................ 19 5.2.1 -' Waste-Package Lifetime.................... 19 5.2.2 Cl adding Fa il ure.................................................................. 19 5.23 Oxidation of Uranium Dioxide Matrix............................................... 19 5.2.4 Release of Dissolved Radionuclides from the Fuel.................................. 20 5.2.4.1 Water contact fraction..... 20 5.2.5 Release of Gaseous Radionuclides............................ 20 53 Model Selection and Justification................ 21 53.1 Model for Dissolved Radionuclides................................................. 21 53.2 Limitations of Model for Dissolved Radionuclides.................................. 21 53 3 - C-14 Release Model................................. 22 5.4 Source Term Inventory....... 22 -6 Flow and Trans port Model s............................................................... 23 6.1 Introduction............................. 23 6.2 Definition of Issues for Selecting Performance-Assessment Transport Models........... 23 6.2.1 Site Concepts... 23 6.2.2 Pathways........... 25 -6.23 Flow and Transport Pathway Phenomena............. 25 y NUREG-1327

4 i I . CONTENTS (continued) Page 6.23.1 Liquid transport... 26 6.23.2 Gas transport............. 27 6.2.33 Direct transport................. 27 63 Computer Program Review and Selection............................ 28 6.3.1 Liqu id Pathway.............................................................. 28 63.1.1 Regional groundwater flow programs...........,.......................... 28 63.1.2 'l\\vo. Phase flow programs.................... 29 6.3.13 G eochemical programs........................................... 29 63.1.4 Transport programs.................................................. 29 6 3.2 G as Pa t h way.............................................................. 29 63 3 a)irect Pat h way............................................................. 29 7 - Methodolop for Scenario Development............. 31 7.1 Introduction.................. 31 7.2 M e thodology.................................... 32 73 Application......................... 32 7.4 Conclusions............................................................. 33 . 8 Auxiliary Analyses Summaries............... 41 8.1 I n t rod uct ion..................................,................. 41 8.2 Carbon-14 Analysis (Appendix D).................. 41 83 Statistical Convergence (Appendix E).... 41 8.4 Analysis of Hydrologic Data (Appendix F)............., 41 8.5 Two-Dimensional Flow Simulation (Appendix G). 41 . 9 Analysis and Result s........,...................... 43 9.1 Treatment of Scenarios........................... 43 9.1.1 Introduction....... 43 9.1.2 Discussion............................ 43 - 9.2 NEFTRAN Source Term Modet........... 44 93 Flow and Transport Models................ 45 93.1 - 1.iquid Pathway.........,. '45 93.1.1 - NEFTR AN network implementation............... 45 93.1.2 Implementation of matrix and fracture flow in NEFTRAN....... 45 93.13 Implementation of transport phenomena within NEFTRAN. 50 93.1.4 Spatial variability of flow and transport parameters.... 51 93.1.5 Effective values of flow and transport coefficients.............. 52 93.2 Gas Pathway.... 53 933 Direct. release (Drilling) Pathway.. 53 NUREG-1327 vi

A s CONTENTS (continued) b Pare i 53 9.4 Param et ers................................. 53 9.4.1 ; Liq uid Path way................................................................. 53 9 4.2 Sampling hrameters for NEFFRAN Analysis......................................... 54 9.4.2.1 Waste.packa ge lifetim e.................................................. 54 9.4.2.2 Solubility of uranium dioxide matrix......,............................... 54 9.4.23 D ispersivity............................................................. 54 9.4.2.4 I rfilt ra tion rate...................................................... 57 l 9.4.2.5 Fraction of water contacting waste.......................................... 57 I 9.4.2.6 Saturated hydraulic conductivity............................................ 59 9.4.2.7 Spatial correlation of saturated hydraulic conductivity..................... 59 9.4.2.8 Po rosi'y............................................. 59 9.4.2.9 B rooks-Corey coct ficients............................................ 59 9.4.2.10 Retardation coefficients.......................................... 60 9.4.2.11 Sol ubilities.......................................................... 60-9.43 Direct. Release (Drilling) Pathway........................................ 60 9.5 Sensitivities and Uncertainth s for Liquid-Pathway Analysis................................. 60 9.5.1 I n t rod u ct ion................................................................. 60 9.5.2 Statistical Uncenainty Analysis...................... 61 9.5.2.1 l atin Hypercube Sampling (LHS)............................................ 61 . 9.53 ' Ad Hoc Sensitivities...................,........................... 62 9.53.1 Se nsitivity to infiltration................................................. 62 ' 9.5.4 Sensitivity Analysis Using Regression................................................ 63 9.5.5 Average importance of Radionuclides............................................ 63 9.5.6 Sensitivity to NRC Performance Criteria..... 63 9.5.6.1 Effects of NRC performance criteria on CCDFs.............................. 74 9.5.6.2 Average contributions by radionuclide........................................ 9.5.63 Ad Hoc Sensitivities to NRC Criteria........................................ 74 '75 9.6 Total Syst em Result s................................................................... 75 9.6.1 : I n trod uct ion.........................................,....................... 76. 9.6.2 Partial CCDF Results...................................................... 9.6.2.1 Undisturbed or base-case conditions.................... 76 76 -- 9.6.2.2 Pluvial conditions ~....................................................... 76 9.6.23 Drilling under undisturbed conditions................ 76 9.6.2,4 Drilling under pluvial conditions 76 9.63 Results for the Total CCDF....... r vii NUREG-1327

l l l CONTENTS (continued) Page 10 Preliminary Suggestions for Further Work............. 87 10.1 Improvements and Extensions to Modeling................ 87 10.2 Improvements and Extensions to Auxiliary Analyses............................... 90 10.3 Recommendations for Additional Scientific Input...... 91 R. REFER EN CES.......................................... 93 APPENDICES rase A System Code Review.............. A-1 n Source Term Code Review........... n-1 C Flow and Transport Code Summaries....................... C-1 D ' Gaseous Releases of Carbon-14......... D-1 E Testing Statistical Convergence..... E-1 F Analysis of Hydrologic Data........ F-1 G wo-Dimensional Cross Sectional Flow Model..................... G-1 H Analysis for Drilling Scenario.......................... H-1 I System Code Steps............... 1-1 J Documentation of Files and Programs on INEL CRAY XMP/24 for Repository Performance Calculations...................... J-l FIGURES ES.1 Components of a total system performance assessment. xviii ES.2 Composite CCDF curve for the scenario classes considered in Phase 1 of the Iterative Performance Assessment............. xix 1.1 Components of a total system performance assessment......... 3 4.1 Flow diagram of external and internal modes of operrition. 14 4.2 Determination of scenario probabilities from the probabilities of fundamental cvents. 15 4.3 Simplified flow diagram of system code steps. 16 4.4 Detailed flow diagram of system code steps.. 17 4.5 System code data array for normalized radionuclide releases. 18 NUREG-1327 viii

FIGURES (continued) Page 9.1 Conceptualization of a hydrogeologic cross-section through the unsaturated zone at Yucca 46 Mountain (after DOE,1988).................... 9.2 - Hydrostratigraphic units used to simulate the variation in depths and units existing below the 47 repository.............................................................................. . 9.3 Geologic map of Yucca Mountain showing repository drift perimeter (after DOE.1988)............. 48 9.4 Representation of the allocation of repository area and radionuclide inventory of the four N EFTR AN s im u lations.................................................................. 49 9.5 CCDF for Base Case; 500 Vectors,10,000 Years............ 65 66 9.6 CCDF for 100,000_ Years,500 Vectors... 9.7 CCDF for 100,000 Yean, 500 Vectors............ 67 9.8 CCDF for 100.000 Years,500 Vectors. Compare Column D to all 68 9.9 CCDF for Pluvial Secnario; 98 Vectors Only....... 69' 9.10 Base Case Conditional CCDF; 10.000 Years........... 70 9.11 Base Case Liquid Pathway Scenario; 10,000 Years Effects of Groundwater Travel Time on E PA R el ease............................................. 71 9.12 Base Case Liquid Pathway Scenario; 10,000 Years Effects of Release Rate from Engineered Barrier.. 72 9.13 Base Case Liquid Pathway Scenario; 10.000 Years Effects of Engineered Barrier Lifetime.......... 73 -9.14 Determination of scenario probabilities from the probabilities of fundamental events 78 9.15 - Graphical representation of hypothetical CCDF with the EPA containment requirements 79 (DOE,1988)........................................................................ 9.16 Plot of an empirical CCDF against the EPA containment requirements (DOE,1988).............. 79 9.17 Partial CCDF for Undisturbed Conditions................. 80 9.18 Partial CCDF for Pluvial Conditions................................. 81 9.19 Partial CCDF for Drilling Under Undisturbed Conditions 82 l . 9.20 Partial CCDF for Drilli.ng Under Pluvial Conditions........ 83 9.21 Composite CCDF Curve for the Scenario Classes Considered in Phase 1 of the Iterative Performance Assessment...... 84 85 9.22 Total CCDF for Phase 1 of the Iterative Performance Assessment l D.1 Release of C-14 inventory......... D-$ 1 I ix NUREG-1327

FIGURES (continued) Pay D.2 Gas comection velocitics alony the repository center line at 100 years after waste emplacement,, D-7 D.3 Carbon-14 travel time from the repository to the surface for ambient conditions,2.000,10,000. and 50,000 years (modified from Ampter, et al.,1988)... D-8 E.1 Statistical Convergence; 100 vs. 500 Vectors,10,000 Years, Ilase Case E-2 F.1 Histogram of Topopah Springs Porosity Values.. F-2 F.2 Histogram of Calico Hills Porosity Values. F-3 F.3 location map of holes used in Scatter and Variogram Plots... F-4 F.4 Seatter Plot, Hole USW GU-3, Topopah Springs Unit F-5 F.5 Scatter Plot, Hole USW GU-4, Topopah Springs Unit F-6 F.6 Scatter Plot Hole UE25a-1,Topopah Springs Unit. F-7 F.7 Porosity Variogram, Hole USW G-4, Topopah Springs Unit....... F-S F.8 Porosity Variogram, Hole UE25a-1, Topopah Springs Unit. F-9 G.1 Hydrogeologic units and boundary conditions used in the cross setional simulation using the VAM2D computer program..... G-2 H.la Intercept geometry for vertical emplacement 11-3 H.lb Intercept geonatry for horizontal emplacement H-3 ~ H.2 Geometry for volutie excavated by a borehole through a horizontally emplaced waste canister. H-5 H.3 Conceptual view of dt? ting through waste canister and contaminated rock below the repository and drilling into contamaated rock. H-11 H.4 Depiction of quantity of waste in packages and host rock from the time of waste emplacement until the end of teaching (all waste leached from waste package) H-12 TABLES 5.1 Radionuclide Initial Inventory (Doctor, et al.,1992). 22 6.1 Identification of Liquid Pathway Processes and Estimated Effect on Calculating Cumulative Release from the Liquid Pathway. 27 7.1 List of Processes and Events 34 7.2 Descriptions of Processes and Events 36 NUREG-1327 x l

h TABLES (continued) 4 Page 9.1 Columns Representing the Yucca Mountain Repository........................................ 50 9.2. Examples of Known and Suspected Correlations.............................................. 54 9.3 ' Input to latin Hypercube Sampling Program................................................. 55 9.4 . infil tra tion Estimat es................................................................... 58 9.5 Mean and Standard Deviation (S.D.) of Log ks................................................ 59 95 Mean Porosity for Hydrogeologic Units................................................... 59 9.7 B rooks.Corey Coeffici en ts.........................................,..................... 59 9.8 Retardation Coefficients for Fractures 60 9.9 Steps to Perform Uncertainty and Sensitivity Analysis....................................... 61 9.10 Regression of Liquid Pathway Cumulative Releases (Raw data cortclations)...................... 63 . 9.11 Average Importance of Radionuclides to EPA Release Limits................................. 64 9.12 Fractional Contribution by Radionuclide to EPA Release Ratio for Unrestricted Vectors and Rose Restricted by NRC Performance Criteria............................................. 75 D.1 - Calculated Zirconium Oxide Thickness............... D-3 - D.2 Release Fraction as Function of Release Time............................................. D-9 G1 Hydraulic Properties Used in the Two-Dimensional Simulation of a Layered Tuff Site > (Prindl e. 19 87)................................................................... G-3 G.2 Ratio of Horizontal to Vertical Flow at the Interfaces Between Different Hydrologic Units over Differing Infiltration Rates '............................................................ G-3 7 xi NUREG-1327

EXECUTIVE

SUMMARY

ES.1 INTRODUCrlON ES,2 PURPOSE Given this background, the primary focus of the Phase I he objective of this effort was to expand and improve the activities was to demonstrate the staff capability to con-U.S. Nuclear Regulatory Commission (NRC) staff capa-duct a total system, performance assessment in an inde-bility to conduct performance assessments independ-pendent fashion. Ily demonstrating such an independent ently. By expanding and developing the NRC staff capa-capability, the NRC staff has provided evidence of a de-bility to conduct such analyses, NRC would be better able gree of readiness for the forthcoming review t f licensing to conduct an independent technical review of the U.S. material to be provided by the DOE. In addi" m, by exer-Department of Energy (DOE) licensing submittals for a e sing this capability for independent res ' v, the NRC high-level waste (HLW) repository. staff has accomplished several secondat, objectives, in-cluding: These activities were divided initially into Phase 1 and 1. Performing an evaluation of the adequacy of existing Phase 2 activities. Additional phases inay follow as part of analytical tools, both methodologies and computa-a prcgram of iterative performance assessment at the tional methods: NRC.The NRC staff conducted Phase 1 activities primar-ily in CY 1989 with minimal participation from NRC 2. Obtaining valuable insights into the need for further contractors, ne Phase 2 activities were to involve NRC development of methodologies and computational contractors actively and to provide for the transfer of IU"E technology. Phase 2 activities are scheduled to start in CY 1990, to allow Sandia National Laboratories to com-3. Obtaining valuable insights into the data needed, plete development and transfer of computer coocs and from the DOE Site Characterization Program, to the Center for Nuc! car Waste Regulatory Analyses conduct performance assessments, including the (CNWRA) to be in a position to assist in the acquisition of priority of these data needs. (llecause of the unwr-the codes, tainties in the analysis, these msights are limited, especially for this Phase 1 effort.) The results presented here have had limited peer review, have numerous simplifying assumptions, consider only a ES3 SCOPE limited number oficenarios, and are based on limited data; thus, the numerical results should not bc.aken as ne performance assessment is considered to be com-representative of the performance of the proposed re. prised of two parts: pository at Yucca Mountain, Nevada. The analysis is also replete with uncertainties regarding conceptual models, 1. quantitative estimation of total system performance data, physicochemical models, and models and data for through the use of predictive models, and predicting scenarios. In the conduct of this limited study, . the authors did not encounter any problems indicating 2. documentation and detailed auxiliary analyses, that the U.S. Environmental Protection Agency (EPA) where appropriate, to support the assumptions, standard could not be implemented.110 wever, due to the data, and modeling approaches used to obtain quan-incomplete scenario analysis in this demonstration, not all tilative estimates of performance, aspem of the standard were tested (e.g., the difficulties in estie.ng scenario probabilities). Therefore, taking Both aspects of performnce assessment were addressed these tentative results of a preliminary analysis out of in the Phase i effort. context, or separating these tentative results from these De focus of this Phase I demonstration was the EPA caveats, may lead to the inappropriate interpretation and containment standard that requires the total system per-use of the results. formance measure for an HLW repository to be ex-pressed by a complementary cumulative dis'ribution His report is intended to demonstrate the capability to function (CCDF) of radionuclide releases to the accessi-conduct a performance assessment. He report is not ble environment, weighted by a factor approximately pro- . intended to provide guidance on performance assessment portional to radiotoxicity, integrated over an appropriate methods or on the conduct of NRC staff reviews of per-period of time (10,000 years is the current regulatocj formance assessments. Furthermore, it should not be requirement).This peiformance measure was estimated considered as NRC staff guidance on the interpretation by following the steps outlined in the irJormation flow and implementation of NRC rules and rebulations. diagram (Figure ES.1). For the Phase 1 effort, these steps xiii N UREG-1327

Etecutive Summary were all executed, but some (steps 2 and 3) were only sis, their bases, and the implications of their uses executed to a limited degree. 'Ihese steps are described explicit. briefly below: Two types of uncertainty are usually treated explicitly in 1. System Descriptiem-In this step, the various impor-the generation of the CCDF;(1) uncertainty due to future tant components of the waste-disposal system-the states of nature and (2) uncertainty in the values of pa-waste form, the engineered barrier (the canister, the rameters determining system perf ormance. Modeling un-repository, backfill, if any), and the site-are de-certainty, including concept ual model uncertainty, is usu-scribed in tetms useful to mo& ling radionuclide ally not treated expbcitly in the generation of the CCDF. migration to the environment.'lhis step usually re-The CCDF is a curve of the likelihood that the conse-quires the synthesis of information from many dif-quence is more than a certain magnitude. For the reposi-ferent disciplines in the natural sciences and enpi-tory system, considerable uncertainty exists concerning

neering, the values of parameters used to estimate the conse-quences of the repositoryflhis uncertainty is displayed on 2.

Scenario Analysis-Scenanos representing alterna-the CCDF, by combining the probability of a given scc-tive futures fo'r the system and possible future states nario with the probability of a given set of input parame-of the environment are screened and chosen. Prob-ters for that scenario, abilities are estimated for the scenarios chosen.This step usually requires the synthesis of information Ilecause of the complexity of the calculation of the from many different disciplines in the natural sci-CCDF, the staff deemed it appropriate, but not abso-ences and engineering. lutely necessary, that the generation of the CCDF be performed by a computer code. 3. ConsequenceAnalysisalhe consequence in tet ms of As explained previously,only a rudimentary performance cumulative release of radionuclides to the accessible assessment was intended for the Phase 1 effort, because environment over a spectfied time period (usually of 1 mited data, resources, and time, and because aid from 10,000 or more years)is calculated for each scenario NRC contractors, which could contribute to the Phase I and usually numerous realizations of possible pa-goals, was not currently available, Because of the con-r meter des. straints on this activity, the scope of the effort was limited. ~' 4. Performance Measure calculation (CCDT)-The con-sequences for each scenario, in terms of normalued Only a prehminary analysis was intended in Phase 1, cumulative releases of radionuclides to the environ-Use of currently available modeling tools was to ' e ment over a specifico period of time, are c;dculated and the results are displayed in a curve of conse-maximiicd; additional computer code developmei 1 quences versus the probabihty that such conse. was to be minimized. = quences wdl be exceeded. Compliance with the per- .lhe analysts were to take advantage of the linuted formance criteria is determined by companng the data available for the Yucca Moun'tain Site. CCDF to a compliance curve, which the L CDF must The scopes of the cauyses were constrained by the not exceed. time and resources made available to do it; the effort 5. Semitivity and I/ncertainty Analysis-Sensitivity was scaled dow n from the original plan for this work, analysis investigates the change in performance As many components of the methodology as possible measures caused by incremental changes m the val-were to be executed, given the limited time and re-ues of input parameters and data. Uncertainty analy~ sources available; this required reducing the depth sis attempts to quantify the uncertainty in perform-to which certain aspects were demonstrated. ance estimates in terms of the major sources of For the Phase 1 effort, the FPA containment stan-uncertainty, including uncertainty in input parame-ters, uncertainty in modeling (both the conceptual dard was to be the major focus; other regulatory model of the geometry and characterization of the standards were considered only incidentally. system and the process model of what physico-chemical processes occur and how they are inani-Phase I was executed by NRC staff only; other than existing reports, papers, and computer software fested), and uncertainty about future states of na-ture, Mrxleling uncertitinty was not quantified in packages already delivered, no contractor help was avadaNe for Phase 1, except mfrequent and short Phase 1, personal communication. CNWR A involvement in Phase I was primanly as an 6. Documentation-The most ef fective documentation must make clear the assumptions used in the analy-observer, but was expected to become more actn e as NURFG-1327 uv

lixecutive Summary the CNWRA performance-assessment capability c. the consequences of C-14 gaseous releases, expands. and d. the statistical analysis of available hydrologic ES.4 ACCOMPLISilMENTS data for input to flow and transport models The NRC staff demonstrated its capability to conduct independently performance assessments for an 11LW re, ES.3 TENTAl,l\\,E RESULTS pository. Figure ES.2 shows how a CCDF for the total system can be constructed from curves for separate sce-in considering these tentative results, some important nario classes. Ghe caveats stated on the next page mdi-caveats should be recogr'ized. Taking these tentative re-cate why this CCDF is not considered to be representative sultsof a preliminary a iysis out of context.or sepamting of total system performance of a proposed Yucca hioun-these tentative results from these caveats, may Icad to the tam, repository.) In domg such assessments, the staff inappropriate interpretation and use of the results. gained insight into the capabilities and limitations of the currently available performance-assessment mett.odol-He results presented here have had limited pect 1. review, have numerous simplifying assumptions, and ogy, in achieving this primary objective, the NRC staff are based on limited data; therefore, the numerical also achieved the following major accomplishments dur-ing Phase 1: results should not be taken as representative of the performance of a repository at Yucca hiountain, Nevada. 1. hiodeled a potential liquid pathway of the undis-turbed scenario class for the Yucca hiountMn re-2. The analysis was replete with uncertainties regard. pository using: ing: a. the NEITRAN computer code, to simulate conceptual models, e transport in the unsaturated zone. e

data, physicochemical models, and e

b. four vertical transport legs under the reposi-models and data for predicting scenarios. tory to acceunt for spatial variability, 3. Only a limited set of scenario classes was incorpo-c. a modified treatment of waste form dissolution, rated in the modeling, so the total CCDF presented and in this report cannot truly represent total system performance. d. a nonmechanistic model of waste package fail-4. The modeling of waste-package failure was non-ure. mechanistic and rudimentary; therefore, this aspect This liquid pathway modeling was extended to treat of rep sitory performance is probably not ade-pluvial conditions, quately represented. 5. The liquid flow and transport models used at-

2. - Developed and used a total system code, to repre-tempted to simulate key amts of the performance sent total system performance, as a CCDF, for a of a repository at Yucca hbatain, but did so indi-limited set of scenano classes, usmg preliminary data and numerous assumptions.

rectly through modifications of transport analysis for saturated rock. A more direct representation of flow

3. - Developed a model and the corresponding com-and transport in partially saturated, fractured rock is needed to ensure more confidence in the results.

- puter code to treat human intrusbn by drillmg. Given the caveats just stated, the reader is reminded that 4. Performed a preliminary statistical analysis of re-the following tentative conclusions should tv; used only salts (sensitivity and uncertainty) using several tech-with t hese substantial limitations kept in mind. Based on a niques including bitin Hypercube Sampling (1. lIS) preliminary analysis, the staff has reached some tentative and regression analysis methods, conclusions: 5. Executed several auxiliary analyses, which kioked at: I.. The fact that the Yucca hiountain repository, like others, is designed so that the waste is emplaced over i a. the potential for non vertic:d flow, a substantial area, appears to be an important aspect determining performance, and should be included in b. the sampling requirements for CCDF genera-models of performance: important aspects appear to

tion, be areal variability of:

xv NtJRl!G-1327 l

lixecutive Summary i waste package fadure, DOli: other items could be pc'fonned by NitC, DOli, or e depth of rwk to water table, and 4hird pany.1hese suggestions are based on the work e potential of rock units to sustain fractsre flow. described in this report; they havs at been conclated e with other NitC stafI views or with th DOli site charac-2. 1he gaseous release of carbon.14 could be an impor. tentation p'.igrarn.1herefore, these sugestions ale not tant factor in repository performance, but mote intended to and should not be taken as indications of analyses'and data are needed to determine how im. deficiencies in the doi! Site Characterization Plan. ponant. 'these recommendations for technical improvements in-choc the following: 3. Two-dimensional mateling of the llYDitOCOIN Yucca Mountain description resulted in significant I:S.6.1 Iterommended imprmements to Modeling of lateral movement of water for unsaturated gro. Performance undwater infiltration rates greater than 0.2 rntn!yr. General Nonvertical flow couh W an important factor in repository performance, which warrants additional 1. Add the cap.,bility for rmxleling additional scenario analysis and data. classes. 4. For scenario classes at n ',t releases along the lig-2. Test the system code, using the consequence codes uid pathway, the most significant contributors to the as subroutines. instead of ge nerating data sets exter. consequences represented by the CCDF were iso-nal to the sy< tem code, topes of plutonium. As plutonium behaviorin poorly understood, large uncertaintici, exist regarding: 3. Acquire, test, and evaluate codes developed by San. dia National laboratories (SNL) for a repository in e

colloids, the unsaturated rone, retrograde solubility, and e

sensitivity of chemistry to oxidation state. 4. lixplore, with the CNWitA the adaptation of the e FPPA(Fast Probabilistic performance Assessment) 5. For releases niong the liquid pathway, the important methodology to pnerate the total system CCDF. input parameters appear to be: 5. livaluate adJitional wdes, which could not be ae-quired and evaluated du.ing this short time effort, e infiltration flux, to determine whether existmg codes can meet the fraction of infdtrating groundwater contacting NltC ruodeling needs or whether addnional code e the waste, development is needn uranium matrix solubility, and e saturated hydraulic conductivity for the Calico e llills Vitric unit. Flow and Transport 1. Itefine groundwater modelmg (e.g., by considering 6. Consequence codes used in this study may not be higher dimensions). sufficiently efficient to allow analyzing numerous scenarios, each with many input parameter vectors, 2. Incorporaa a model of gas pathway transport in the so that total system performance is adequately char. cidculation of the CCDF. acteri7ed. 3. Include flow and transport through the saturated '"" C - ES,6 PRELIMINARY SUGGESTIONS FOR FURTIIER WORl( 4. Directly model transport through a partially satu. rated, fractured rock, instead of the indirect, ap-Itased on this preliminary analysis and the limitations pnnimate representation used in phase 1. noted, the authors have some preliminary suggestions about the directions for further technic >d work to take. 5. lixplicitly model fracture / matrix coupling, ,these do not represent an official NitC position, but ne the view 3 of the individual staff members who wrote this Source Term report. Several of these suggestions relate te aspects of the methodology that are missing c.r need improvement 1. Attempt to develop or use a previously developed or that have not yet been incorporated into the NitC mechanistic model of waste-package failure. performance assessment capabiitty. Other suggestions relate to the general lack of data for Yucca Mountam. 2. Develop a mechanistic model of contact between Some of this suggested work is clearly Ihe responsibility of groundwater and the waste. NUlt!!O-1327 xvi ~ _ -

_ _ _ _ _ ~. _ _ _ _ _ _. t I xecutive Summary 3. Treat the repository as a usurce of radionuclides I;S.6.3 Herommendations for Additional Scientlinc distributed in time and space,instead of as a point input (Some of these items could be performed by either IX)ll or NHC, whereas others are clearly the responsibility of FS.6.2 Hecommended imprmements to and DOII) I.stensions of Ausillary Analyses 1. Develop and demonstrate a mathematically rigor-1. Perform detailed geochernical analyses to investi-ous, scientifically robust, method for scenario analy-gate: sis. the use of k 4"istribution coefficients), 2. Obtain geoscience data for imideling volcanism. d the effects of spatially varying saturation on 3. Obtain geoscience and hydrologie data for modeling radionuclide migration, faulting, uplif t, and subsidence at Yucca Mountain. the waste form, groundwater, tu!f interactions. the waste psLpc degradation, 4. Obtain laboratory chemical analysis to detcrmine the o.;idation of the i,gnt fuel matrix, and OC N # " "8"I"0""" # CU"""d""' # "P"* o ments of the spent fuel waste form. M c geochemical behavios of plutonium. i 5. Obtala field and laboratory data on phenomena im-2. livaluate heat effects at early time periods; estimate portant to the near field behavior of the repository, the thermal, hydrologie, and geochemical environ. especially the effects of heat. rnent of the repository at early times. 6. Obtain more data on plutonium geochemistry. 3. livaluate importance of thermally and barometri-7. Obtain a better understandingof waste pacLare cor-t cally driven air flow on repository performance at Yucca Mountain. rosion in the unsaturated rone. 8. Obtain field and laboratory data and perform analy. 4. Perform detailed hydrologic analysis for Yucca ses to investigate the issue of non vertical flow at Mountain, to provide better data to the transport Yucca Mountain, analysis and te examine, in more detail, various ul-ternative hypotheses regarding hydrology at Yucca 9. Obtain field data or, the transport of gaseous radio-Mountain. nue?Jes(C-14)at Yucca Mountain. xvii NUR11G-1327

D 2 C_ = l' f MINED GEOLOGIC REPOSITORY T SYSTEM DESCRIPTION a M WASTE SITE FORM SYSTEM .i ? 1 r 'A y 9 CONSEQUENCE 9 SCENARIO 7 ANALYSIS l g ANALYSIS 4, DESCRIPTION 2, 5

• SCREENING y

y E + PROBABILITIES i = E PERFORMANCE l 3 CALCULATION FLOW & TRANSPORT h

• CCDF

= E

• SYSTEM CODE a

8 I 5 5 seo SENSITIVITY & COMPARISON UNCERTAINTY TO REGULATORY ANALYSIS STANDARD

- i 5 TOTAL CCDF 5 (10,000 years)

c. a, > o E s';; 8 a eo e n=

I o r e $ s7 Total CCDF-EPA Standard j B v

3. e., o :r N

t an' \\ EEEEO

1 Pluvial / Drilling Case oggO

\\ fFEM ~~s

c. 3 E 2 y g ir 3

'10-2] Undisturbed / Drilling %3 r Case l S 5:[ r., m -5E39 'l I E&w o .t E E = I '_ s # g c-a

LRng, y 10_41 Demonstration P!at

s. g a =

2 5.

i 5 7 "-

n. see caption

c. m y 5 o

1 l r e w n, x -na w ww-n a o :::= 5 e G. - e. 61 1 s - c_ E? 82 i Undisturbed Case p # g 2. o ~ - - -. - - - - - - 8 5 n ". 1 i w cr-g r-c - [ l Pluvial Case - s5So T ~%s u-10-83 %-A a,a s ~ ,24 s s t sm \\ z n,o,-w =cc o %Acr 1 1 eom o I i 1 5359 ~ a 7 5 - c. t M g g g-g 1 0 -10 ...,.m, .m, .m, .m, p u- ? 1 0 -10 10-s 10-s 10-4 10-2 1 102 g I =0?.- o 2 e E" g' 3 y S T t 3 1E Summed Normalized EPA Release ? l-x am a a =oss a T' pomn a w o:=n q ~ f M L f

~- -. - - _. ~. --.- _-.. - lhecutive SummDry - I. - l l 7 ) i 1 7 I l r ? ? L >t-l I l l I' l-(- NUREG-1327 xx o l I e-r--W'-- m r e n-e ,-wx. w- . - +. -w,y -.6--. c -- -*vu ve.

• mda-n

e 9 e** m e me' r +-- - * - - e +w - rw--ww-w r---vi'wm-r e- -e--- -m-

ACKNOWLEDGMENT 'the work desenbed in this report was a team effort that J. Ilradbury, NhtSS T. hio, Nh1SS includm! input from the authon as wc!! as is om additional J. Tiapp, NMh5 K. Chang NhtSS staff members.1he authors would like to c.sprers their C Petemin. NMSS T. MarguHes IU.S collective appreciation to those NitC staff who contrib-uted to this report: N. Coleman, NMSS J. Pohle, NMSS 1 i. xxi NURiiti-1327

l i

1. INTRODUCTION l

This report describes the results of the Phase 1 demon-was continued by sorne staff for a time, sustained effori by i stration of the U.S Nuclear llegulatory Commission staff on this Phase 1 demonstration did 1,01 tesume until (NRC)capabihty to execute a performance assessment of August / September 1989. At that time, the effort we a high. level nuclear waste (lit.W) teposi.ory. restructured. 'the major features of this restructuring included: 'ihis dernonstration was performed as the initial step in a Concluding the Phase I work in 3 months, i.e., no sequence of planned iterative performance assessments to be undertaken by the NRC staff and NRC contractors. later than November 30,1964 Performance assessment of an llLW reposnory, like Attempting to execute as many steps as possible in other systematic safety assessment methodologies, bene-fits substantially by bemg conducted in an iterative man-the performance-assessment methodology, while at ~ ner, primanly because the lessons learned regarding the same time tailoring the activities to fit into the '1 modehng improvements, data needs, and methtdology time and resources allowed. crm be addressed in subsequent iterations. 'this activity I!stablishing a smaller core group of participants to was undertaken to maintain and to enhance the inde-pendent NRC staff capability to evaluate performance be respcmsible for the work. The involvement of assessments submitted as part of a license application. other staff and continual peer review, as originally envkioned in late 1968, would be deferred until al This capabihty consists of at least two aspects: (1) the capabihty to provide an independent check on key aspects ter November 30,1989, to expedite the effort. of the licensce's assessment and (2) the capability to 1)ividing the wmk into five parts: probe the licensee's assessment for potential weaknesses, based on a familiarity with the mcinods, data, and as-sumptions used in the assessm;nt. 1. Scenario Analysk 2. 1710w and Transport in addition, these iterative performance assessments are 3. Source Term expected to provide insights helpful in developing reFula' 4* System Code tory pnslucts, including: (1) technical positions, rulemak-5. Auxdian Analpes ings, and other regulatory guidance; (2) cvaluations of site characterization activities; and (3) evaluations of ihe Working groups, or teams, were set up to e mduct the NRC research program. Phase 1 analyses for the first four of these topical areas. e teams mugW mnesrxmd to the methodological Currently, two phases, of potentially several iterations, s@ of puformance assessment shown in I tgure 1.1. are planned for these iterative perfo nance assessment N auxWay an lpes wem mnducted by individual staff activities. Phase 1, a demonstration, was intended to: (1) tesult'in a framework for performance-assessment modeling;(2)with the limited resources alkicated to per- ~Ihis report is largely structut ed along the same lines used form this activity, provide a rudimentag demonstration of to organize the work. Sections 11S tbrough 3 provide in-a performance assessment modeling capability; and g g.lhese sections are: (3) be accomplished with a mimmum of techmeal input and interaction with NRC contractors, except for work O ~ h ""li*' *.**87 and pmducts already documented and delivered to NRC. Phase 2, is intended to:(1) be accomplished in l'Y 90 and ..l.0 - Introduction beyond; (2) incorporate significant products to be deliv-2.0 - Purpose and Scope cred by NitC ermtractors, most notably the Tuff Perform

• 3.0 - Organization and Staffing

. ance Assessment Methodology currently under develop-ment 'by Sandia National laboratories (SNL); and Sections 4 through 8 of this report describe the work (3) provide a more complete. accurate, sophisticated, and performed by the various teams: realistic performance-assessment modeling capability. Additional phases (iterations) may be added as this work 4.0 - System Code _ proceeds. 5.0 - Source Term 6.0 - I' low and Transport Models An interdisciplinary, integrated approach was envisioned when the initial plans for this activity were developed in 7.0 - Methodology for Scenario Development late 1988 to early 1989. Although a portion of the work 8.0 - Auxihary Analysis Summaries 1 NURiiG-1327

llecause Phase I was a demonstration of capability, these at Yucca Mountain are given or mtended to be given 'lhe sections may be taken as a status report on progress made authors did not encounter any proNems indicating that to date.'lhey should in no way be taken as the desenption Ihe U.S.17.nvironmental Protection Agency (l!PA) stan. of a definitive approach to these components of perform-dard (40 C1'It Part 191)could not be implemented. Ilow. ance assessment. cver, due to the incomplete scenario analysis in this dem-onstration, not all aspects of the standard were tested Section 9 " Analysis and 1(esults," presents the limited (e g., the difficulties in estimating scenario probabihties). results of this Phase 1 demonstration. Section 10 " Pre-liminary Suggestions for 1 urther Work," presents some preliminary thoughts on the direction for l'hase 2 efforts. 17inally, Appendices A through J provide more detaded liceause of the hmited nature of the analysis, no conclu-materialon the auuliary analyses and other aspects of the sions or recommendations about the proposed repository Phase 1 work. NUlt110-1327 2

MINED GEOLOGIC REPOSITORY SYSTEM DESCRIPTION ~ GINEERED WASTE SITE BARRIER FORM SYSTEM 3T g-i , r t ' r a

3 SCENARIO CONSEQUENCE 3'

g - ANALYSIS ANALYSIS = l DESCRIFTION l 4 E

• SCREENING SOURCE TERM E

. PROBABILITIES w M PERFORMANCE CALCULATION l I FLOW & TRANSPORT ^ o g

• CCDF i

N

• SYSTEM CODE n

ra i. s I 3 i-E COMPARISON SENSITIVITY & j i TO REGULATORY UNCERTAINTY + z t ANALYSIS 5 STANDARD 1 5a .L M ~

b R I I

2. PURPOSE AND SCOPE "Ihe primary purpose of Phase 1 of the iterative perform.

(2) conwderation of a range of conditions and events that ance assessment activity was to demonstrate the capabil-could affect Iuture r tformance. ity of the NitC staff to conduct, independently, a per-formance assessment of a proposed llLW repository. An The CCD1 was estirnated by followmg the steps outlined independent assessment capability is considered to be an below, which are shown m Ii ure 1.1. For the Phase 1 t important aspect of the licensing review to be conducted effort, all thcsc steps were performed, but some (for by the NRC staff. To achieve these goals, a limited. pre. eumple, Steps 2 anJ 3) were necuted only to a limited liminary total system performance assessment was con. degree, and only portions of others (for example, Step 5)

ducted, were done.

L Symm Dmriptionalhe repositog is broken intoits 'Ihe performance assessment was considered to be com_ prised of two parts: comp (ment parts for the purposes of modeling.

• 1hese components include the waste. the mined (1) the quantitative estimation of notal system perform.

repository, and the portion of the geosphere sur-ance through the use of predictive models, and roundmg the repository through which the radi-onuclides, m time, migrate. The system description (2) documentation, including detailed atailiary analy. must include informahon to support development of ses, where appropriate, to support the assumptions, models descnbmg repository performance and to data. and modeling approaches used to obtaiq quan. determine parameters upon which the models de-titative estimates of performance. pend. 2. Scenario Analms-Scenanos representing alterna-i lloth of these aspects were addressed in the Phase 1 tise futures for the sysicm and possible future states effort. of the cnvironment are screened and chosen. Prob-a m are ennaW for & cWn uns In accomplishing the primary goal of Phase 1, some worthwhile secondary goals were achieved: 3. Corocqucnce Analyds -Models are developed to de-scnbe the performance of the repository. 'the conse-

• the existing analytical too's to conduct a perform-quence, in terms of cumulative release of radio-ance assessment (both methodologics and computer nuclides to the accessible environment over a codes) were evaluated.

specified time period (usually 10.000 or more years), is e:dculated for each scenario and usually for nu-insight was obtained into the needs for the develojw merous realitauons of possible parameter values. In e ment or improvement of methodologies, addition to being incorporated by way of cumulative releases into the CCDF (Step 4). certain types of Insight into the needs for site characteruauon was conseqecoces miyht also be considered separately, obtained for comparisons to standards for maximum doses to individuals and for maximum concentration in The total system performance measure for an Hl.W re-groundwater. ('ihese were beyond the scope of pository can be expressed by a complementary cumulatn e Phase 1.) 1 or purposes of dividmg up the woik, the dMribution function (CCDF) of radionuclide releases to consequence analysis was conducted by the Source the accessible environment, weighted by a factor appro6 Terrn and the Flow and Transport Teams. mately proportional to radiotoxicity, and integrated ovs r an appropriate period of time (10,000 years is the curre it 4. Pciformance Mcasure Calculation (CCD/bThe con-regulatory requirement). 'Ihis performance measure is sequences for each scenario, in tenns of normalized mandated by the liPA standard (40 CMt Part 191) for 1 he cumulative releases of radionuclides to the environ-containment of waste by an Hl.W repository,'this per-ment over a specificJ period of time, are caletdatedi formance measure is incorporated into NitC's regulati m, and the results are displayed in a curve of conse-i 10 CFR Part 60, along with additional performance me ts-quences versus the probability that such conse-utes telating to: (1) waste-packape lifetime, (2) fractional quences might be c>ceeded. Compliance with the ~ release of mdionuclides from the engmeered barrier sys-performance enteria is determined by comparing tem (lillS), and (3) groundwater travel time. The repre-the curve to a comphance curve that provides limits sentation of repository performance by a CCDF of that the cedculated curve must not exceed. weighted cumulative releases incorporates: (1)considera-tion of the various components impeding the movement 5. Sensninty and Umcriainty Analyn,s-Sensitivity of radionuclides to the accessible environment; and analpa mvesugates the change in performance 5 NURl!O-1327 -, - - - - ~.-- - -. - -. -.

l

2. purpose and Scope measures caused by incremental changes in the val-ters, the magnitude of the parameters,or the shape of the l

ues ofinput parameters and data. Uncertainty analy-dntribution), depending on the scenanot sis attempts to quantify the uncertainty in perform-ance estimates in terms of the major sources of Ilecause of the compleuty involved m the CCDi calcula-uncertainty, including uncertainty in input param-tion. i was decided that the teneration of this curve would eters. in modeling(both the conceptual model of the be performed using a computer code. At a mimmum, such geometry and characterization of the system and the a code would be needed to: (1) sequence through all the process model of what physicochemical processes scenarios to be considered; (2) choose the consequence occur and how they are manifested), and in future models and parametric distributions correspondmg to the states of nature. Uncertainty in modeling, including scenario being analy/cd; (3) sample the parameter space conceptual rmdel uncertainty, was not quanttfied m appropriate to the gnen scenmo; (4) estimate conse-phase 1, quences based on the models and parameter values for the scenatio; and (5) combine the parametnc and scenario 6. Documentation-Documentation must make clear probabilities and the calculated consequences to gener-ate the CCDI-the assumptions used in the analysis, their bases, and the implications of their uses. Although the primary focus of the Phase 1 demonstration was the epa containment standard and the awociated Two types of uncertainty are usually treated explicitly in performance measure (cumulatis e releases to the acces+ the generatinn of the CCDF:(1) uncertainty due to future ble environment), some calculations of performance states of nature, and (2) uncertainty in the values of pa-

• neasures related to the NRC subetem requirements, rameters determining system performance. In a safety such as groundwater trasci time, fractional release rate, analysts for a more conventional type of system, the re-and waste package hfetirne w ere performed. These calcu+

sponse of the system to any single future state of nature to lations w cre performed to demonstrate the capabilitics of be considered would be a single valued estimate of system the performance-assessment methodology and ihe abihty performance (in the parlance of the repository system, a of the staff to esercise the rnethodology.'These calcula-single value of consequence), System performance would tions are intended as examples and shou 1J not be consid-then be described by the plot of consequences sersus the cred to be methods for calculatmg quantities in a regula-likelihood of the future state of nature (scenario)produc-tory context that the NRC staff wnsiders acceptable. ing that consequence; such a curve would be the distribu-tion function.The integral of such a curve over probabd-As explained in Section 1, only a rudimentary perform-ity would yield a cumulative distribution function; i c., the ance assessment was intended for this Phase I demonstra-likelihood that the consequence would be of a certain tion, because of limitcJ resontees and time, and because magnitude or less.The CCDF would be the curve of the input from NRC contractors that could wninbute to the likelihood that the consequene: would be more than a accomplishment of the goals was not avadabic. llecause certain magnitude. of these constraints, the scope of the offort was hnuted. Some of these limitations on the phase 1 effort were: For the repositor' system, considerable uncertainty exists concerning the values of parameters used to estimate the Only a preliminary an:dysis would be performed. o consequences of the repository. Traditionally, the uncer. Tbc effort would be scaled down f rom the original tainty from tius source t3 also displayed on the CCDF by: g . ;g gg (1) describing some or ali of the parameters used to esu-mate consequences as distributions of values rather than Only currently available modehng tools would be e pamt estimates;(2) choosing a value of each parameter used; computer code developrient would be mini-required to describe system perfortnance from these dis. mired. tributions representative of some portion of the vanous distributions; (3) estimating performance based on a The analysts would take advantage of the limited e given realization of parametric values; (4) noting the con-data available for 1he Yucca Mountain site. ditional parametric probability, i.e., the jotnt probability density for the given reali/ation or region of parameter As manycomponentsof the methodology as possible e space (for uncorrelated parameters, this would be the would be caecuted, gnen the limited time and re-product of the individual parameter probabilities); and sources available; this would require reducing the (5) calculating the CCDF, using the parametric probabil-depth to which certain aspects w cre demonstrated. ity multiplied by the probabihty of the scenario. This process is complicated further when consideration of dif. ~lhe !!PA containment standard would be the only ferent scenarios makes it necessary to vary: (1) the conse-performance standard considered. 'lhe !!PA stan-quence models for different scenarios, and/or (2) the dards for indiudual protection and groundwater distributions of parameters (either the range of parame-protection would be investigated at some later time, NUREG-1327 6

-. ~. - _. . - -.. ~. -. - -.

2. Purpose and Scope as would the subsystem requirements of 10 Cl~R tractor contribution would be via personal commu.

60.113. nication. To perform this preliminary performance assessment and .lhe demonstration would be executed by NRC staff, demonstrate the staff capability to conduct such work, the e only. 'the Center for Nuclear Waste Regulatory follow ng rypes of activities were performed:(1)computa-Analyses * (CNWRA) mvolvement would be pnmar-tions and support, including: data input, model setup, ily as an observer. c4de development and testing, code execution, and out-put analysis: (2) auxiliary analyses, including: evaluation e Other than existir.g reports, papers, and computer of assumptions and preprocessmg raw data: and (3)docu-software packages already delivered, the only con-mentation. i W 7 NURl!G-1327

3. ORGANIZATION AND STAFFING NRC staff members from both the Offices of Nuclear of technical staff to technical efforts in Phase I was done Materials Safety and Safeguards (NMSS) and Nuclear without regard to office affiliation.

Regulatory Research (RES) worked on Phase l. To coor- 'lhe project manager and technical coordinators facili. dinate the efforts of the two offices, the organizers of tated communications among the various technical par. Phase 1 designated an administrative project manager ticipants and managers. 'lhe technical coordinators also from NMSS and two office technical coordinators, one proposed plans for technical activities, schedulen, and from NMSS and one from RES. The technical staff in-staffing for Phase 1, for approval by NMSS and RES volved in phase 1 came from both offices /the assignment management. l t l t I i l l 9 NURl!G-1327 i .._.,r.... ,.4

i

4. SYSTEM CODE 4,1 Introduction 3.

To trcat uncertamty in future states of nature prop. erly, the system code must be able to treat diff erent scen nos (or mop properly scenario classes) that In the Phase 1 demonstration, a systern code was devel-attempt to desenbe those future states and obtain oped to process information needed to generate a CCDF the corresponding data on cumulative releasch of representative of the performance of an llLW repository, mdmnuchs. To obtain the CCDF, this code treated sequentially a set of scenarios that represented possible future states of 4. To treat properly the uncertainty related to the vari-nature. Consequence modules associated with the avail. able relcar,e pathways calculated the cumulative radion. ability in parameters used in the umsequence mod-uclide release, over the 10,000 year simulation t!me, for els, the system code rnust be able to collect and each of four scenarios analyred in Phase 1.These conse. process cumulative release data rencrated frorn quence modules were products of work puformed under multiple sets of parametne input vectors. the Source Term and the Flow and Transpon Tasks, that are documented in Sections 5,6, and 9 0f this report. liach 5. Hecause many scenarios are expected to allow radi-scenario yielded numerous cumulative release values, re-onuclide releases by mor e than one pathway (c,g.,in sulting from the multiple parametric input vectors used in groundwater, by gas, and/or by direct release), the a realization. Probabilities assigned to each consequence system code must be able to ohtaln cumulative re-within each scenario were then combined with the likell-leases corres1xmding to the specified pathways. hoods of the scenarios themselves, to form the CCDP. 6. The system code should have built-in protection to in au:ompthhing these tasks, the system code handled ensure the consisteney of the assumptions used two types of uncertainty inherent in a CCDF. First, it within e single simulation.1 or example, the per-treated the uncertainty in the future states of nature by fonunce time period (10,0Ml years for the cuirent 13PA standard) should be the sarne for all scenarios looking at sets of scenarios that attempted to describe those future states. Secondly, the code handled the un. and pathwap in any given representation of the certainty related to the variability in model parameters by repository to which the system code is applied. using multiple sets of parametric input vectors when exe-cuting the pathway consequence modules. 7. 'the system code should be able to present results in both tabular and graphical formats. 4.2 Requirements for the Development 4.3 Survey of Existing Codes of the System Code - The staff evaluated several computer codes to determine The development of the syrtem code is a continuing proc. their suitability (as a whole or in part) for use as a system ess. consistent with thc ongoing iterative performance as-program in this effort. Although all the surveyed codes sessment activity. Throughout its development, this code were not " system codes" per se, each was reviewed in should meet certain minimum requirements: terms of how well it fit the requirements expressed in Section 4.2. Appendix A provides a detailed discussion and description of the codes evaluated. 1. The computational modules for calculatmg conse-quences, comprised of one or more computer codes Based on the results of the review, the staff decided to for the source term and the flow and transport calcu-develop its own system code rather than to adopt an i lations, produce output in terms of cumulative radi-existing one. There were several reasons for this choice. onuclide release to the accessible enytronment.The First, adapting an existing program to meet the staff's system code must be capabic of receiving these data. needs and to be compatible with the NRC computing environment would likely be as time-consuming as devel. 2. The system code raust be able to treat two of the opment of a new cale. Secondly, an NRC-written code types of uncertainty incorporated in a CCDF charac-could be more closely tailored to the specific require-terizing repository performance:(1) the uncertainty ments and needs of the project than one developed out-in future states of nature, and (2) the uncertainty in side NRC. Finally, the more promising system codes for model parameters used to estimate cumulative re-potential use in this work would not be avmlable to the leases. staff within the timeframe set. 1I N Ulti!G-1327 -~

4. System Ode -

4,4 DeScripilott of the System Code ^ scenario's Pn'tubihty is estimaied by combining the probabdities of the processes and events making up the scenario. I or this demonstration, the staff nadeled four 4.4.1 Introduction scerano classes tused on two fundamental events: a plu-vial penod (or not) and drilling at the site (or not). Fig- 'Ihis section presents a brief dewription of the system ure 4.2 shows the protubilities awigned to the events and code developed by the staff for this demonstration.1he the scenarios. 'lhe probabdity of occurrence of drilling manner of code execution (i.e., internal vs. external). the was auumed to be independent of the protubihty of the input data requirements, the type of output available, and occurrence of pluvial conditions. a brief outline of the n$ tem program are presented. 1he liPA release limits were taken from 40 Cl R Part 191, i M etric ? ^J"N' I' """ "' '"'I'" '*I'"d P I'* ^PP'"@l ons of Ilcavy Metal (M 111M), these limits 4.4.2 internal vs. External Huns 'lhe system code can be executed in either the "internato converted in the system mJe to release limits for the in inntory of 70,000 MTitM assumed for this dem-or the "crternal" mode (Figure 4.1).1his distinction e-fers to the tirne at which the output files from the conse-onstration. FPA ratios were calculated using thev: limits, for each rcleased mdionuclide. quence modules are generated. In the internal mode, consequence modules are run and cumulative radion. .the cumulative rcleases of radionuclides were calculated uclide releases calculated as the system code is executed. Ihts requires that the modules be mtegrated closely with by the consequence modules, which modeled the reposi. the code. For external runs, however, the modules are tory release via the available pathways assigned to each separate from the system code, and as a result, the cumu. lative releases can be generated and placed in files at any time before iteration of the system code. Internal execu. 4A4 Operation tions would appear to make sensithaty analyses easier, because simulation parameters are global. 'lhus, changes .Ib obtain a CCDF estimating repository performance, to the input files for subsequent runs need be made only the system nide treated a set of scenarios desenbmg pos-sible future states of nature, and accessed the estimated once. This decreases the opportunity for error, while of-fering increased convenience and quality assurance to the cumulative releases corresponding to each scenano, the code next combined this data from the scenarios into the analyst. Simulations in the external mode offer the oppor-tunity to repeat earlier runs as long as the output files CCDI, and printed the CCDF as a graph and/or table, from the consequence modules are uniquely identifiable. .this section, along with I igures 4.3 and 4.4, pmvides a in addition, external runs would appear to be more eco-more detailed explanation of how the system code accom. nomicalin terms of both computer time and money, since plished these tasks. they do not require the execution of either the Iatin .lhe effects of each scenario were assessed in the follow-I crcube Sampling (LHS) routine or the consequence ing manner. First, had the s)$ tem code been set to run in the internal mode, consequence modules for the poten. tial release pathways specified for a scenario would have in Phase 1, although both modes of operation were al* been executed. Next. the cumulative releases calculated lowed, the system _ code was demonstrated only in the by the modules, either internally or externally, using external mode. l.IIS generat ed input vectors, were read into the program and stored in temporary arrays. 'lhen, each radionuclide-4,4.3 InEut I'"* P k *"S C"* P '"d '" i'S E P^ '^^* h *i ""d

• corresponding normalized I!PA ratio was calculated by

'The Phase I system code required input data in the fol-the following formula; lowing five areas: (1) general run information, (2) scenar-io specific information, (3) probabilities of those scenar-Omatized Release curuutauve Release of Radionuchde i los occurring, (4) EPA release limits for the initial of Radionuclide i EPA LJmit for Radionuchde i radionuclide inventory, and (5) cumulative radionuclide releases due to the effects of the scenarios. 'lhese normalized releases were then placed into a four-dimensional array arranged by scenario, radionuclide, The analyst created a file consisting of both the general vector, and release pathway (Figure 4.5) Once the effects ' run data and the scenario-specific information. This file of all scenarios had been modeled, this array was used as a supplied the execution mode, the simulation time perimi, data base over which different summation routines took and the amount of output desired, as well as the scenarios place. 'these routines created a second array of summed (total number, names, release pathways) to ctmsider, normalized HPA releases, ordered by scenario and vector, NUREG-1327 12 - ~. -. - - - - -. - -

4, Systern Code by addir.g up nortnaliied releases for all radionuclides factored in by multiplying the probability of each conse-over all release pathways. quence by the likehhood of its scenano. l'inally, the results from all scenarios considered were 'Ihen, for each H;enario, pnibabilities were calculated for combined, the summed nonnalized releases with their the coni,equences associated with a particular input vee-probabihties ordered and wrted, and a running sum of tor 'Ihese likelihoods were based on the assumption that the probabilities created. 'lhis outcome was graphed as a every vector within the scenario wns equally ptobabic. l'or CCl)l on a log log plot of summed normahied I PA exarnple, given this assumption, the likelihood of occur-release against cumulative probabihty. rence of a single vector within a scenario containing 500 vectors is equal to 1/$00 or.002. I ollowing the assign-4.4.5 Output ment of probabilities, the consequences within each scc-natio class were sorted, duplicates climinated, and the 1(esults generated by the system code were wntten to two likelihoods adjusted accordingly, output Itlet in addition to the data needed to graph the total CCl)lt, these files contained normalized teleases broken down by secnano, vector, release pathway, and 'the array for each scenario then contained unique or-radionuclide. or various combinations of these cat epones. dered consequences with associated likelihoods of occur-tence. lhen,in order to obtain a representative cumula-A commercially avadable software graphics package was tive distribution function, scenario _ probabilities were used to plot the CCl)l. l l i l. 13 NUltlGl327

3

3..,

j 3 9 SYSTEM CODE ~E 3 t CONTROL DATA o n (specify scenarios, g w I output desired, 1 r operation rWe ) I e 3 OUTPUT .E EPA RELEASE CENTRAL r LIMITS COMPUTATION PRINT & L PLOT 1 E CUMULATIVE Y RELEASES OF j ( 4 RADIONUCLIDES 9 FOR EACH - t E$

SCENARIO, 2

INPUT VECTOR, 3 AND PATHWAY I i E Q i 5, o l = F1 j c SYSTEM CODE 1 ~' CONTROL DATA { (specify scenarios, I E output deelred. 3 7 a '~ operation mode ) E CENTRAL l 2 EPA RELEASE OUTPUT e 5 LIMITS COMPUTATION PRINT & PLOT a ac. S 1 F J L t f f I l LHS CONSEQUENCE i coca i l I l i

._ - _ _ _ _ _ _ _ _ _. _ _ _ _ _ _ __ _. _. _.. _ _ _ _. _ _ _ __ _.q I

4. Syste n Ccde DETERMINATION OF SCENARIO PROBABILITIES FROM THE PROBABILITIES OF FUNDAMENTAL EVENTS P

P 0.9 0.1 Scenario Scenario class # 0 class # 1 D 2.3 x 10-7 Probability Probability = 2.0 x 10-7 = 2.3 x 10-s Scenario Scenario class # 2 class # 3 D ~ 1.0 Probability Probability ~ 0.9 ~ 0.1 P is not pluvial' D is no drilling P is pluvial D is drilling Scenario class # 0 is no drilling, not pluvial Scenario class # 1 is no drilling, with plucial Scenario class # 2 is drilling, not pluvial Scenario class # 3 is drilling and pluvial Note: Probability combinations assume that fundamental events have independent probabilities of occurrence; this is not a general restriction. Figure 4,2 Determination of scenario probabilities from the probabilities of fundamental events.This figure pre-sents results from an initial demonstration of staff capability to conduct a performance assessment. l The figure, like the demonstration. is limited by the use of many simplifying assumptions and sparse data. l 15 NURiiG-1327 2

e Z 5 OBTAIN DATA Y E Cumulative Release j 9 for each: y' C

• Scenerlo 0
• N3d*

SIMPLIFIED FLOW q.

• Pathway DIAGRAM OF
• vectnr SYSTEM CODE Form EPA Ratio, R, for each:

{ 7 3

• Vector P
• Scensno a

B ? Form (R, c. l p c. Probability) s pairs for each E Scenario 3 2, 3 l A Form (R, I a Probability) E pairs for all 2 Scenarios j combined OUTPUT Plot & Print f l

t i f i i Obtain Cumulative Normalize each Obtain EPA l l Releases for each Cumulative Release Limit Values { Scenario,1 by the EPA Limit for Cumulative l Radionuclide, J for that Release of each [ } Pathway, K Radionuclide, J Radionuclide, J j Vector, L v i t k E i h Sum Normalized Compare Arrays as Sum Normalized Cumulative Releases - Checic Attach to Each Cumulative Releases ^- O Over Nuclide J First Normalized Cumulative Over Pathway L First l f l Then Over Pathway L Release a Probability Then Over Nuclide J F - Calculate the EPA of 1/LTOT. ivhere - Calculate the EPA 4 Ratio, R, for each LTOT(Il is the Total Ratio, R, for each s dinput Vector, L, and. Number of Vectors Input Vector, L, for I Each Scenario, L for Scenario I each Scenario, t l G f 3 Eliminate Multiple Multiply Probability a Arrange Normal.ized. A Values of EPA Summed Cumulative Values of Each i a a-s;; Rat.io, Multiply.ing Scenan.o by the Releases.m th = Probability of the l l f Ascending Order ~ . p 9ba b the ScenanoI of Magn _itude c i n Number of Multiples t 1 .2 j ? I f r 8 Report Out - Plot Sum Probabiiities Assem e alues of and Print Values from Highest EPA Reorder (sort) and EPA Ratio, R and of the EPA Ratio Ratio Values to El.iminate Multiples + the Associated + Lowest to Obtain + I and the Cumulative. Probability of the Cumulative p g; j as Done Before for z of Exceeding any Probability of Each Scenano, I ingle ge Array j Value - THE CCDF Exceeding any Value r g w m Y y .t tJ k i

y f ? 3 a 3 1-O M a u DATA ARRAY FOR SYSTEM CODE INPUT 5 P RN J { f f f j / / / / / / / / / E RADIONUCLIDES / / / / / Array for p/ 3 cg RN 2/ / / / / Scenario # 0 a Y f f // / [ Vector 1 0.719 0.111 9.055 [ Vector 2 0.003 0.102 6.333 / 8 Vector 3 0.001 5.991 8.421 ^[ / g / / 5 k // F // /j/ m i Veetor K 0.223 0.00G 3.759 ? / a f 1 2 3 h PATHWAYS i

5. SOURCE TERM 5.1 Introduction Inespective of these measures, canisters may still fail.

Some of the mechanisms that might lead to failure are: 'Ihe demonstration of the performance. assessment Mechanical damage by excavation failure, earth-e r rnethodology depended in part on developing or adopting a source term model that could consider the rate of re-quakes, magmatic intrusions or human mtrusions Corrosion by hot steam or water dripping through lease of the radionuclides from the engineered barrier o system. 'Ihe staff reviewed several assessments of the fractures Yucca Mountain site performed for the U.S. Department Corrosion by direct contact of canister with rock; e of Energy (DOE) by national laboratories. Other source e4,Ims of air gap because of spaHation of nd m term models not developed for thy Yucca Mountain case were reviewed also. Appendix 11 provides a synopsis of inhlling by water borne sediment Certosion by immersion because of rise in regional these reviews, None of the models were completely satis-e factory, because important data on actual spent fuel un-water table or perched water table the expected repository conditions was not yet avail-g 'Ihe staff's model drew on features found in these assess-ments. In many cases,it was necessary to make simplifying Most of the spent fuel will be protected by thin cladding, assumptions /lhese assumptions were believed to lean on usually rirconium alloy, but in some cases stainless steel. the side of conservatism. In a small fraction of the cases, the cladding will be flawed by pinhole leaks or damaged (Van Konynenberg, et al., 1987). 'the cladding is an additional layer of corrosion 5.2 Review of important issues for resistance for the fuel, protecting the fuel from oxidation m water contact for a time. Since it is very thin (typically Selecting Source Term ModelS 0.6 mm) relative to the canister thickness, claddmg has 'Ihe radioactive waste, consisting mainly of spent light. usually been ignored in perfonnance. assessment studies. water reactor fuel, will be stored in metal canisters. A Aside from the potential corrosion protection offered by typical canister, according to current DOH plans, is about the cladding, the claddmg itself is likely to contam C-14, 4.8 meters long. 0.f 6 meters in diameter and has a wall produced by activation of impunti,s in the zirconium thickness of i cm (Site Characterization Plan (SCP). Sec-metal or picked up from the circult. ting water in the reac-tion 73.13) (DOH,1988). Small amounts of nuclear tor. Claddmg corroston thus might prove to be a source wastes in other forms, such as vitrified defense wastes, for the release of C-14 from the waste. Itcleases of gase-may also be stored in the repository, but the demonstra-ous C-14 are discussed in Appendix D. tion focused only on the spent fuel wastes. *lhe source term model must account for the processes in the near 5,2 3 Oxidation of Uranium Dioxide Matrix fic!d that determine the rate at which radionuclides are released, including corrosion and physical destruction of - Uranium dimide is unstable in an midizing environment the easte package, oxidation of the cladding and the spent (Grrmbow,1989). llecause the repository will be located fuel, gaseous releases, c(mtact betwern hqmd water and in unsaturated mek, there will be mygen available to the fuel, and transport of the released radionuclides be-oxidize the uranium dioxide after failure of the waste yond the confines of the engmected borrier. package and cladding. Before failure, the canisters will most likely be filled with en inett gas to prevent oxidation, 5.2.1 Waste Package Lifetime ahhough it is possible to have oxidation directly from water that might be contained in the fuel nxts, particu-The canisters will be scaled and most probably filled with larly those fuel mds that have already failed. 'the rate of an inert gas.~Ihey must first be breached before there can midation depends, among other things, on temperature, he any release of radionuclides. Several measures will be so the time that the waste package fails might be impor. l used to reduce the likelihood of (_unister breaching: tant. Oxidation of the uranium dioxide is potentially (1) the unisters will be made of corrosion-resistant mate-important to the performance model, because uranium in rial; (2) there will be c.n air gap between the canister :md higher valence states is much more solubic than in low toe. host rock to prevent any direct contact with pore valence states If the fuel is immersed in water, the rate of water; and (3) the decay heat may crute a dry zone for oxidation may be the limiting rate for congruent several hundreds of years after emplacement, further dissolution of the fuel matrix (Doctor, et al.,1992). In l-isolating the cimisters from contact with liquid water. addition, midation >f tl : fuel under dry or moist steam l 19 NUltlE1327

5. SourceTerm conditions can cause an increase in its volume and poros-into contact with hquid water, but at the same time, di-ity, with the consequence that the case at which the gase-recting water away frorn other waste packages.

ous radionuclides such as C-14 could be released might increase. The significance of the issue of thermally 4! riven water circulation is difficult to determine at this time. If all neat 5.2,4 Release of Dissolved Radionuclides 8*"",*'d W'

• I##' * "" *#"' I"* #""P"'"'h"M' from the Fuel water the flux would far exceed the likely infiltration rate,11 may be the case that these phenomena are short-lived and unimportant during the period of canister Initially, the canisters and the spent fuel are likely to produce sufficient heat to dry out their surroundings or integrity, when most of the water dnven off would be

- create a dry steam environment. Ilventually however, diverted from the canisters rather than returning. Of liquid water might come into contact with the spent fuel, course, the relationships between heat production, tdlowing it to dissolvc and release its inventory of radi-evaporation, and circulation are far from simp!c, and onuclides to the environment. Most of the inventory of must be approached with sophisticated modeling tools, radionuclides will be entrapped by the uranium dioxide Models such as 'lOUGli (Preuu.1987) would be re-matrix of the iuel. and will be released slowly as the matrix quired to carry these arguments iurther.,they wcre be-disintegrates. Some of the radionuclides released frorn yond the scope of the Phase i study. but should be the matrix might precipitate immediately because of their planned for subsequent studies. low solubility, thereby limiting their release (Ogat d. et al., 1983), or may form colloids (Ihompson,1989). Some of 'lhe water contact fraction was characterued by a ratio the more-volatile radionuclides such as C-14, ecsium, and relating the amount of water infiltrating the site to that iodine tend to migmte from the rr. atrix and collect at coming into contact with the waste canisters. Sirnple cal-mtergranular boundaries and in the gap between the fuel culations were performed to estimate the fraction of the and the cladding, particularly while still in the reactor. waste canisters exposed to purely vertical infiltration by These volatile radionuclides will be released more qmckly taking the ratio of Ihe cross-sectional area of t he camsters than those released by congruent dissolution. to the total area of land surface projected by the re;wi-topimtio wheulatN to be approximately 0 0011. 'this simple figure does not capture the true nature of 5.2.4.1 Water contact frartion water contact, because, in its uninterrupted state, infal-DOE plant, to emplace the canisters in the host rock in a trating water is !3 cly to flow around the canisters because manner that it considers can reduce the likelihood of of thematrixsuction of the unsaturated:ock.Theanalysis in Doll's emironmental assessment of Yucca Mountain water coming into contact with the waste (SCP, Section (DOI:,1986) assumed a contact fraction of 0.025, but no 8.3.5.9) (DOII,1988). A proposed emplacement, plan basis was specified for this choice. Other analyses have would have the camsters stored vertically with an att gap specified that all water infiltratmp the site contacts the between the canister and the rock walls. l'urthermore, DOli considers that the heat generated by the waste may wast e (Doctor, et al.,1992). DOli design poals specify that create a sigmficant zone af dry rock around the cantsters. for the first 300 years following closure,95 percent of the isolating them until such time that the water can infiltrate canisters should be essentially dry, and the remaining 5 the rock again. Water might still come into contact with percent have contact with less than 5 liters of water per - the canisters by several mechanisms: year. From 300 to 1000 years after closure, up to 10 percent of the canisters can have 5 liters per year contact (SCP, Section 8.3.5.9)(DOE,1988). Section 8.3.5.10 of o Circulating water generated by the decay heat the SCP allows contact of less than 20 liters of water per i 4 0= Infiltrating water flowing through fractures'and year per canister for up to 10 percent of the canisters, dripping onto the canisters however. His figure was estimated as 80 times the ex-pected maximum flux for canisters eruplaced vertically, less of the air gap caused by failure of the emplace-o , ment holes through mechanical and thermal 5.2.5 Release of Gaseous Radionuclides stresses, or mineral and sediment infilling There are several gaseous radionuclides in spent fuel. Here are other possible sources of water available to the although many of these are short lived and of no 10: g-fuel, other than vertically infiltrating precipitation, but term concern. The most significant radionuclides are these were not included explicitly in the Phase 1 calcula-carbon 14 and possibly iodine-129 (only at elevated tm-tions.71%o potentially important sources are: (1) lateral peratures). C-14 would be released from the cladding, inflows from areas of perched water and (2) liquid water the cladding-fuel gap, and the matrix. The gaseous re-circulation caused by heat-driven evaporation and con-leases would be partitioned hetween the groundwater and densation. Lateral infiltratioa might givert infiltrating air, depending on environmental factors such as satura-groundwater, causing some of the waste packages to come tion, temperature, and concentration of bicm bonate ions. NURIIG-13d - 20 ~

5. Source Term None of the models reviewed in Appendix 11 handles the Theleachrate AL was determined liy the combmation of releases of C-14 iri a very sophisticated way,'the models the infiltration rate (1), the fraction of water contacting cither treat the C-14 as a component of the fuel released the waste (f). the surface atca of the repwitory (A), the to the groundwater by congruent dissolution of the fuel solubihty of the waste form (Su), and the initialinventory matrix, or all is released instantaneously upm failure of of the waste form (Mo):

the waste canister. AL = 1 x f x A x Su/Mo. (5.2) 'the release of C-14 from the repository is of interest to disposal in unsaturated sock because there is at least the if the solubility limit would be exceeded by the release possibility of a fast pathway to the accessible erwironment calculated byliquation 5.1,i.e.,if R (t) > S lAf,then the i i through fractures, excavations, and tunneb. Two models release rate was cut off at the solubility limit: of transport of C-14 in the geosphere of Yucca Mountain R (t) = S lAf. (5.3) indicate that the time for C-14 released at the repository i i level to reach the atmosphere would be on the order of where Si - the solubihty of radionuclide i. hundreds to a few thousand years, too short a time to depend on decay to diminish the importance of C-14 'the release rate R;(t) became a flux luundary condition cumulative releases to the accessible environment to the transport equation. (Knapp,1987; Amter, et al.,1988). An assumption of instantaneous release from a failed canister may too pes-53.2 Limilations of Model for Dissobed simistic. On the other hand, the assumption that all( -14 Radionuclides is contamed in the matrix and released only as the matrix dissolves may be too optimistic, because a substantial '1he most significant hmitations of the dissolved radio-fraction of the C-14 may be contained in places other nuclide source term model are believed to be: than the rnatrix, e.g., the cladding.1 atmratory data on the 'lhe model ignores the diffusion. limited case, where location of various radionuclides in spent fuel under dif-e ferent conditions wdl reduce this modeling uncertainty. there might be the buildup of a loundary layerlimit-ing the release of solubihty limited radionuclides (this mechanism would apply only if there were a 5.3 Model Selection and Justification wntinus Hquid witter path between the fuel and the rock). 53.1 Model for Dissolved Radionuclides For I rger infiltration rates, the model cannot tepre-e sent phenomena that would tend to limit the rate of 'The source term model provides cdculations of radion-9"" " I" U *' uclide releases to the flow and transport calculations. For this study, the source term model incorporated in the Network Flow and Transport (N1!!TRAN) computer code (longsine, et t .17) was adopted. Radionuclide The model assumes intimate contact between the releases would occm u...y after failure of the waste pack-age, characterized as a single failure time tr (it was recop-Froundwater and the waste, ignoring features such nized that waste-canister f ailure would probably be dis-as the air gap, designed to prevent such contact. It in tributed in time and space, but the Niit TR AN model was effect assumes there is no protection for the fuel incapable of dealing explicitly with the source term in this from the water, even though the fuel has multiple manner). layers of protection, including the air Fap, vaste package, and claddmg. Upm failure of the enginected barrier at time it, radio. C The model incorporates a single time to failure, even nuclide release would be governed by either the teaching though it is more likely that waste packages would rate determined by the rate of dissolution of the waste fail in a distributed manner in time and space. form, or limited by the solubility of the indwidual radi-Releases from the matrix of low-solubility radi; onuclides, S. For the former, the rate of release would t onuclides might result in colloid formation rather be: than precipitation. The model does not take into account radionuclides Rj(t): = At Mg(t), (5.1) that might not fit neatly into the three compart-ments (unleached, undissolved, and dissolved), such where M - theinventoryat time t of the radionuclide m as t hose collecting in the grain bour.daries and in the the waste, cladding gap. 21 NURiiG-1327 - _ _ ~._ _ -,._ _ _ _. _.. - _. _ _, _ _ _, _ _ _ _ _

5. Source Tenn The model ignores the guentially significant spent fuel indicate a rapid, small telease of t'-14 upon amount ofliquid water circulation through evapora-failure of the fuel nd, and very slow release therealter tion and condenution of groundwater that might be (Van Konynenberg, et al.,1984). llecause of the specula.

caused by the repository heat, i.e., a " heat pipe." tive nature of the C-14 release model, gaseous release pathways were not included into the overall systems

• lhe use of this moJet was based on expedieng, because analysts, but wer e treated separately as an auxiliary analy-the fundamental framework was already in place in the sis (see Appendix D).

NEITRAN code and reouired a minimum cf reprogram-ming to adjust the coefficie*2 to represent the Yucca 5.4 Source,I,crm inventory Mountain case. Adjustmeni Vf the coefficients of the model allowed a wide latitude of potential source tenn 'lhe irnentory of radionuclides assumed for the source conditions to characterize either congruent dissolution of term in the Phase 1 study was typical of previ,us analyses the uranium matrix or solubility limited releases. f the performance of repository at Yucca Mountain and is given in Table 5.1 (Doctor, et al.,1992). 'the list was 5.3.3 C-14 Release Model restricted to 29 isotopes, chosen from a more extensive list of fission and activation products found in sp:nt fuel, Very little is known about the long-term release of gase-on the basis of half lives, potential inventories and ous radionuclides from spent fuel under conditions antici-radiotoxicity (in terms of their l!PA (omulative release pated at Yucca Mountain.'lhe only data on releases f rom limits). Table 5.1 Itadionuclide Initial Imentory (Doctor, et al.,1992) Itadionuclide llatf. life,3 rs inve ntory, Cl Curium-246 5.501 03 2.451iO3 Cunum-245 9.30E03 1.261104 Americium-243 7.95E03 9.80l!05 Americium-241 4.581iO2 1.121:0S Plutoniurn-242 3.79E05 1.12E05 Plutonium-241 1.321i01 4.831:09 Plutonium-240 6.5SE03 3.15F07 Plutonium-239 2.44 E04 2.03E07 Plutonium-238 8.60E01 1AGEO8 Neptunium-237 2.14 E06 2.171iO4 !Iranium-238 4.51!!O9 2.24E04 Uranium-236 2.39E07 154E04 Uranium-235 7.101:08 1.121!03 Uranium-234 2.47E05 5.18E03 Uranium-233 1.62 E05 2.66E00 'Ihorium-229 7.34E03 1.961bO3 Thorium-230 8.00E04 2.8711 01 Radium-226 1.60E03 5.lC E.0 t

1. cad-210 2.23E01 4.901605 Cesium-137 3.00E01 5.25E00 Cesium--135 3.00E06

$.89104 hdine-129 1.59E07 c..N11:05 Tin-126 1.00liO5 3 3 F.34 Technettum-99 2.15E05 9 MD5 Zirconium-93 9.50E05 1.19;:05 Strontium-90 2.90E01 MiEDO Nickel-59 8.00l!O4 2.10E03 Carboel4 5.731 03 9.80E04 NURiiG-1327 22

6. FLOW AND TRANSPORT MODELS 6.1 Introduction eastward tilted volcame rockt s' opes are locally stecr(15 to 30 degrees)on the w est facmg side of Yucca hiountam

'lhe quanttitcation of the consequences of 1II.W disposd and along some of the valleys that cut into the more gently is anticipated to require the analysis of groundwater flow slopmg (5 to 10 degrees) east side of the mountam. and transport of radionuclides in liquid, gas, and direct release pathways. 'these analyses typically will be based l~or this study, the hydrostratigraphie units of interest at on site conceptual models that are then implemented in Yucca hiountain were comprised primarily of ash flow computer programs for calculational use in a perform. and ash fall tuffs that originated from croptions during ance assessment.1 or this study, a review of information the development of calderas. The amount i f weldmg, on unsatur..cd fractured tuff and transport pathway fractunng, unit thickness, and chemical alteration varies phenomena, and flow and transport computer programr. greatly Imm one layer to the next. 'Ihe major was conducted O select computer programs to provide hydrostratigraphic urnts beneath Yucca hioumain, stan-calculational tools with which to demonstrate the ing at the surface and moving downward are: alluvium. performance-assessment capability."Ihe purpose of this Tiva Canyon welded unit, paintbrush nonwelded umt, section is to desenbe the information that was collected Topopah Spongs welded unit. Calico liills (vitric and and used to select the programs for quantifying conse. icolitic) nonwclded unit, Crater llat welded and non-quences. welded unit (prow pass member and llullfrog member) (see 1 spute 6.1). It should be pointed out that the defmition of the site conceptual model(s) will typically be based on detailed ~lhree broad categories that desenbe these tuffs are: laboratory and field investigations for the site under con. densely welded tulfs, nonwelded vitne tuffs, and non-sideration. 'lhis site conceptual rnodel(s) will undoubt. welded icohtired tuffs. The densely welded tuffs are edly be the most important factor in selecting a computer highly fractured "Ihese tuffs have a very low saturated program for site analyses. liowever, a well charactented matnx conductivity (less than 1 mm/) car)and a saturated llLW site does not currently exist. As described above, conductivity for the fractures that is probably several or-the model selection, for this study, was based on a resic w dets of magnitude or more higher than the matrix value. of published information.'the authors do not c( nsider the 'the nonwelded vitric tuffs. have fewer fractures and a review comprehensive and do not mte"d the model selee. higher matrn saturated conductivity (100-10,000 mm/ tion to represent an endorsement of any particular con. year). 'the fr actures for this unit would have a relatively ceptual model(s)foi the Yucca htountain site or a recom. Iow saturated conductivity.1he nonwelded reohtized mended approach to modeling flow and transport in tuffs have few fractures and low matnx saturated conduc-unsaturated fractured tuff, tivity (less tSan 1 mm/ year) anJ low fracture saturated ~ conductivity. The contacts between these tmits rencrally tend to occur over short distances and involve large dif fer-6.2 Definition of issues for Selecting ences in hydrologie emperties (prindle,1987). Performance-Assessment Based on current information on hydrogeologic units and I,ranSport Models theories d Dow at Yucca hinuntain, DOli (from page ' C""Y The definition of the technical issues for defining and "'"E"P" ""U "*I selecting flow and transport models was based on the characteristics of an unsaturated fractured tuff medium and the pathways anticipated to be analyzed. This infor-j, ..Dow m the TSw unit is expected to be essentially mation was obtained from published reports concerning vertical and under steady-state conditions to occur as performance assessment m geologic media and ground-flaw withm the matrix for fluxes less than some criti-water flow and transport in the geosphere, with an em-ml value of flux related to the saturated matnx hy-phasis on unsaturated fractured media. dntulic conductivity, and predominantly as fracture flow at fluxes higher than the entical value. 6.2.1 Site Concepts 2. I ateral flow may be induced in the TSw unit at its 'lhe Yucca hiountain site is located on and irnmediately contact with the underlymg Calico Ilills nonwelded adjacent to the southwestern portion of the Nevada Test unit (Cl ink 1he cir cumstances under which this may Site. Yucca Mountain is a prominent group of north-occur depend on the magnitude of t!.e flux in the trending, fault-block ridges. The terrain at the site is TSw unit and whether this umt is underlain by the largely controlled by high-angle normal faults and low-conductivity /cohtic facies (Clini) or the 23 NUllIli-1327

. ~. _ _ _

6. Irlow and Translut Models fQ6n/s.

,Ct CAL ~~ _ a

"

.ii;,;;... ;;;;;........

.

.

.

• .

  .:.:.. "!.:, ::,:,:d,*1"n"::

15* ;

;!;;;;;;;;;;;

Q

: I!...

m . s s s s s s g s s s"s"% % % \\..... ;.;.;.;.;.;.;.;.;;.;.;.;:.;.;! *:;; "::j,::::::::::::::: 'ssssssss\\\\ sggssss

ggggggss\\s\\\\\\\\\\\\s\\\\\\\\\\\\\\\\\\\\\\\\\\\\'NN\\s\\'s%\\ m\\\\\\\\' sss"

:::::::::::::::::

\\\\\\\\ \\ NNggs NNN\\sCM\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\' .s g s\\ \\ \\ g g gsssss\\NNN\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ \\\\ \\s ssssggggg ssgsssss s sss\\s\\NN\\\\\\NN\\\\\\\\\\\\\\\\\\\\\\\\ s\\NN\\gsgggggg,sss sg g g g g g g ggs \\\\ \\ \\ \\.' \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ggg s s sss sss sssss s, ssss ,\\\\\\\\\\NNggggggg ss \\\\\\\\\\\\\\\\\\\\\\\\\\\\ g o ss's'N' gg \\

• ss**

s s s s s,s \\>\\ \\ \\ \\ \\ \\\\ \\\\\\\\\\\\\\;' M \\ / >>>>>\\>>>>>>>\\ \\ v" _ f /-A,. ' u- ,p- % gATEB 1 WEST EAST UNSATURATED ZONE HYOROGEOLOGIC UNITS N . d '. '.. OAL ALLUVIUM a i,;/, TCw T!V A C ANYON WELDED UNIT X-C9 PTn PAINTDRUSH NONWELDED UNIT C www TSw TOPOP AH SPRING ' WELDED UNIT mus M s\\\\ W CHo CAUCO HILLS NONWELDED UNIT M CFu CRATER FL AT UNDIFFERENTIATED UNIT NORM AL F AULT 7 UNIT UNCERTAIN i igure 6.1 Conceptuahration of a hydrogeolope crowsection through the unwiturated ione at Yucca Mountain (after DOli,1958) NUlmO-1327 24 . ~,

6. Ilow and Transport Models relatively higher-cond uct hity vit ric facie; (Clinv) of infdtration.'the current review indicated that the effects the Clin unit. At low Duxes within the 'l$w unit, of fractures on groundwater flow and of flow diversion at i

lateral flow may be pnMuced by capillary barrier lay er tvundaries will need to be assessed and their senu-i effects within the matrix of the 'lSw unit where it tivity to infiltration rates determined. Ilowever, for the overlies the Clinv unit. At high flutes, efficient frac-present study, it was assumed that groundwater flow tur e flow in the *lSw unit may prodt.cc lateral fkiw as would be one dimensional and in the vertical direction. L well as vertical flow where the low c(mductivity Clin I unit underlies the'ISw unit. 'Ihe role of fractures and flow diversion at unit bounda-ries could have significant affects on flux rates through a 1 3. Flow in both the Clinv and Clinz units is predorni-repository. Although flow diveruon was the subject of a i nantly vertical through the matrix (although a lateral limited auxiliary analysis (see Appendix 0), future analy. component may occur parallel to the bedding within ses will need to consider fracture-matrn interactions and j the vitric Clinv unit) and continues directly to the further consider flow diversion where fractures c:m affect water table wherever the latter transects the Clin the flow. unit. Where the Clin umt lies above the water table, flow is presumed to proceed vertically downward to 6.2.2 Pathways i the water table through the Crater 11at undifferenti-ated unit (Cl~u). 'lhe assessment of a repository in the unsaturated zone i could involve the following three pathways: (1) liquid, 4. 'Ihe nearly vertically oriented fault zones and their (2) gas, and (3) direct.'Ihe most obvious release path for associated fracturing may be highly effecthc path-radionuclides away from the repository is thc liquid path-ways for vertical moisture flow, especially in the way. For the present study, it was assumed that radio-competent TCw and TSw units. Ilut faults may im-nuclides would be transported vertically in the unsatu-pede lateral flow and may thus produce perched-rated zone toward the water table, and re! cases were water txxiics w here the faults transect zones or hori-calculated at the water table. zons of significant lateral Dow." The gas pathway is a potential concern for a repository l Additiomily, very httle data are available on estimated located in the unsaturated zone because of the presence infiltration rates and deep percolation rates past the re-of carbon-14. It is present in the emplaced waste in quan. pository Estimates of deep percolation rates past the tities at least one order of magnitude greater than the repository horizon are described on page 3-205 of the release litre spccified in Appendix A of the liPA stan-SCP (DOE,1988) as: dard. It can exist as one of several gasses (CO, methane, 2 - acetylene), and could therefore move relatively rapidly Wilson (1985) reviewed available site and re-comparcd to its half life (5720 years) through the unsatu-gional hydrogeologie data in order to set con-rated fractured rock and along pathways such as access servative upper limits on the present, net verti-tunnels and excavations. In addition, unlike most of the cally downward moisture flux below the other radionuclides in the waste, transport in the geo-repository horizon at Yucca Mountain and on sphere is not likely to depend strongly on the influx of the present rate of nel recharge to the satu-water to the repository, and can proceed under totally dry rated zone in the vicinity of Yucca Mountain. conditions. (llowever, release from the waste may de-Wilson (1985) concludes (1) that the liquid-pend on the water influx.) -water percolation flux, directed + ally ' downward in the matrix of the 'ISv, ,ow Finally a release pathway could occur as a result of a the repository horizon, probably b 6than0.2 " direct" release. The " direct" release path *vay encom-mm/yr and (2) that the area ave %cd rate of passes a couple of possible scenario types such as a re-net recharge to the saturated zone in thaicin-lease due to drilling into the repository and a release due ity of Yucca Mountain probably is less than 0.5 to a disruptive event like a magmatic eruption.The con-mm/yr. Although Wilson (1985) considered a' sideration of the consequences due to volmnic activity number of processes, such as upward water. was too involved to be included in the current study; vapor flow in the fractures of the 'ISw tmit at therefore, the direct release pathway considered only re- . the repository horizon, these upper bounds on leases due to drilling. Releases resulting from volcanism percolation and recharge fluxes must be re. - will need to be addressed in future work. garded as preliminary estimates that have as-yet unknown limits of uncertainty. 6.2.3 Flow and Transport Pathway i Phenomena It was considered that the definition of a conceptual model for flow ar'd transport in unsaturated fractured tul'f Performance assessment of potential releases on.adioac-depended on fracture matrix interactions and the rate of tivity from nuclear waste requires an understanding of a 25 NUI 11G-1327

6. Flow and Transport Models number of complicated transport phenomena for the that under high suction the dominant groundwater path-pathways under consideration. He transport pathways to way will be in the matrix (i.e., the fractures will be dry).

be analyzed are the liquid pathway, the gas pathway (pri-However, it is worth pointing out that many factors (la n-marily involving the tmnsport of carbon-14), and a direct sient infiltration rates, fracture coatings, fracture dimus - release pathway (due to a drilling scenario). His section sions, and the presence of perched water) can dramati- ' describes, in a preliminary way, some phenomena associ-cally influence the degree of fracture flow and validity of a ated with the transport of radionuclides in groundwater single continuum model for unsaturated fractured media. - and the phenomena considered in this study, Many assumptions that preclude fracture flow under un-saturated conditions have not been substantiated by labo-6.2.3.1 Liquid transport ratory or field data and, therefore, cannot be ruled out as a possible transport pathway in unsaturated, fractured A common starting point in the' development of a trans-rocks (Parsons, et al.,1990). port model is a qualitative statement of the conservation of mass in the liquid phase for an elemental volume Based on the lack of information to support a detailed - (Freeze,19 /9): fracture flow model, a steady-state flow model was as-sumed, where the fractures contribute to flow only when cNgfor - - soYute out - s(Nmass the infiltration rate exceeds the saturated conductivity. ut into + ma within ofthe the . due to reac-Further work will need to determine 0,e degree of con-the element element _ element tions and sinks servatism or pessimism in this assumption. and sources. Whereas advection moves solute in the direction ofibw - The processes that control flux into and out of the ele-hydrodynamic dispersion and matrix diffusion affect sol- - mental volume te advection (transport via the bulk mo-ute concentrawn along b hw pah HyMynam& tion of the groundwater) and hydrodynamic dispcrsion persion includes dilution due to mechanical mr,xing and - (transport resulting from mechanical mixing and molecu-molecular diffusion. Mechanical miving (a direct result of lar diffusion). Chemical reactions and radioactive decay - a tortuous path, variation in pore sizes or fracture aper-wdl affect the loss or gain of solute mass (for the present tures, and surface roughness)is idated to the heteroge-analysts, phenomena such as Knudsen diffusion and cou-neity of the geologic media and is typically characterized - pled proces:es were considered of mmor importance). by the dispersivity. "The transformation of the aforementioned qualitative statement into differential equation (~s) typically involves a For the present analysis, it was assumed that dispersivity number of simplifying assumptionswith respect to dimen. can be represented with a single dispersion length. nts - sionality, variability, and processes associated with the treatment was assumed adequate, because the perform- = intended application.This section will review some of the ance measure of interest (cumulative release at the acces- - ptucesses associated with the pathways to be considered sible environment over 10,000 years) would generally be in this study. insensitive to longitudinal dispersion, when the cumula-tive releases include a majority of the_ waste, and small cumulative releases are n t as important as large re-(a) ' Physical Processes leases. Ihe degree to which this was or was not a conserv-It is assumed genera h that the bulk movement of fluid ative assumption will need to be examined in further will be the primary source of transport away rom an work. ~ f l_ HLW repository. In a porous medium, it is assumed :om-monly that the average rate of solute transpoti by advec. Matrix diffusion couples the solute concentration in the - tion'is equal to the average linear velocity of the fluid fracture and matrix systems and is generally thought to - - multiplied by the concerMtion.The presence of a frac- . provide a retardation of radionuclide transport in the i ture system complicates t,e advective flow system by pro-fractures. As with the flow of water across the fracture-p viding a high permeability flow path separate from, but matrix interface, a large uncertainty in evaluating this _ interacting with, the matrix path. phenomenon is determining the effect of fracture coat. ings on the diffusion rate. Quantification of the effect of p fracture coatings will be needed to better determine the in the unsaturated zone, water is held in the pore space by surface tension. Geologic media are comprised of a vari-best approach foi performance assessment. For the pre-ety of pore space and fracture dimensions: therefore, sent study, matrix diffusion was assumed not to occur, these volumes will not empty at the same suction. During This assumption should be conservative for the situation drainage, the large pores (or larger fractures) will empty when contaminant being transportst in the fractures is at low suctions, whereas small pores (smaller radius of diffusing into the matrix, lh wever, this assumption may curvature) will empty at higher suctions. Most models of not be conservative when contaminant is diffusing from p _ unsaturated flow in fractured media, therefore, assume the matrix into the fractures. 1 NUREG-1327 26 b ,.-n

f

6. How and Transpott Models (b); Chemical Processes more, gas-phase source terms (i.e., carbon-14, tritium, krypton-85, and iodine-lF

'ild potentially be re- %cre are Teveral chemical processes that affect the leased f ; 1 spent fuel but Ncca Mountain. Gas-movement of radionuclides in groundwater. One of the phase caroon-14 in the fon o ' m dioxide appears to - most significant chemical processes that occurs is sorption be the mostimportant for ct .itionr, of performance (Parsons, et als 1990). Solute species adsorb to the mat 6x assessment.The half-lises of namm and krypton-85 are or fracture surfaces tr forming bonds with the molecules relatively short (12.3 years and 10.7 years, respectively), 'i on the solid surface.He strength of these txmds and the and it is possible that elemental iodine could quickly par-kinetics depend on many chemical factors such as: tition into the liquid phase. Because of the complexity of -(1) electric charge of solute and solid,(2) saturation of ' the issue and the relatively poor state of knowledge about bonding sites, (3) pH. (4) oxidation and reduction poten-gaseous release and transport, cartxn-14 release to the tial, and (5) teraperature and pressure of the hydroge-atmosphere was not included in the total system analysis. ologic system (Freeze,1979) - An auxiliary analysis for carbon-14 release to the atmos. Adsorption can be physical (generally considered a re-versible process) or chemical (generally considered an 6.2.3.3 Direct transport irreversible process). At any moment, some of the s dute particles are bonded to the solid surface and some are Potentially significant scenarios for the assessment of re-free to move with'the groundwater. The adsorption-pository performance involve the possibility of voletmism desorption process has typically been represented in most in the form of a disruptive event such as a magmatic groundwater transport models, using a retardation equa-eruption, or an intrusive event involving human drilling ~ ion that employs a distribution coefficient.The assump-activities. Both scenario classes involve events whose esti-t tions in this model include instantaneous and reversible mated likelihood of occurrence and consequences are adsorption and desorption (equilibrium), linear sorption very uncertain over the regulatory period of performance isotherms, and single-valued - eption isotherms (i.e., no for the repository (i.e.,10,000 years). Considerations for - hysteresis effect)(Rasmussen and Evans,1987). The dis-magmatic events and human intrusion are discussed next. tribution coefficient model was adopted for this study. Ilowever, due to the complexity in understanding and Future work will need to perfonn supporting geochem-predicting magmatic events, simulation work in this area ical analyses, to determine the degree of validity of the was not performed in this study. . present approach. (a) Magmatic Events The model ignored precipitation of radionuclides along the flow path, although solubility was taken into account . Basaltic eruptions are noted to have occurred near the in the source term. Tnis assumption is conservative be-Yucca Mountain site and west and south of it during the cause it would overestimate the cumulative release. . Quaternary period. Basalt flows and cinder cones have been observed on Crater Flat,'and volcanic centers in Amargosa Valley have deposited ash falls as recently as Table 6.1 20,000 to 30,000 years ago. The consequences, assuming Identification of Ligeld Pathway Processes that a magmatic eruption occurs, are very uncertain; how- - and Estimated Effect on Calculating ever,it is believed that this class of scenarios would need Cumulative Release from the Liquid Pathway.- to consider the following in estimating consequences: (1) entrainment of the waste and deposition on the sur-Estimate of - face, for example, as a result of a physical (steam) explo-Processes - Importance sion; (2) dispersal of fine-grained ash and radioactivity into the atmosphere; (3) mechanical and thermal loading - L 1Advection ' High - that can affect rock stresses and permeabilities and flow-2; Sorption, High conditions for radionuclide migration from the repository

3. - Radioactive Decay and Production High to the accessible environment, even if the event does not 4.'

Fracture-Matrix Interactions -High. compromise the structural integrity of the repository;

5.

Matrix Diffusion Medium (4)the relative amounts of radioactivity that would be 6. Precipitation of Radionuclides Low released due solely to the occurrence of this natural 7. Dispersion; low event; (5) potential barriers to flow or water table level changes; and (6) the source term. - 6.2.32 [ Gas transport The source term depends upon many factors, including: Mix of waste forms for the repository (spent fuel and De gas pathway is an alternative pathway for-radio-nuclide transport to.the accessible environment. Further-high level waste from defense activities) 27 NURl!G-1327

6. Flow and Transport Mcdels o

Spent fuel inventory characteristics (reactor type modelt used in a systems model. Some examt les of the and burn.up) types of programs needed are: (1) two-phase flow pro-o Time of emplacement gram for analyzing thermal effects; (2) tww or three-o Emplacement configuration dimensional programs for simulating regional flow; ^ E# "*" E W"** "E ^ o Rock geochemical properties nomena; (4) a program that mcludes the tnfluence of o

• 1_tme of eruption or intrusion fractures or allows for an interaction between fractures o

Extent, location, and geometry of volcanism and matric and (5) an efficient transport program for use in the multiple simulations of a performance assessment, For scenarios involving the intercepw - waste pack-ages by feeder dikes, estimates of the distribution and size 'lhe review of computer programs is divided into the of these dikes (resulting f rom the feeding of basaltic cin-following four sections:(1) regional or far. field ground-der cones) are needed, in addition to estimates of their water flow programs; (2) two-phase flow programs; times of occurrence (to account for radioactive decay). (3) geochemical programs; and (4) transport programs. ~ The ability of the varioJS programs to deal with the prcS-(b) lluman Intrusion ence of fractures will be discussed unde, the individual programs. A summary of the review and the selection lluman activities such as deep exploratory drilling of rationale is provided in the subsequent sections, whereas boreholes could potentially provide direct releases of ra-individual program summaries are provided in Appen-dioactivity to the environment. It is believed that this dix C, issue is primarily a source term issue which depends on the amount of radioactivity brought to the surface by drilling. In general, the waste package material, emplace. 6.3.1.1 Regional groundwater flow programs ment configuration, age of waste at time of interception A number of unsaturated flow programs (e.g., FEM-by a drill bit, altogether contribute to estimating the ra-WATER and UNSKl'2) were developed approximately dioactive source term. Estimates of radioactivity brought to the surface in contaminated cores from those 10 years ago, to analyze unsaturated flow m near-surface boreholes that intercept the repository are also needed s ils (Ih mas, et al.,1982). NRC participation in the for a more complete consequence analysis. In order to nternational project HYDROCO!N (Cole, et al.,1987) revealed sigmficant numerical limitations in these pro-estimate the risk, one needs to combine the consequence information with a probabilistic analysis of the drilling grams m simulating unsaturated problems m, velvmg large non-linearities (e.g., mfiltration into a dry soil and large rate end penetration depth. permeability contrasts)Jihese and similar-type programs were not exami=d further due to their numerical defi-6.3 Computer Program Review and ciencies, which would be unacceptable in evaluating un-Selection saturated fractured media. A new generation of unsatu-rated flow programs has been developed to better handle The analysis of any complex system often involves the use the non-linearitics encountered in unsaturated flow. of computer-implemented mathematical models to assist t,he analyst in presenting an " adequate" description of the The computer programs entitled S(TER A, V AM2D, and risk or performance of the system /Fhe analysis of hydro-TRACER 3D were selected as representative of the logic systems has, over the last 20 years, created a number newer generation of unsaturated flow programs. The of computer programs for analyzing a variety of problems three programs employ similar Darcian approaches to (until recently little attention has been paic to an unsatu. simulating fluid flow in porous media.The ability to simu-rated, fractured, and uneconomic rock such as tuff). late fracture flow could only be accommodated through a Based or' the path-phenomena and types of scenarios single-continuum, composite porous medium approach. anticipated for the . lysis of a repository in unsaturated (Currently, there are no existing programs that simulate fractured tuff, computer programs were reviewed for fracture-matrix interactions with an approach dtiferent their applicability in a performance assessment. from the single-continuum approach. SNI, under an NRC contract, is descloping a flow program that will account for the fracture matrix interactions m a more 6.3.1 Liquid Pathway rigorous fashion than is currently available.This program The evaluation of the liquid pathway could involve a suite is scheduled for completion in April of 1990.) of computer programs.'Ihe complexity of flow and trans-port in unsaturated fractured tuff could dictate the use of The VAM2D program (lluyakorn,1989) was selected for a set of models. A specific model could be used to evalu-use in modeling regional flow because of the efficiency of ate a specific performance question, assist the assignment the non-linear numerical techniques employed and the of model inputs, or justify the assumptions of simpler availability of the program for NRC staff use. NUREG-1327 28

6. I' low and Transport Models 6.3.1.2 Two. Phase flow programs number of simulations r.ecessary for sensitivity and uncer-tainty analyses (see Appendix 171'esting Statistical Con-Assessing the ther nal period of the HLW repository will vergence"). Some of the simphfications being considered require programs that can simulate the flow of air, liquid are: use of a one-or two-dimensional analysis; limited (tf water, and water vapor. TOUGil, NORI A, and PITIROS am) interaction between fractures and rn' :.trix; steady-are existing programs that sohe the two-phase flow and sta'te flow; and limited peochemistry (typically a lumped energy-transpon problem. A detailed SNL review of tetardation factor that is intended to account for all the these programs (Updegraff, T) discussed the difficul-geochemical interactions).

ties of running two-phase flow models and the relative strengths and weaknesses of the individual programs. A number of existing programs, which employ many of Overall, one program was not superior to the others. the aforementioned simphfications, hase been reviewed Ilowever, TOUGH successfully ran most of the test prob-(see code summaries in Appendtx C). The review lems, whereas NORIA and PirlROS could at best simo-included numerical solutions such as SPARTAN, late approximately half the test problems. NEl'1R AN, and TOSPAC, as well as closed-form solu- ~ tions such as the UCil programs. 'lhe NiilPIR AN The TOUGH program (Pruess,1987) was selected to (l ongsine. ct al.,1987 ) progt am, develope'J at SNI. under analy7e two-phase flow problems because of its ability to NRC funding, was selected because of: (1) availability on handle a variety of problems (Updegraff,19S9) and the NRC computer systems; (2) ready access to the SNL de-current availability of TOUGii to NRC staff. (Due to the velopers; and (3) efficiency of the program, and compatb complexity of two-phase flow problems, simulation work bility with the 1.IIS computer program for analviing was not performed in this study.) model sensitivity. 6.3.1.3 Geothemical programs Although all the reviewed programs did not fully describe fracture-matttx interactions SNI. is currently modifying The geochemical behavior of the Hl.W repository could NEl71RAN to irAlude fracture-matrix mteractions (to be have a very strong effect on the movement of rad 10-completed by March,1990) Staff use of the current ver-nuchdes. Unfortunately, current geochemical programs sion of Niil'IR AN wdl assist technology transfer of the are not amenable to most performance-assessment s)s-new version of Niil71RAN in 1990. tems programs because of their complexity. The pnmary use of the geochemical programs will be to aid the under-63.2 Gas Pathway standmg of the geochemistry of the site and the assign-ment oilumped parameters in the simpler transport mod- 'Ihe pas pathway was treated as an anuhary analysis and is els. presented in Appendix D. The current study did not consider complex modeling 633 I)irect Pathway associated with geochemicalanalyses. Summaries of vari-ous programs are included in Appendix C. Selectian of a Computer programs for evaluating the consequences of particular program was considered inappropriate until drilling into a repository were not able to be acquired in a more specific performance issues or questions with re-timely fashion. A model was developed that accounts for spect to geochemistry could b y made, the anticipated important aspects of a drilling scenario. The model accounts for a drilling rate, radioactive decay, the areal extent of the repository, waste package em-6.3,l A Transport programs placement orientation (horizontal versus verticalk and The use of a transport program in a systems code for the boreholes intercepting both the waste package and con-performance assessment will require a number of simpli-taminated rock. A detailed dtseussion of the drilling fications of the real system to accommodate the large model is provided in Appendix 11. 29 N U R1 E 1.427

7. METilODOLOGY FOR SCENARIO DEVELOPMENT 7.1 Introduction they were initiated. Those phenomena irutiated in the accessible environment were classified as external pertur-When this study was initiated, the staff intended to ac.

bations of the repository system, even if the effects of the complish two objectives: (1) identify a methodology that phenomena occurred within the repository. Thus, fault could be used for scenario development, and (2) demon-movement within the controlled area of the repository strate the utility of the methodology by application to the was classified as an external event because the tectonic Yucca Mountain site. Due to limitations on availabihty of forces responsible for the movement wcre external. Simi-staff resources, only limited proress was made on appli-larly, Jrilling into a repository was classified as an external cation of the methodology. This section is, therefore, event because the drilling was initiated outside the con-trolled area. Phenomena mternal to the repository sys-primarily a status report of on-going work, and consists tem, such as corrosion of waste canisters, were assumed to primarily of a description of the selected methodology. llecause application of this methodology to the Yucca be addressed in the development of model(s) and data ~ Mountain site was not completed, there is no correlation base (s) desenhing the wpository system, and therefore between this section of the report and the scenario classes were excluded from corweration for scenario develop-ment. hypothesized for analysis in other sections of the report. 1 or this antdyns, the boundary of the repository system An important part of a performance assessment for an was chosen to be comcident with,the boundary of the HLW repository is an evaluation c.* the uncertainties in accessible environment, for two reasons. First, many of projected performance, Two general approaches ire the uncertainties withm this boundary mvohy processes available for analyses of uncertainties in repository per-rather than discrete disruptive events. Simulatton of proc-formance. Such analyses can be carried out by incorporat-esses and their associated uncertainties is often faisly ing the uncertainties directly into the model(s) and data smple, sometimes involving no more than specification base (s) describing the repository system, or uncertainties of a range of values withm the data base for the repository caa be approximated as " scenarios"-i.e., descriptions of (e.g, a range of mosion rates). On the other hand, alternative ways in which the repository system might phenomena outside this boundary are often rare, discrete perform in the future. Most analyses use a combination of events such as fault movement or volcanic activity. Simu-the two approaches, although there are generally no ex-lation of such events withm the model of the repository plicit criteria for which way to treat a specific source of system may be awkward, especially when Monte Carlo or uncertainty. Thus, lists of processes and events to be related simulation techniques are used. In such cases, the included in scenarios often include phenomena such as number of simulations needed to obtain a good represen-waste-canister corrosion, even though such phenomena tation of repository performance may be so large that are likely to be evaluated directly within the repository accurate appronmations of performance are not practi-model(s) and data base (s), rather than as scenarios. cal. This study distinguished two aspects of an uncertainty The second, and more important, reason for selection of analysis: (1) uncertainty about the characteristics of the tne repository system boundary involved the way in which repository system and its environment as they exist at the the rqdtory is perceived by regulators and by the pub-time of analysts, and (2) uacertainty about the future lic. Iloth groups tend to view the repository system as evolution of the environment withm which the repository ending at the accessible environment boundary and to will exist far into the future. For the purposes of this visualize phenomena occutting outside this boundary as study, scenario analysis was limited to the second type of external perturbations of the repository. Scrutiny of re-uncertainty. All uncertamttes of the first type were as-pository safety tends to take the form of "What iP' ques-sumed to be incorporated directly into the model(s) and tions-e.g., What happens to the repository if a volcano data base (s) that describe the repository system. erupts nearby? livaluation of external phenomena through scenario analysis directly answers such questions, The term " scenario" is defined here as a description of whereas incorporation of external phenomena into the one of the many alternative ways in which the environ-repository system model(s) or data base (s) would tend to mem af a repository might evolve in the future.The goal obscure the results of the analysis, of a scenario analysis is, then, to identify a set of such scenarios, to be used in uncertainty analyses, that is suffi-It is important to note differences between the approach ciently complete to support a regulatory decision on the adcpted here for scenario development versus those pro. acceptab;1ity of the repository. posed by other analysts. Ihidgkinson and Sumerling (1989) describe an approach for scenario development in in this study, phenomena were considered to be either which no distinction is made between " internal" phenom. " internal" or " external." dependmg on the location where ena and those that occur outside the repository. In their i 31 NUlWG-1327

7. Methodology for Scenario Development approach, processes such as waste-canister corrosion on repository performance, if the process or event would be treated as phenomena to be combined into were to occur.

scenarios for analysis. Because these authors combine internal with external phenomena, their list of " events, 4. Scenario Construction. Processes and events surviv-features and processes" to be combined into scenarios ing the screening of Step 3 are combined to form contains approximately 150 entries and, even after " scenario classes" using the event tree approach screening out unimportant entries, the number of scenar-described in NUREG/CR-1667 (Cranwell, et al., ios that could be constructed from a list of this length 1982). (Each " scenario class" is a unique combina-would be quite large. Treatment of internal phenomena tion of processes and/or events without regard to the within the repository system model greatly reduces the order in which they occur.) For this study, different potential number of scenarios, keeping the complexity of permutations of events that comprise separate scc-the repository analysis within manageable bounds, narios were not considered. Instead, judgment was used to select a permutation to be representative for Hodgkinson and Sumerling also describe an alternative the scenario class. For the illustrative purposes of approach, referred to as "em>ironmental simulation," in this project,it was planned that le only scenarios to which an attempt is made to incorporate all identifiable be formed would be those consisting of zero, one, or uncertainties into the repository system model. As dis-two processes or eventsw.e., scenarios containmg cussed previously, it appears that such an approach would three or more events would not be fortned. have difficulty satisfying the information needs of regula-tors, and could require excessive numbers of simulations 5. Scenario Probabilities. Scenario probabilities are esti-in order to provide accurate approximations of repository mated by combining the probabilities of the proc-perfowance. esses and events that comprise the scenarios. If the processes and events comprising a scenario are inde-pendent, the scenario probabihty is determined by 7.2 Methodology multiplying the probabilities of the constituents. If the processes and events are not independent, cor-The scenario development approach adopted for this relations or causal relationships may be considered study is an adaptation of the event tree approach used in when determining s:enario probabilities. probabilistic risk analyses, and consists of the following steps: 6. Scenario Screening. Scenarios are screened using the same critena as for screening processes and events 1. Identification of Processes and Events. This step in. in Step 3. volves identification of a comprehencive set of proc-esses and events that could adversely affect reposi' 7.3 Application tog performance. Only " external processes and events occurring (or initiated)in the accessible envi-Application of the selected scenario-development meth-ronment arc included. Processes and events internal odology for Yucca Mountain was largely limited to the to the repository system are assumed to be treated as first step-identification of processes and events. He uncertainties within the model(s) or data base (s) primary source of information used to compile a list of describing the repository system, and therefore are processes and events was the staff's knowledge of the not meluded. When the time of occurrence of a Yucca Mountain site, although limited references to lit-process or event (e.g., volcame activity) is expected crature describing similar scenario development efforts to have a sigmficant effect on repository perform-for Yucca Mountain were also made. Some progress was ance, the time is specified as part of the description also made on the third step, involvmg screening processes of the event, and occurrences at several different and events. However, because probability assignments times could be listed as separate "subevents." were not completed, screening was conducted only on the bases of physical reasonableness and insignificant conse. 2. Estimation of Probabilities. Probabilities of the proc-quences. Combination of processes and events into sce-esses and events are estimated from historical data, narios, development of scenario probability estimates, models of the processes and events, or expert jud - and scenario screening (Steps 4 to 6) must await develop-8 ment. ment of probability estimates for the phenomena com-prising the scenarios. Table 7.1 summarizes the candidate 3. Screening of Events and Processes. Where possi';te, list of processes and events identified, including those processes and events are eliminated from it.e i;t that were later screened from the list. (As additional compiled in Step 1, using the following sNeening knowledge about the site is acquired, it will obviously be criteria: (a) lack of physical reasonablene.,s; (b) low necessag to periodically review both the completeness of probability of occurrence; and (c) insignP.icant effect the list and the specific descriptions of processes and NUREG-1327 32

7. Methodology for Scenario Development events making up the list.) Table 7.2 provides a more sive, definition of "subevents, as was done for vol-detailed description of each process and event and, v.here canism in this analysis, provides a way to remove appropriate, the basis for screening.

conservatism and to generate a more realistic ap-proximation of expected repository performance. 7A Conclusions 3. As for other risk analyses, no way has been found to ensure completeness of the initiallist of processes Although only limited progress was made in applying the and events from which scenarios are formed. An selected methodology for scenario development, several approach similar to fault tree analysis, in which the tentative conclusions have been reached. repository system is examined to identify potential failure modes, seems a useful way to check on the 1. The methodology appears to be workable. The dis. completeness of process and event identification. tinction between " internal" and " external" proc. esses and events appears to have merit for determin-ing which uncertainties are to be incorporated 4. Alternative approaches to scenario analysis, such as directly into the model(s) and data base (s) describing those described by thidgkinson and Sumerling the repository system and that are to be addressed in (1989), appear to differ primarily in the degree to scenario descriptions. This distinction also appears which they address uncertainties in the model(s) and to be capable of limiting the number of processes data base (s) describing the repository system or in and events in the scenario analysis to a manageable scenario descriptions.The approach selected for this study was intermediate between the extremes pro-level. posed by others, and appears to be a reasonable 2. Scenario descriptions are necessarily only approxi-trade-off betweca the desire for a highly detailed mate descriptions of future repository performance, simulation of repository performance and the need and must incorporate significant conservatisms in to limit resources expended on the simulation.~lhe order to limit the number of scenarios to be evalu-selected approach also appears to have advantages ated. In particular, the time at which a process or over alternatives for producing information in a event is assumed to disrupt a repository may be form that corresponds to the nceds of NRC's regula-highly conservative. If such conservatism is exces-tory pmcess. 33 NURiiG-1327

=. ~. _- 7[ Methodology for Scenario Development c Table 7.1 ' List of Processes and Events * = II. l Tectonic A. - Volcanic

1. ~ Extrusive a! On-site-

- Ei Years 0-100 ii. Years 101-1,000 iii. Years 1,001-10.000 b. Off-site 2. Intrusive - a; Upgradient b. Downgradient c; Intersecting repository B. Regional Uplift & Subsidence - 1. Increased rate of uplift

2. _ [ Subsidence]

Cc Fault Movement ' 1. . Fault within controlled area a'. Within underground facility 3 b; Outside underground facility 2. Fault outside controIIed area a. Location alter:: groundwater flow b. Effects limited to ground motion ~ ~ II. Climatic A.. [ Current climate-extreme weather phenomena] B. - [ Increase in frequency or intensity of extreme weather phenomenal - C._ - [Glaciation) .li [ Covers site with ice] 2 [Causes sea level change] D. Change in precipitation 1. Piuvial period

2. -Drier period

Ill. Human-initiated -

- A. = Greenhouse effect 1. ' Increased precipitation. - 2. Reduced precipitation-

D-

[ Climate controll i c-C. [ Weapons testing at _ Nevada Test Site (NTS)] , D.~ ' D;illing j. -L ' Intersects canister L-2. Misses canisters . NUREG-1327 34

7. Methodology for Scenario Development Table 7.1 IJst of Prucesses and Events * (continued)

E.:- _ Mining 1. [Above underground facility)- 2. .At or below underground facility F. Withdmwal well(s) at or beyond controlled area 1. Small, single-family drinking water well 2. Large drinking water well (addition to las Vegas supply) 3. Agriculturalirrigation wc!! IV. ' Other A.. [ Meteorite impact], B. ????

• Proccues and events enclosed in apare brackets are acreened frorn further analysis.

35 NUlWG-1327

7. Methodology for Scena ia Development 1

bble 7.2 Descriptions of Processes and Esents" Process or Event Description 1 On-site extrusive volcanic activity. A tusaltic volcano erupts through the underground (L A. I.a.) facility 'lhe volcano is fed through a dile. Waste canisters within the dike mix with the magria, and their contents are erupted. The size of the dike is assumed to be , which is sufficient to eject from the underground facility percent of the originally emplaced waste. This size is the worst credible, and is taken to be representative of allless disruptive events.Three "subevents" are defmed, based on the assumed time of occurrence. (a) Subevent la, occurring immediately after repository closure, represents all occurrences during the first century after closure, (b) Subevent Ib, occurring at year 101, represents all occurrences between year 101 and year 1,000. and (c) Subevent Ic, occurring at year 1,001, represents all occurrences between year 1,001 and year 10,000. Screening on the basis of likelihooJ is done only on the overall probability of occurrence of the event during 10,000 years-not on the probabilities of the subevents. The probabihty of Event 1 is estimated to be 2 Off. site extrusive volcanic activity. Off-site activity is a likely candidate for screening (LA.I.b) from the list because potentially detnmental effects seem unhkely. Ilowever, the event was retained pending a more thorough consideration of potential effects such as alterations of regional or on-site hydrological or geochemical conditions. 3 Upgradient intrusive volcanic activity. An igneous intrusmn at ('I. A.2.a) Occation) upgradient from the undeground facihty forrrs in a way that alters groundwater flow downgradient fro a the hicapm of tbc intrusion.1he intrusion is in the form of a dike with dimensions of , and reaches to a depth of below the ground surface.The hication and dimenstons are the worst credible values, in terms of effects on repository performance, and are taken to be representative of all less disruptive intrusions. He temperature of the intrusne materiid is _ , causmg thermal alterations of surroundmg groundwater flow conditions. De probability of livent 3 is estimated to be 4 Downgradient mtrusise volcanic activity. An igneous intrusion forms at (l.A.2.b) (location) downgradient from the underground facility. F.xcept for h> cation, this intrusion is identical to that of Event The probabihty of Event 4 is estimated to be 5 Volcanic intrusion intersects underground faJlity. An igneous intrusion identical to (LA.2.c) that of Event 3 forms beneath the undergrounu facdity, intersecting emplaced waste, but not reaching the ground surface. The probabt.ly of livent 5 is estimated to be 6 increased regional uplift.The existing rate of uplift at the repository site increases (LIL1) to a rate of immediately after repository closure and then remains constant for 10,000 years. This same uniform rate of uplift also occurs within a surrounding area of dimensions The probability of process 6 i3 esumated to be Subsidence. Subsidence was screened from the list because potentially disruptne (LH.2) effects could not be identified. 7 Fault movement within underground facihty. A fault mtersectmg the underground (l.C. I.a) facility moves immediately after repository closure, resultine in an offset of along the fault. (Should the type of fault, dimensions, etc. be specificJ7 Is simultaneous movement on a series of faults within the underground facihty posuh!c and. if.so, should that be the descnption here?)This event is taken to he representatise of all similar esents with less detrimental effects on waste isolation. NUREG-1327 36

7. Methulology for Scenario Development Table 7.2 Descriptions of Processes and Events"(continued Process or Event Description

'lhe probability of Event 7 is estimated to be 8 Fault movement within controlled area. A fault intersecting the controlled area, but (l.C.lb) not the underground facility, moves immediately after repository closure, resulting in an offset of along the fault. This event is taken to be representative of all similar events with less detnmental effects on repository performance. The probability of Event 8 is estimated to be 9 Fault movement outside controlled area alters groundwater flow. A fault located (l.C.2.a) outside the controlled area moves immediately after repository closure, altering groundwater flow characteristics in a way that potentially influences waste isolation. The location of the fault is and the effset along the fault is . This event is taken to be representative of all similar events with less detrimental consequences. The probability of Event 9 is estimated to be . (NOTE: If both upgradient and downgradient locations of fault movement capable of altering groundwater flow are credible, separate events might need to bc defined analogous to Events 3 and 4, previously mentioned.) 10 Fault movement outside controlled area causes ground motion. A fault located outside (l.C.2.b) the controlled area moves, causing ground motion at the underground facility and shaft and borehole seals. The maximum acceleration and the frequency of motion are . This event is taken to be representative of all similar events with lower acceleration or less detrimental frequencies. The probability of Event 10 is estimated to be . (NOTE: It might be possible to combine Events 9 and 10 although, in ~ general, these events will be different since Event 9 depends strongly on the location of the fault movement, whereas livent 10 is concerned with the ground motion produced by an event at any location.) (ll.A) Current climate-extreme weather phenomena. Extreme weather phenomena, such as tornados, hurrie:mes, etc., were screened from the list because potentially detrimental effects on waste isolation could not be identified. (II.B) Increased frequency or intensity of extreme weather phenomena. 'Ihese phenomena were also screened from the list because potentially detrimental effects on waste isolation could not be identified. (11.C) Glaciation covering site with ice or causing sea level change. Glaciation causing the site to be covered with ice was screened from the list because of lack of evidence of occurrence during previous glacial episodes. Sea-level changes caused by glaciation were screened from the list because potentially detrimental effects on waste isolation could not be identified. 11 Pluvial period. A period of increased precipitation begins immediately after repository (ll.D.1) closure and continues for 10,000 years. Precipitation at the site and throughout the surrounding r,;gion is increased by 50 percent compared to current levels. This event is taken to be representative of all similar events of later onset, shorter duration, or smaller changes in precipitation. The probability of Event 1I is estimated to be 12 Drier period. A period of reduced precipitation begins immediately after repository (ll.D.2) closure and continues for 10.000 years. Precipitation at the site and throughout the surrounding region is reduced by 50 percent compared to current levels. This event is taken to be representative of all similar events of later onset, shorter duration, or smaller changes in precipitation. The probabihty of I! vent 12 is estimated to be 37 NLIR EG-1327 1

7. Methodology for Scenario Development Table 7.2 Descriptions of Processes and Eunis"(continued)

Process or Event Description 13 Greenhouse effect-increased precipitation. The greenhouse effect causes precipitation (Ill.A.1) to increase by 30 percent above levels that would have otherwise prevailed. He increase begins immediately after repository closure and continues for 10,000 years. his event is taken to be representative of all similar events of later onset, shorter s duration, or smaller changes in precipitation. He probability of livent 13 ts estimated to be s 14 Greenhouse effect-reduced precipitation.ne greenhouse effect reduces Oll.A.2) precipitation by 30 percent compared to levels that would have otherwise prevailed. The decrease begins immediately after repository closure and continues for 10,000 years.This event is taken to be representative of all similar events of later onset, shorter duration, or smaller changes in precipitation. The probabihty of livent 14 is estimated to be Oll.H) Climate control. This event was screened from the list because of low likelihood. It is presumed that the institutional controls required by 10 CFR Part 60 wdl be sufficiently effective to prevent any events of this type that could detrimentally affect waste isolation. (III.C) Weapons testing at NTS. This event was also screened from the list by presuming that the institutional controls required by 10 CFR Part 60 will be sufficiently effective to prevent any events of this type that could detrimentally affect waste isolation. 15 Drilling intersects a canister, Wildcat drilling fcr petroleum breaches a canister OII.D.1) allowing part of the canister contents to be brought to the surface in drilling fluids. Wildcat drilling for petroleum is taken to be representative of all potential dalling at the depth of the underground facility. The frequency of drilling at the repository site is estimated to bc , and the probability that any one drilling c ent will breach a canister is estimated by the geometric relationship between the area of the waste cimisters and the total area of the underground facility. 16 Drilling misses canisters. Wildcat drillmg for petroleum penetrates the underground GII.D.2) facility, but misses all canisters. This type of drilling is taken to be representative of all potential drilling at the depth of the underground facility. The frequeng of dnlling at the repository site is estimated to bc , and the probabihty that any one drilling event will miss all canisters is estimated by the geometric relationship between the area of the waste c misters and the total area of the underground facihty. (111.11.1 ) Mining above the underground facility. his event was screened from the hst because effects potentially detrimental to waste isolation could not be identified. 17 Mining at or below the underground facility. Construction of shafts and other mining (Ill.E.2) activities are assumed to be carried out only if direct contact with wastes does not occur. If wastes are directly contacted, it is assumed that their character will be recognized, mined openings will be sealed, and mining activities will be abandoned. The frequency of mine construction is estimated to be , and the probability that mining activities will contact waste crmisters is estimated by the geometric relationship between the area of the waste canisters and the total area of the underground facdity. NURl!G-1327 3x

l

7. Methodology for Scenario Development Table 7.2 Descriptions of Processes and Esents" (continued)

Process or Event Description 18 Small water well. A small, single-family drinking-water well is assumed to be located (Ill.F.1) at the downgradient boundary of the controlled area and is used as a year round domestic water supply. 'lhe well is assumed to be dnlled 100 years after repository closure, and is used continually for the next 9900 years. Use of the well enhances the hydrologic gradient within the controlled area, potentially affecting transport of radionuclides to the accessible environment.The probability of Event 18 is estimated to be 19 Municipal drinking-water well. A municipal drinkmg-water well is assumed to be drilled (Ill.F.2) at the boundary of the controlled area at year 100 after repository closure, and the well is assumed to be used until year 10,000 after closure (or until depletion of available groundwater supplies). The effect of this well on repository performance is limited to potential alterations of regional groundwater flow characteristics. It is assumed that current requirements for monitoring the quality of municipal water supplies will continue, so that remedial actions will be taken tf radioactive contamination of water supplied by the well occurs. The probability of this event is estimated to be 20 Agricultural irrigation well. The assumptions regarding this well are identical to those (Ill.F.3) for Event 19 except that monitormg for potential radioactive contamination of the water is not assumed to occur. Therefore, remedial actions will not be taken to stop potential releases of waste via this well Because no monitormg of water quality is assumed, both releases of radionuclides and alterations of hydrologic gradients within the controlled area may be more severe than for Scenario 19 The probability of this event is estimated to bc (IV.A) Meteorite impact. This event was screened from the list because of low protubility. Several references in the technical literature demonstrate that the probability of impact by a meteorite large enough to disrupt a repository is extremely small. "Utanks indicate infonnation to be developed later. 39 N UIEG-1327

l l l

8. AUXIIlARY ANAINSES SUMMARIES 8.1 IntrodiletiOn Approximately an order of magnitude more simulations than the rule of thumb would indicate were requited for in general, the auxiliary analyses conducted for this dem-tM ment problem.The most likel) reason for this re-onstration were directed toward the evaluation of the sult, was the very few simulations that provided a non-appropriateness and limitations of various computational vero result in the high consequence part of the CCDF.

approaches and the analysis and interpretation of data used in this study. These analyses included: the to dimensional cros5 sectional flow simulation of a layered SA Analysis ol'ilydrologie Data porous site, the analysis of hydrologic data, and the analy-(Appendix F) sis of statistical convergence for a CCDF. Additionally, a separate analysis of carbon-14 releases was performed to An auxiliary analysis of hydrologie data was conducted to supplement the liquid and direct pathway analysis. The determine if spatial correlations could be identified for aforementioned auxiliary analyses are discussed in detail porosity and hydraulic conductivity parameters. A large in the appendices. A bnef description of the analysis are scale trend of decreasing porosity with increasing depth provided next. was identified in data from three holes drilled into the Topopah Spring umt, and a small-scale correlation length of less than 40 meters was identified in data from two 8.9 Carbon-14 Analys.is (Append.ix D) holes drilled into the Topopah Spring unit. llowever, this The release of cirbon-14 ' rom waste packages is a poten-analysis did not idenufy any spatial correlation with depth tial concern for a repoutory located in the unsaturated for Calico Ilills porosity data or for saturated hydraube zone because of the presence of a fast pathway (gas conductivity in either the Calico llills or the 'lopopah through the fract'.res) to the accessible environ' ment. Spring units.This was relevant to the flow 1md transport Due to the com;aexity of the source term comiderations modehng. because long correlation lengths lead to a of this proble a, the analysis was not considered appropri-broad travel time distribution for each column (see Sec-ate to be included in the total CCDF. However, it was non 9.3.1.4). Very short correlation lengths lead to the - considered important to perform some simple calcula-conclusion Inat there is a smgle groundpater tiavel time tions to obtain a better appreemtion and understanding of per column and little hkelihood of long. last, groundwater the magnitude of the problem and some of the concerns. flow paths. In the flow and transport modeling, it was assumed that there was no apparent spatial correlation The analysis identified elease mechanisms and the geo. for saturated hydraulic conducuvity beyond 10 meters chemistry of calcite precipitation as areas where data sep;uation (see Section 9.3.1.5). collection and further investigation (vould be most fruit-fuh 8.5 Two-Dimensional Flow Sinmlation (Appendix G) 8.3 Statist.ical C,onvergence (Appendix E) ^ two-dimensional flow simulation was conducted to en amine the potential for flow diversion at unit interfaces or There are rules of thumb for determining the number of the propensity for non-vertical flow, The analysis, w hich Monte Carlo simulations to perform to provide statisti-considered only matrix flow, showed that considerable cally representative results. Ilecause of the highly non-non-vertical flow would occur at interfaces where the linear problems currently being tackled. it was deemed saturated conducuvay of the lower unit was 75 percent or appropriate to investigate the number of simulations re-less of the infiltration rate. Future work wdl need to quired to obtain statistical convergence. consider the effect of fractures on non-vertical flow. l 41 NUlt!Gl327

9. ANALYSIS AND RESULTS Previous sections have described the methods and ap-These types of events were selected, in part, because they proaches for c>timating performance and the evaluations would demonstrate interesting aspects of repo;itory per-used to select the various methods and approaches. This formance and because the mohling variations needed to section describes the implementation of the methods and accommodate these events were not excessive. Thus the results obtained. *lhe following correspondence exists treatment of the scenario classes invohmg climatc between the previous sections and the current section:

changes, called pluvial conditio,x in this study and repre-sented by increased infiltration a..J a rise in the water Previous (Methods) Current (Implementation) table at Yucca Mountain, were accommodated by slight modifications to the data used as input to the model Section 4 System Code Subsection 9.6 Total CCDF representing groundwater transport.11xcavation of radio-Section S Source Subsection 9 2 NEI' IRAN activity contained either in the repository or in contami- 'T,erm Source nated host rock could be relatively easily modeled to what Term is believed to be an acceptable degree of accuracy. In Modc! addition, excavation of radioactivity is an archetypical Section 6 l' low and Subsection 9.3 Flow and direct-release event, representative of the type of model-Transport Transport ing anticipated for similar direct-release mechanisms like Models Models volcanism. Section 7 Methodology Subsection R1 Treatment 'the two classes of fundamental events were combined to for Scenario of Scenarios form four classes of scennios: Development Scenario # 0 : Base case conditions and no drilling Two additional subsections are added to this section to Scenario # 1 Pluvial conditions and no drilling comp'ete the exposition of implementing the mcthodol-Scenario # 2 : Base case conditions with drilling ogy. Subsection 9.4, Parameters, describes values und in Scenario # 3 : Pluvial conditions with drilling. the analysis, and Subsection 9.5, Sensitivities and Uncer-tainties for Liquid Pathway Analysis, describes a demon-Consequences for the base-case scenario were estimated stration of analytic methods. by the output of the NEFIRAN code, as described in Sections 6 and 9.2 of this report. The pluvial case was estimated by the NEFIRAN code, but with input modi-9.1 Treatment of Scenar,10S fled to simulate a higher water table and greater infiltra-tion rate. Because the drilling removed so little radioac-9,1.1 Introduction tivity from either the repository or the host rock, the [ consequences of drilling, to a first approximation, could A general approach for analysis of scenarios is discussed be calculated independently of the details of the migra-in Section 7 of this report. Because work on this part of tion of radionuclides. Ilowever, some of the same factors, the performance assessment was delayed, a less system' such as the removal of waste frorn the repository, influ-atic approach to the treatment of scenanos was taken in enced both the groundwater and direct release pathways, the interest of expediency. In particular, the steps of: so parameters important to these factors were used in (1) identification of processes and events: (2) estimation calculating releases from both pathways. For Scenario of probabilities; (3) screening of events and processes; Classes 2 and 3, consequences from these two pathways (4) scenario class construction; (5) scenario class prob-were calculated and subsequently added together by the ability estimation; and (6) scenario class screening, were system code, collapsed into a more direct approach. Because of the limited time available to perform the Phase 1 analysis. The probability of occurrence of dnlling was considered significant new modeling initiatives were not possible. to be independent of the occurrence of pluvialconditions With this in mind, a small number of interesting scenario (see Figure 9.14; and dtscussion in Subsection 9.6). Al-classes were chosen for incorporation into the CCDF to though drilling boreholes for purposes of aquifer detec-demonstrate how this is done and how results from vari-tion or water exploration and extraction probably would ous scenario classes are combined. depend on the climatic conditions at the site, drilling for purposes of mineral exploration probably would not. Fol-9.1.2 Discussion lowing the guidance provided by EPA in Appendix B of 40 CFR Part 191, a constant drilling rate of 0.0003 Two classes of fundamental events were selected. These boreholes per square kilometer per year, a repository events wer. (1) changes in climate at Yucca Mountain, area of 5.1 square kilometers gave 15.3 as the expected and (2) human intrusion by drilling exploratory boreholes. number of boreholes over 10.000 years.This means that 43 NUREG-1327

9. Analysis and Resuhs the probability of no boreholes at the site over the same no radionuclides would be released, lleyond that time.

time period is very small. Using a Poisson distnbudon to the waste was assumed to be fully accessible to the envi-desenbe dnlling, the probabihty of no boreholes was esti-ronment and could be leached and dissolved. Upon expo-mated to be 23 x 10 7,'Ihus,the probability of drillmgwas sure to the environment, the raJionuclides in the waste ' very close to 1.0.This probabihty may be oserestimated, were assumed to be contained in the uranium dioxide because exploratory dnlling may be done preferentially in matrix and to be released at a rate determined by the teach time T,, i.e.. the time for the matrix to be totally more level terTain (which is not accounted for in the t ave %ge drilling rates) and because the repository's mark-dissoh ed at a constant rate. The leach time is simply the eri may be more effective than was assumed for this stuJy. reciprocal of the leach rate A.. 'The leach rate was esti-L llecause there were no readily available data, the prob-mated on the basisof the totalinventoryof the matrix Mo ability of occurrence of pluvial conditions was assumed to (in grams), the infibration rate 1 (in m/yr), the total sur-be 0.1, and the non-occurrence of pluvir.1 conditions was face area of the s.te, A (in m2), the fraction of inftitrating assumed to be 0.9. water contacting the waste f (unitless), and the solubility of the matrix So (in grams per cubic meter of water): Had the scenario analysis procedure discussed in Sec-2.=1AfS /Mo. (9.1) tion 7 of this report been followed for this Phase 1 dem-1 onstration, the event of no drillmg probably would have been screened out, because of low probability of occur- .l'he rate of radionuclide release was governed by either rence. Attematively, the two scenanos involving no drill-the dissolution rate of the UO matrix or the solubihty of 2 ing probably would have been screened out for the same the mdividual radionuclides. Most of the radionuclides reason,These non-drilling scenarios were retained in this must first be released from the matrix before their solu-analysis for demonstration purposes, and because the bihties become limiting. Since more-oxidi/cd f uel is likely ^cenario analysis effort had not progressed sufficiently to to bc more soluble, this solubility may be a function of tr e the scenario screening procedure. An interesting re-umt Re rate of fuel dissolution might be controlled sult shown in Section 9.6 is that these non-dnlling scenar-either by the amount of water entering the canister, or if ios had a negligible effect on the total CCDF, which is there is ample water, by the solubility of the fuel deter-dominated by the scenarios with drilling. mmed by its oxidation state. The two fundamental events selected for treatment here Once released from the waste matrix, the NiiFIRAN illustrate the striking differences in the importance of program determines if the concentration of the radia-various scenarios to the CCDF that are to be expected nuclide exceeds its solubihty lirnit. If so, then the "undis-when the probabilities of occurrence er non-occurrence sched inventory" for that radionuclide increases, and the of a particular event (such as drilling or pludal conditions) flux leaving the source is hmited by the solubility, lhe p are nearly equal or are orders of magnitude different. undissolved inver. tory can be released later if the concen-u Alsca note that the treatment of drilling consequences,in combination with consequences fromliquid-pathway re_ tration of radionuclides leaving the source term drops below the solubility limit. All variables for the source leases, as a separate pathway depended on the viability of term model, except the initial radionuclide inventories in the assumptions that: (1) the amount of radioactivity re-the spent fuel, were random, generated externally to the leased by drilling is small compared to the total mveritory progmm by the L1IS routine. in the repository and host rock, and (2) drilling boreholes had no substantial effect on the mechanisms important to liquid pathway releases. Had these assumptions not been Several of the radionuclides, notably C-14,1-129. and good approximations, a far more complex treatment of cesium, are known to collect outsde of the uranium diox-the combination of fundamental events would have been ide mat rix (SCP, Section 8.3.5.9), and could be treated as necessary, being solubility-limited rather than teach.hmited. 'lhe inventory fraction of these radionuchdes available for immediate release was assumed not to be sufficiently 9.2 NEFTRAN Source Term Model gre t en ugh to ffect cumulative releases over 10.000 years; thus, no changes to the NiiFIRAN code were needed to accommodate this quick release. For this study, NEFIRAN (t ongsine, et al.,1987) has three built-in the entire inventory of these radionuclides was assumed source term models; solubility-limited, leach-limited, and to be contained in the UO matrix. However, the vari-2 mixmg cell.The solubihty-limited and teach-limited mod - ation in the h> cation of the C-14 invemorv wasconsidered els were adopted for the demonstration. n the gaseous pathw ay auxihary analysis (' Appendix D). In Phase 1, the gaseous releases of C-14 were treated sepa-The engineered barrier lifetime (Tebb a randomly sam-rately from the proundwater release of radionuclides. pled variable in the calculation, was the time before which including C-14. NURliG-1327 44

9. Analysis and Results 9.3 Flow aiid Transport Models centage of the repository determmed to be above each column (see Figure 9.4). Table 4.1 presents the hydroge-The movement of radionuclides could occur in the liquid, ologic units within the columns (labeled A through D),

gas, and direct pathways. As discussed in Section 6, the the thicknesses of the units, and thus the distance from liquid pathway was simulated with the NEFlRAN com-the repository to the water table for each column, and the puter program (longsine, et al.,1987), and a computer fraction of waste present in the repository areas repre-code was developed by the staff to simulate the direct sented by the columns. release of radionuclides due to exploratory drilling.The gas pathway was analyzed as an auxiliary analysis, as dis. There are 6 hydrogeologic units in Column A,5 in Col-eussed in Appendix D. umn 11,4 in Column C, and 2 in Column D. Note that in Column D, the only layers present have very small aver-age k, and that for high infiltration rates, the transport 3 9.3.1 L.iquid Pathway might be dominated by fracture flow, and therefore con-tribute to potentially high radionuclide releases to the Yucca Mountain is a complicated, multilayered system in three dimensions. NEFIRAN can represent the' site only w ater table. Column C is only slightly better, with two thin yers f the Calico Hills Vitric and Prow I ass % elded as an interconnected series of flow tubes. Although capa-ble of representing a two-dimensional flow situation, uma present NEFIRAN is further restricted to having only a single Some timitations of this one-dimensional network model-flow path for radionuclide nugration. Ilecause much ing eppr ch (i.e., simulating the groundwater, release important detail would be lost in the analysis of the com-pathway as four distinct columns of vertical flow, with plicated site with a one-dimensional analysis, the mdenm-tr mpon to the water table)are: NEFIRAN code was modified and run in a manner to partially overcome the limitations of thc one-dimensional 1. lateral flow caused by the diversion of water along structure (i.e., simulate the spatially varying and uncer-interfaces between units and/or obstructions of flow tain conditions at Yucca Mountain). This specialized im-near faults is not taken into account. plementation can be divided into the followmg areas: (1) geometry or network set up, (2) phenomena, and 1 Flow and transpon oi radionuchdes in the satumted (3) data input. zone from the water table beneath the repository to the accessible environment is conservatively ne-9.3.1.1 NEI' IRAN network implementation glected. For this study, the design of the one-dimensional network The source term was conservatively considered to start at for NEFIRAN was based on current information on hy-the boundary of the disturbed zone,25 meters lower than drogeologie units and theories of flow at Yucca Mountain. the plane of the waste emplacement, and therefore closer in the SCP (DOE,1988), the flow at Yucca Mountain is to the water table (NRC Draft Technic d Position on the ~ conceptualized as essentially vertical and under steady-Disturbed Zone,1986). state conditions within the matrix for fluxes less than the matrix-saturated conductivity, ks, and as fracture flow at 9.3.1.2 Implementation of matrix and fracture flow in higher fluxes. (The potential for lateral flow at the contact NEFTRAN between hydrologie units when a higher-conductivity unit is underlaid by a lower-conductivity unit was examined as The NEFIV AN code was developed primarily for reposi-an auxiliary analysis in Appendix G.) tories hicated in saturated media (e.g.,in bedded salt and basalt). It represents groundwater flow and solute trans-flased on the assumption of vertical flow and the fact that port through a network of flow tubes.The groundwater the repository is envisioned to have a slope similar to the flux and transport within each flow tube are considered to surroundmg geologie unit (see Figure 9.1), the analysis be fully saturated and at steady state, with each steady-was comprised of four separate networks. The network, state vekicity determined by Darcy's 1 aw. In this form, it designed to represent the hydrologic units existing below was not well-suited for the present unsaturated flow eid-a portion of the repository and extending to the water culations, because steady-state saturated conditions are table, is depicted in Figure 9.2. This representation took not anticipated. into account the assumption that one end of the reposi-tory is 299 meters above the water table whereas the other The guidelines for the present phase of this work limited end is 155 meters above the water table and that different the staff to using currently existmg computer codes wher-units exist below these two extremes. Additionally, the ever possible. Rather than develop a new code capable of areal extent of the repository is rather complex (see Fig-simulating the Yucca Mountain case, modifications were ure 9.3). The percentage of waste inventory was parti-made to the NEFIRAN code to facilitate the simulation tioned among the four columns. based on the areal per-of unsaturated flow and transport. First, all coding within 45 NUR EG-1327

9. Analysis and Results

$0ttTA.RC CANYON FAU(T A A U$w H 5 tsoo - USW H4 gr 2$ cit 7 C UE-2sett WELL DesioN 4tpostrony N *M

• JU r-,.m TOPOPAH SPRING MEMBEI 0F PAINTBAUSH TVFf y

W CAttCo ggg .....5..==='

• • ow%PQ ~

..... m f BUltrag

• BER or

/ m

• BER or rtw,,

1. E*8ER or

[ LAVA e447gq, TWr y y ,o ~...._ ****.. co f *, e C 0 500. go10 g VERT 6 CAL EXAGGERATION 3a 0 1 2 WILES i t i I l t 5 0 1 2 3 KILOMETERS HORIZONTAL SCALE ....V....... . WATER TABLE STRATIGRAPHIC DOUNDARY, DASHED WHERE UNCERTAIN UNITS UNCERTAIN 7 [ FAULT USW H4 ORILL HOLE Figure 9.1 Conceptuali/ation of a hydrogeologic cross-section through the unsaturated zone at Yucca Mountain (after DOI!,1988) l NUREG -1327 46

.o A A TSw B m5 3 TSw C CHnv z5= E OE TSw D CHnv hkk a >*B 299 m CHnz CHnv S@C-8,d TSw PPw QG:rJ'E~ CHnz EEs CHnz 155 m E. E E CHnz 0 PPnw PPw ,a e ,8y 8 u ~BFw PPnw ppw hi y j Ei9 h Water Table j adj S 5 ji .- h5 m :3. .= mu DE E' o !? ' E 5' Cn c. TSw - Topopah Springs welded unit e 4 ji CHnv - Calico Hills nonwelded vitric unit g CHnz - Calico Hills nonweided zeolitized &e PPw - Prow Pass welded member or PPnw - Prow Pass nonwelded member k 2c BPu - Bullfrog welded member a ? -h c0 l -.m

9. Analysis and Results I

upn i s..-.tr .,, 3 .e .y.. -; :r- ,3 . A% * : w.,t ' .

• U

. i.':.u.:M.,. o. a,s. 6 p.o* ?.,6. ' 3.V.*.*. o,L* 2

U *

~. y .,.... I, ; v, e+ e t.. s .v>y e. p.. :'( e tc. i.,, s .-g4 L .1 e ,J. y :. ;;o. : t. . n.,..c. e.y'. yt,,',e t;s.or ss,,, R, ;;., c.Op. e&... ,g....

-a%,;

h+W: ~ Q k i 1 \\ ..:"f..d.': v}; %Q>': Q l y' \\ s;,.,$1~.y.j

• f.

a y. w. % s: e i ;.2.:.; ,m:s. ;.;g (e..; ., ; {c.. .r. .N9.g;i s.< -..n %y ; >3... N. i , i. k,ss .'S;.gl%. %.,tg g A bQ ,,.......A,%. m r.. a. .,, ' (.. y ; g...a.. '. l..,.s :.- :g,C... r, :

9

\\;g .*-t.. s .ss,.. .. s ,g. 4v, f-. g j c. a.- 6L. 2 .c s.: n ; i l,,,, a. >g* 15 . vJ..:.rc.,i. M. c. '- %v ?.. s,Q h3.. v,:' w C'.~ 2. V ;%;. c,,

1. J:6';p 8.

t A, i ~..... J,. g .:.,.....,.v ,i.,;..... e,. .s.. ..c e. c,....,..s. .).' ... 1f.... y. i., a. b... h kh. N,8

• g t t WQi:, '

- ' 3 / e.'.[gP.' '

<@. j :. [. 4,! ai g = r?.f.y >;. -> f.. ..;:N,'y y~7,.. d k.\\ : N, ',r J ' ***.* R .:. Pg ~ v.t - v. y:y,,9: g,q h. s'qf:.:~. ;c;vp: s$W UZ D. v-.u.s% o t%n&. R s U A.' 6 ."i w :. g .,: h

v. p,,..

.: a s- . c. ~s's': ns4:,.... 3w w g 3 ..o.*ff p..: ~* ~ m. .b

P:'

K:j.,. g, y: 3 g.hW y(s... l%.. fsc':l usw H sh * '.K.,r.o. V*\\ '. g.): J We. a '..,, *U O'.!Id, 'q"es +7 e 9 I,. 4 * ? ' 4%* a .ude. D e 8 f:9.. 'i$.V}:J ?[ h '.' i WM,fN 0 .'N;.. ~. y(j$ g.,7'4:s4 W a(('#

• p :"0
• ye

)C ? 4 y:c; i k. G4 i gU$

f. M Q ut s s d.7 f41 k.ip.

,c 1 w ....f. a, m. 4 - : Q.~.~. .%w. i s.:..,.e: w : r,::.::iv:.. u v. oe u ~~ e .. ':. c 9 . ::.f t.:.' : p. : 1.1,f.y ps.. 4 5 usw H4 g n. W:i. :::::.$. lf.s.?... m w. .i,. g

s. 9:n

' b"W.&.:.;.b$:,z$ $e% :: s. L st.,.. g.e:. ..n : n .n g.. A E di is..% ; cs .,::: y.; '.:*; M. D"' 1 )CW,3 %g e*G;:< ! FC.j D,.:z.rK:: W. ' - 4

'.4. ' :

.b t s hk. .@!:" & Jg,Mq' J *.'.y'W: E., /, ?A[h.q/ -9 3 G- . u.,,%g .. =.-e. }w:.8. J. O h.e.h.Q1p w, \\ v'h.Y 8 4*n' [ f.DWe

• [: y. ? a.:.2 e <a,kW

%. 4 9 s, g:c.

x... ' sn:.-

s '.b.. f Yi. ^ f. e$. 1 m.],$ h.. 5 .h K,,

"g"p,

., i . - gg Ej u $ 6 /" o "4 ea i ALLWwtuu amo COLLVvivw Quattahany L. NORW al. # AULT C & %Mf C e nt A t R NO W N t] Lava #LCat on pes tant e, cottt D wntat 00NCt A6to. s aLL AmD D A A Om C0w=Tanonie 3:08 .a .t......f..t. m _,,,,.t.,L,... L,. ...... t.t m.C;d 7 T'utt t edCyasf alN TUf f > MIC Ct NS u k$wm 04 INEERRf D. ammowl SHOW O' '. p aint e n.no Tup p 064t CTioN Os mov t utaf. owtaif 0 malat $(Nlt of wCTion 45 Sat 0u ativt O atm-FLQw Tuf f. PRt PapsTenutM fu87 pgaiutits omFT 00unoaav 0 1 2 MILES i I I B 0 1 2 KILOMETEriS Figurc 9.3 Geologic map of Yucca Mountain showing repository drift perimeter (after DOli,1988). NUREG-1327 48

9, Analysis and Results TSw - Topopah Springs welded unit CHnv - Callco Hills nonwelded vitric unit i CHnz - Calico Hills nonwelded zeolitized l / PPw Prow Pass welded member I I I PPow - Prow Pass nonwelded member lB j g BFw - Bullfrog welded member I M% i l I I ID I IC 10Ak l A 40% l l #* \\ l \\ I \\ I/ I a TSw / TSw CHnv TSw y CHnv 299 m Chat CHnv TSw PPw CHnz E* CHnz PPnw N pp, ~BFw PPnw PPw 0 y Water Table Figure 9.4 Representation of the allocation of repository area and radionuclide inventory of the four NiilTR AN simulations. 49 NURI!G-1327

9. Analysis and Results I

i l Table 9.1 Columns Representing the Yucca Mountain Repository Average Matrix Saturated Column A 11 C D Condarti ity Topopah Spring Welded 45 m 60 m 55 m $5 m 0.72 mm/yr Calico Ilills Vitric 100 50 10 0 107 Calico Hills Zeolitic 20 70 120 100 0.54 Prow Pass Welded 34 45 10 0 88 Prow Pass Nonwelded 90 20 0 0 22 Bullfrog Welded 10 0 0 0 118 Fraction of waste -J.4 0.33 0.17 0.10 NEFTRAN that calculated satutr.ed flux through the (b) Infiltration Excccdmg Saturated Ilydraulic flow t ubes was eliminat ed. Instea/, the flow rates through Conductivity the network along the path c' radionuclide migration were calculated from the infatration flux. Second, the in this case, the matrix will be incapab!c of carr,,;ing all the fl w; therefore, a part of the flow will be carned by the NEFTRAN eode was modified to examine predominantly interconnected fractures m the tuff.The matrix portion o. downward bifurcated flow. Flow occurred either througli the flow would have a transport velocity defined by: the matrix or fractures, depen &g on the rate ofinfiltra-tion relative to the saturated hydraulic conductivity of the matrix, ks. Flow through a vertical column would be v = k /n. (9.4) 3 e driven by the infiltration rate. Since the column is one-dimensional, all flux must pass through each layer. If the .Re fmemre ponion of the flow would be; infiltration rate is greater than the matrix-saturated hy. draulic conductivity of an individual unit, then the frac. v' = (1 - k )/ng, (9.5) 3 tion of the infiltration exceeding the saturated conductiv-whue ng s the effective porostly of the fracture. This i ity was assumed to flow in the fractures (in this case, all radionuclide transport occurred through the fractures, parameter should also depend on the infiltration rate. ignoring any radionuclide transport through the matrix). Ilowever, for the present set of calculations, ny will bc The possible subcases for this flow are: taken as a constant,0.0001, representative of a smail value leading to short travel times in fractures (Lin,1986).

• P"" "'

'"'O "E' (a) infiltration Lower Than Saturated Hydraulic times is in the matrix, a relatively small fraction of flow in c Conductivity the fracture may completely dominate the transport for In this case,it was assumed that because of matrix suction, btfurcated flow, Therefore, only the fraction of the infil-water will flow entirely within the matrix, so that the tration carried in the fractures affected radionuclide velocity of a non-sorbing tracer without dispersion will be transport for cases w hen both matrix and fractur es should equal to the infdtration rate I divided by the water content occur.The reasons for this choice are covered in the next 4; i.e. section. vu1/$. (9.2) 9.3.1.3 Implementation of transport phenomena ' thin N13TRAN The water content is related to the unsaturated hydraulic Radionuc. les will be transported in the matrix or in the conductivity through a constitutive relationship. In the fractures, if infiltration exceeds the saturated conductiv-present case, the Brooks-Corey formula was assumed: ity. If flow occurs in the fractures, the matnx and the fractures would be coupled by hydraulic and chemical e (1/k )l/8, (9.3) processes. The effect of matnx diffusion on the transport @=n 3 through the system would depend on the transfer rate of where e is the Brooks-Corey factor for each hydroge-radionuclides between the fractures and the matrix.The ologic unit and n is the saturated effective porosity (Lin, net effect of this transfer can be characteri/cd in three e 1986). ways, depending on the rate: NUREG-1327 50

hk

9. Analysis and Results High transfer rate l'or the preliminary analyses of the Phase 1 effort, the e ffects of matrix diffusion are ignored (the transport strat-At one extreme, transfer between the matrix and frac-egy is expressed by the *No Transfer" case). The reasons tures would be high, leading to the concentration in the for the choice of this ap;' roach are:

fractures being identical to that in the mairix. For plug flow (i.e., no longitudinal dispersion in the direction of 1. The approach is conservative. Transfer from the flow) the rate of radionuclide movement would be the fractures to the matrix would likely retard radio-flux divided by the total water content 4, i.e., the total nuclide transport. T volume of the void water-filled void space: 2. Preliminary sc eening analyses showed that, for cases where fracture flow was important, the great-v=1/t. (9.6) est contribution to dose was likely to come from T transuranic elements, such as plutonium and ameri-Partial tran$r cium.1hese elements are known to have a tendency to form colloids (Thompson,1989). The molecular For the intermediate case, the concentrations of the ma-dtffusion coefficient of colloids is much less than that trix and fracture wotdd be coupled by a process allowing for dissolved molecules and ions, so matrix dtffusion the transfer of radionucl!Jes from the higher to lower may not be effective (colloid transport is not mod-potential; i.e., if the concentration of radionuclides in the cled explicitly in the present exercise, however). fracture was greater than in the matrix, there would be transport of the radionuclide into the matrix by molecular 3. Fractum cmtings would lead to a diminished effec-diffusion. This phenomenon is generally called matrix tiveness of both the diffusive transfer of radi-diffusion. onuclides and water flow from the fractures to the matnx (Carlos,1985). By judiciotis choice of parameters, the NEFIRAN code can be made to approximate matrix diffusion using a sim. lackmg experimental data on the actual magnitude and ple two-zone model (Van Genuchten and Wierenga, rates of matnx diffusion at Yucca Mountain, this process 1976).The staff assumed that the water contained in the was not included in this initial demonstration. This was matrix is essentially immobile, because fracture flow is so believed to be a conservative assumption. much faster.The model accounts for the loss of the radi-onuclide from the mobile fluid to the immobile fluid by 9.3.1.4 Spatial variability of flow and transport transfer across a boundary between the fracture and ma. parameters trix. The concentration in the mainx and fracture is as' To maintain a high degree of efficiency in the Monte sumed to be untform, and does not vary with distance Carlo analyses with NIil:TRAN, the complicated, spa-from the interface.The model is only a rough approxima-tially varying repository was represented as four vertical tion of true matrix diffusion, because it ignores concentra-columns, each with a small number of hydrogeologic units tion gradients lateral to the direction 01 flow, it may through which all the radionuclides must pass. Existing capture salient features of matrix diffusion for our pres-data on tuff layers at the Yucca Mountain site indicate ent purposes, however, and maintains the high efficiency that there are considerable variations in the material of the code. properties. Available data do not support long correlation lengths for the transport parameters at the Yucca Moun-Transport due to matrix diffusion is proportional to a tain site. The data in rnany cases suggest small spa,ial coefficient B. The NEFI'RAN manual suggests that B correlation, or none at all, on the scale for which they can be approximated from the average fracture spacing, a, were collected (see Appendix F). Using constant values of and effective diffusion coefficient, D': transport parameters in the models therefore would be inapprcpriate. Assuming perfect spatial correlation B = 2D'/(a!2)2 (93) vithin a unit could lead to a false conclusion that condi-tions leading to short travel time would apply over the The model does not account for the additiona resistance whole unit in actuality, short travel time might apply only that could be caused by the presence of surface coatings to a small segment of the cohimn and bc countered by the on the fracture. Since fracture coatings at:: common, the presence of a barrier elsewhere in the column.1his ap-coefficient B should be reduced to take into account the plies to a one-dimensional analysis only in which the flow reduction in transfer caused by these barriers. must pass through each segment in series. No tran$r Previous studies have recognized the importance of spa-tial correlation in the assessment of arrival-time distrihu-At the other extreme, no coupling, the transport in the tions. I in and Tierney (1986) estimated the arrival-time matrix and fracture pathways would be separate. distnbution for releases at the Yucca Mountain site by 51 NUREG-1327

- - ~ ~ -

9. Anidysis and Results calculat ng the travel time of particles confined to a series cross-sectional area, and the retardation coefficients for i

of one-dimensional columns that represented the path-each of the radionuclides considered in the presentimaly-ways from the repository to the water table. For each -sis, column, they varied the correlation length by changing the spatial step size, but keeping the hydraulic properties As described in Section 9.3.1.2. groundwater transport constant within a given step.They found from this analysis was assumed to be either entirely in the matrix.at low that longer step sizes lead to a wider arrival-time distribu-rates ofinfiltration, or entirely in the fractures at infiltra-tion: tion rates that exceed kg Since vertical flow under un-saturated conditions was assumed, the primary factor for "The implicit vertical correlation length (10 feet) determining whether the flow in the present analysis is in . of the baseline case is much less that the thick. the matrix or the fracture was the saturated hydraulic ness of any of the hydrogeologic units. This re. conductivity, ks. If infiltration exceeded k, then the 3 sults in a large number of independent random excess flowed in the fractures. variables (travel times through each of the calcu-lational elements) which are added together to Geostatistical analyses of the k3 data presented in Ap-obtain a travel time through a column. Conse. pendix F indicate that th:re was no apparent spatialcor-quently there is a low probability that fracture relation beyond about 10 meters separation distance, the flow will occur through a large number of ele. smallest interval evaluated. As longer correlation lengths ments in any single column from the disturbed are more conservative, ks was assumed to be completely zone to water table..., correlated at a distance of le meters. Each tube i_n the column was represented by a connected series of sub- .tenger correlation lengths affect the tiavel t bes, each oflength L Each sub-tube had uniform prop-time distribution, especially at the tail ends of the C"Ies, but was uncorrelated to the next subtube m the distribution, because of the increasing probabil-senes.The value of ks for each sub-tube was picked by ity of fracture flow through a significant number t he Monte Carlo method from the lognormal distnbution of elements that make up each of the columns,. derived from the available core data presented m Table 9.5. These results indicate high sensitivity of the travel time distribution to the as yet undeter-mined correlation length for velocity in each hy-The ana1ysis was based on the assumption that the Hux of m traung water passed through each of the sub-tubes. drogeologic unit. Generally the sensitivity of the travel times to the correlation lengths suggest Phe travel time across each sub-tube depended on how prudent it is to perform a carefully designed whether the flux was greater or less than ks: testing program for determining the correlation lent h <'aii key parameters influencing flow ve-I I I * ,"S ' locines " At, = ng Al/(I-k ) @) 3 Long correlation lengths led to an overly broad distribu- ~ i dj f i N) tion for arrival time, with some very short travel times at Atj = Atj R the tail of the dis;ributioni At the other extreme <the - assumption of zero correlation length led to the conclu. - for I < ks, sion that there is onlya single gmundwater travel time per column. The aforementioned conclusions apply equally At = + Al/l (9.10) well to radionuclide transport; and therefore, the deter-mination of spatial correlation scales, especially for ks, At. - = At; Rdi, (9.1l) was important to this demonstration. 11 where At, - the water travel time for subtube i 9.3,1.5 ' Effective values of flow and transport coefficients Atjg = the travel time for radionuclide j in subtube i The NEFTRAN code simulates flow and transport nf = the effective porosity of the fractures through a network of connected tubes. For the present (taken to be 0.0001) case,the flow and transport model was represented by up to six tubes in series, with each tube representing a major the water content of the matrix of subtube i hydrogeologic unit; for example, the Topopah Spring welded unit. Within each tube, the unit was represented Al = the length of the subtubes by coefficients expressing its physical properties for flow - and transport, namely hydraulic conductwity, porosity. I-the infiltration rate . NUREG-1327 52

9. Analysis and Results R.-

the matrix retardation coefficient for quency of dnlling), but all other parametric values were dI obtained by reading the NiiFIR AN input files. radionuclide j dj,f = the fracture retardation coefficient for 9,4 Faranleters R radionuclide j. .Ihts section presents the ranges of parameters used in the liquid and direct pathways. Parameter values used for the In this analysis, flow was considered to be either totally g s pathway analysts are presen>d m Appendix D. matrix or totally fracture flow for each sub-tube of lengig Ranges were used by the 1J IS program to generate data Al. Even though there will be matrix flow in parallel with for the source term and transport programs. the fracture flow, in practice the fracture transport prop-erties generally overwhelm the contributions of the ma-trix flow and can therefore be safely left out of the analy. 9.4.1 Liquid Pathway N Using NEFIRAN to imulate the liquid pathway requires the assignment of the following parameters: Then, the individual travel times and radionuclide travel t mes were summed to determine effective values of po-i saturated conductivity f rosity, @e, and retardation coefficients, Rtubes representing the hydrogeologic unitsf or the main g P * '"Y volumetric flux N retardation coeffi,ients ! I At5 solubility limits i=1 NAl solubility of uranium matrix waste-package lifetime N . I At j water-contact fraction i dej " (913) e R dispersivity correlation length for hydraulic properties . I Ati e i=1 For the liquid pathway analysis, the geologie medium was where N is the number of sub-tubes. represented as a series of four vertical columns, each wit h up to six hydrologic units through which all of the radi-There are two levtis of sampling: onuclides must pass. Each segment represented a single [ hydrogeologic unit. Subroutine GETRV in the 1. With.m each sub-tube, a Monte Car'o method was NEFIHAN program contained all the definitions of uscJ to sample for the values of ks, from a lognor-source term and transport parameters necessary to make mal dstribution, to determme tube-averaged prop-the code emulate the unsaturated flow and transport erties of effective porosny and retardation coefft-model. eients, and inputs to NEFIR AN were generated using the 1JIS pro-2. From reabzation to realization, the IliS method gram, which selects random values from the input pa-was used to sample the mean and standard deviation rameter ranges. Several known or suspected correlations of thelogsof ks and the sub-tube length I., to repre-are provided in Table 9.2. Formal inclusion of correla-sent the uncertainty in their values from borehole to tions between variables should be performed in subse-borehole-quent phases of this study. 93.2 Gas Pathway 9.4.2 Sampling Parameters for NEFTRAIN Analysis The discussion of this pathway is presented in Appendtx D. The parameters necessary for this preliminary analysis of the Yucca Mountain repository came from a variety of 9.33 Direct release (Drilling) Pathway sources. but primarily, from published DOE reports, in-cluding previous performance assessments for the Yucca The analysis for the direct-release pathway (via explora-Mottatain and other repositories. Many of these data are tory drilling) is presented in AppenP H. The drilling hiinly uncertain; nevertheless. they represented the hest analysis used parameters specific to orilling (i.e., fre-data available for the demonstration. Sensitivity analyses j l l 53 NURl!G-1327

9 Analysis and Results Table 9.2 For the initial phar,e of this study, it was assumed that Examples of Known and Suspected Correlations the fuel solubility was fixed and not a function of time and temperature. Refinements to account for time-0 fletardation coefficients for similar elements dependent oxidation state and temperature may be incor-o Solubilities of similar elements p rated into the modelin later phases of this effort. A o Solubilities with temperature possible subj,ect for further study would be the potential isolation afforded the waste by the drying out of the rock. o Temperature of canisters with engineered barrier failure time 'lhe waste-package failure for the liquid pathway analyses o Uranium. matrix decomposition (i.e., oxidation, was chosen as normally distributed, with a 0.001-to spallation, dissolution) with waste-package failure 0.999-fractile range for 100 and 1000 years, respectively, time For the gas pathway, two distributions were chosen to o Irach rate with infiltration rate and fraction of demonstrate the sensitivity of the release of C-14 to watercontacting waste waste 9 ckage lifetime (see Appendix D). Infiltration rate with fraction of water contacting o waste form 9.4.23 Solubility of uranium dioxide matrix Assuming the canisters and cladding have failed and water penetrates the canister, the bulk of the radio-performed after the calculations pointed out areas where nuclide release is likely to be from the dissolution of the improvement in data would be imponant in narrowmg uranium dioxide waste matnxJlhe solubility of the waste the ranges of calculated performance. Table 93 shows will be controlled by several factors. Among the more the input ranges and distnbutions of parameters for the important factors will be the oxidation state of the fu'1 NEFFRAN and other analyses, as generated by the 1.11S which is, in turn, a function of temperature, oxygen avail-program LHSVAX. ne following sections describe the ability, and time, it was assumed that the dissolution rate basis for choosing the nmges appearing in Table 93. of the waste was controlled by the rate of disintegration of the uranium dioxide matnx, as characterized by a solubil-ity limit (the disintegration of the fuel matrix may not 9.4.2,1

w. te package lifet.ime as actually be limited by solubility, but by the rate of oxida-No readily availaNe models were accessible to assist in "A

'I E""I *' I II the choice of the waste-package lifetime in the unsatu-

• "^'i" #E "

1. • ~

I"" ## "" "* "" rated zone.The NEFTR AN code is simplistic and able to accept only a single value of lifetime for each run, even U" *

• F E'"*

though it is likely that waste-package failure would occur r in a highly distributed manner. 9.4.23 Dispersivity The disper.stvity is a measure of the spatial variance in the Waste-package lifetime will affect the source term in transport speed, particularly that caused by variability in several ways. First, the package must fail for anything to uterial properties, which causes the amval-time distri-be released at all (although failure does not alone imply bution to spread. It is not likely to be an important consid-tha' there will be contact between the waste and the cration in most analyses for cumulative releases. He water). Second, if the package fails in an essentially dry dispersivity was chosen to be normally distributed, be-environment, oxygen from the unsaturated zone will en-tween 0.1 and 10 feet, for the 0 001 and 0.999 fractiles, ter, which might allow oxidation of the UO to proceed respectively. 2 for a fraction of the fuel rods that have defects, The Inore-oxidized uranium would have increased solubility 9.4.2.4 Infiltration rate over tha less-oxidized form (Grambow,1989). Further-more, oxidation could cause an increase in volume of the One of the key variables in the analysis was the rate of pellets, causing splitting of the cladding and spallation of infiltration, which is the main influence on the speed of the pellets and thus possibly increasing sudace area. Oxi-water movement in the vertical column, as well as the dation might also take place in some of the unfailed canis-amount of water coming into contact with the waste. ters, because of the presence of small amounts of oxygen, or the disso ttion of water caused by ionizing radiation (a) Base-case InfUtration (Reed, et al.,1987). This radiation could form hydrogen peroxide or nitric oxide, which are powerful oxidantsflhe At this time, there are no direct measurements of infiltra-time-to-failure of the canister would impact directly on tion at Yucca Mountam. Estimates of present day infil. fuel oxidation, because the reactions are sensitive to tem-tration rates have been calculated from: (1) heat flow perature and radiation, both of w hich decrease with time. measurements. (2) precipitation and elevation data and NUREG-1327 54

9. Anuysis and Results Table 9.3 Input to Latin flypertube Sampling Program Distribution Range Label Normal 100

'lT) 1000 Waste-package lifetime, years Uniform 1.0E-04 TO 1.0E-03 Solubility of matrix, gm/gm water Normal 0.10 'IO 10 Dispersivity, ft infiltration Rare,[t lday 3 Uniform 0.50E + 02 TO 0.25E + 04 Base-case scenario Uniform 0.25E + 04 TO 0.50E + 04 Pluvial scenario Uniform 1.00E-04 TO 1.00E-02 Fraction of water contact Porosity c; Matrix Uniform 0.10 TO 0.18 TSw Uniform 0.04 TO 0.14 CHv Uniform 0.28 TO 0.36 ClIz Uniform 0.26 TO 0.31 PPw Uniform 0.10 TO 0.18 PPnw Uniform 0.13 TO 0.28 BFw Log ks, mmlyr Uniform -0.50 TO 0.25 TSw Uniform -1.40 TO 0.50 CHv Uniform -0.70 TO 1.20 CHz Uniform 1.40 TO 2.20 PPw Uniform 1.40 TO 2.20 PPnw Uniform 1.50 TO 2.50 llFw Standard Deviation oflog ks, mmlyr Uniform 0.60 TO 0.75 TSw Uniform 0.70 TO 1.00 CHv Uniform 0.80 TO 1.00 CHz Uniform 0.40 TO 0.60 PPw Uniform 0.40 TO 0.60 PPnw Uniform 0.50 TO 0.70 BFw 55 NUREG-1327

. - ~

9. Anolysis ond Results Table 9.3 Input ts atin Hpercube Sampling Program (contmued)

Retardation Coefficientsfor Ma:rix Uniform 100 TO 1.0E + 04 Am - Uniform -3000 'ID 3.011 + 04 Cm Uniform 3 7D 2000 Ni Uniform 5 TO 100 Np Uniform 10 TO 100 Pu ~ Uniform - 0.1E + 04 TO 3.5E + 04 Ra Uniform - 0.2E + 04.TO 0.4 E + 04 Sn Uniform 5 TO 10 Tc Uniform 200 TO 0.5E + 04 Th Uniform 5 TO 30 U Uniform 1.0 TO 1.0E + 04 Zr Uniform 20.0 TO 0.1E + 04 Pb Solubilities, gn/pn water Uniform 2.0E-10 TO 2.0E-0 / Am Uniform 1.0E-09 TO 2.0E-07 Cm Uniform 2.0E-04 'TO 1.0E-03 Ni. Uniform 2.0E-05 TO 3.0E-04 Np .. Uniform 5.0E-08. TO 3.0E-06 Pu Uniform 1.0E-08 TO 1.0E-07 Rn Uniform 5.0E-12 TO 5.0E-10 Sn Uniform 0.5 TO 1.0 Tc Uniform 1.0E-11 TO 5.0E-10 Th Uniform. 2.0E-11 TO IJ2E-10 Zr Uniform 1.0E-04 TO 2.0E-03 Pb . Uniform 20.0 TO 50.0' Correlation length, ft. ' NUREG-1327 56

9. Analysis and 1(esults (3) hydrelogic parameters rneasured from core and within flow around the canisters because of the matrix suction of site boreholes. Table 9.4 contains a summary of infiltm-the unsaturated rock, so this f@tre derived from this tior, estimates cited in the literaturelstimates ofinfiltra-Simple approach does not aiptue the true nature of tion rates range from less than 0.1 mm/yr to 10 mmlyt.

water contact. Mor.t of the previous IX)!! analyses have emphiyed infil-DOli plans to emplace the canisters in the host rock in a tration rates in the range of 0.1 and 0.5 mm/) car. Ilow-manner that it considers would reduce the likelihood of ever, because of the considerable uncertainty in the esti-water coming into contact with the waste "these precau-mates presented in Table 9.4, a considerably wider range tions include vertic:d storage and an air gap between the of htfiltration rates than previous dol! analyses was cho-canister W the rock walls l'urthermore, DO!! considers - sen. For the base case scenario, tir infiltration rate was that the hut eenerated by the waste may create a signifi-considered to be uniformly distributed between 0.103 and cant rone of dry rock around the canisters, thus i=oLQg 5.14 mmlycar (500 to 2500 cubic feet per day over the them until cmling of Ge rock at a later time allows water total repository area). 'Ihis range was considered to be a to rewet the rock (SCP, Section 8.3.5.9). Water may till sufficient representation of available data for the pur-come into contact wnh the canisters by other mecha. poses of this demonstration. nisms, including: e (b) I'lmiul Sccwio 1. Infiltrating water flowing through fractures and CzarnecH $85) estimated infi'tration for a future plu-vial climate secnario to calculate the potential rise in the 2. loss of the air gap caused by failure of the emplace-water table. listimates of future precipitation were based ment holes through mechanical and thermal on descriptions of paleoclimates, which indicated that stresses, or mineral and sediment infilling annual precipitation 12,000 to 9,000 years before present, in the modeled

• a, may have been 100 percent greater than modern.

$nual precipitation.*lhis 100-percent ,Iwo additional and potentially important sources of increase with

t to modern-day precipitation was water c uld be (1)1ateral inflows from areas of puched auumed tobe ti abable maximumincreasein the next water, and (2) h., quid water circulation caused by heat.

10.000 years. Cra ecki doubled the rainfall estimate of driven evaporation and condensation. Iateral inflows 1(ush (1970), and then multiplied by the percentage of were assumed unlikely to affect more than a few of the precipitation occurring as recharge that is associated with canisters, since the water necessary for this phenomenon to be viable would be diverted from the vertical infiltra-that higher precipitation range. lie assumed that the in-uon available for all canisters if such a diversion were creased flux across the northern boundary of the mod.: led area occurred because of the increased precipitation in possible, some canisters might receive a greater share of recharge areas to the north. Vertical infiltration into the overallinfiltration at the expense of the remaining Fortymile Wash increased because of surface-water run-canisters being exposed to less water. la, quid water circu-off from its drainage basin. Czarnecki calculated that labon caused by heat is potential!y tmportant, and is dis-( increased recharge from a 100 percent increase in annual cussed further in Section 5 of this report," Source 1,crm." precipitation w-I be 13.7 times greater than estimates of modern-day.rcharge, corresponding to about 7 mm/ For this preliminary analysis, the water contact fraction year infiltration. lie also predicted a rise in the water was chosen to be 0.002 to 0.01, based on the assumed table of 130 meters. wetting of a small fraction of the canisters. i or the purpose of this study, the range of infiltration for 9.4.2.6 Saturated hydraulic conductivity the pluvial scenano was chosen to be 5 to 10 mm! year, with an increase in the water table height of 100 meters Water flow in the unsaturated fractured rock could pro-cced through toth the matrix of the ock at low rates of (- 9.4.2.5 Fraction of water contacting waste infiltration or through the fractures and the matrix at higher rates of infiltration. *lhe switchover frem matrix The ratio of water infiltrating the site to the water actually flow to flow in the fractures may be related to the satu-coming into contact with the wast _c was characterited as a rated hydraulic wnductivity of the rock matrix (ks). Sta-constant. Simple calculations were perfo med to estimate tistical evaluation of the data, presented in Append ~ F, the fraction of the waste canisten; exposed to purely verti-indicated that ks was lognormally distributed. Table 9.5 cal infiltration by taking the ratio of the cross sectional summarizes the available data on saturated hydraulic con-area of the canisters to the total area of land surface ductivities from rock cores at the Yucca Mountain site in - projected by the repository.This ratio was approximately terms of log meara and standard deviations, where there equal to 0.00078. In most cases, infiltrating water could are sufficient data available. 57 NUltl!O-1327

[ 2- .c .. ~g !~ C g 1. 'T i g E M~ w w c. t

=

0 e . Table 9.4 - Infiltration Estimates II t L f M Estimate Incetion Methodology Source i l .4 mm/yr Yucca Mt. precipitation and Rice, 1984 } i elevation data Rush, 1970 i 1-10 mm/yr. Yucca Mt.. drill hol6 thermal data Sass, 1982 ,e l t 2 mm/yr Yucca Flat hydrogeologic parameters 'Ninograd, 1981 j O.5 mm/yr . Yucca Mt. precip. and elevation ,czarnecki, 1985 ,f data t f <0.5 mm/yr-Yucca Mt. core and insitu Wilson, 1985 hydrogeologic parameters 0.5 mm/yr Yucca Mt. maximum for matrix k, data Sinnock,.1984 I O.1 - 0.5 mm/yr USW UZ core and insitu Monterar, 1985 hydrogeologic parameters i 10-I - ' O.2 mm/yr USW UZ-1 core and USW UZ-2 insitu Montezar, 1984 f hydrogeologic ' parameters L i i I i I t i h' t l I t

9. Antlysis and itesults Table 95 Table 9.6 Mean and Standard Deslation (5.D.) ofleg ks Mean Porosity for it)drogt ologic Units Unit Mean of log k:

S.D. of leg ks Unit Arithmetic Mean Porosity m m/yr mm/)r llFnw 020 IlFnw 2.22 0.22 1.38 0.25 1.71 0.59 IlFw 0.13 0.28 IlFw 2.08 Clin 0.36 0.20 Cilny -1.32 0.28 0.47 0.34 0.07 0.29 PPn 0.29 Clinz 1.16 I,P* -0.65 OK 0.26 PP 1.44 2py T5w 0.11 0.13 TSw 0.22 0.72 0.10 -0.4:3 0.61 0.I1 0.I8 9.4.2.9 Ilrooks.Corey coemclents 9.4.2.7 Spatial correlation of saturated hydraulle conducthity 'lhe llrooks.Corey coefficients are used to determine the fraction of saturation for a given infiltration rate, as de-Geostatistical analyses of the ks data, presented in Ap. senhed in Section 9.3. 'lhe values used in the demonstra-pend v F, indicated that there was no apparent spatial tion were taken from IJn and Tierney (1986) and are correlation of the core data on ks atxwe the mimmum presented in Table 9.7. separation distance of 10 rneters used in the analyses. Since larger correlation scales are conservative, a correla. Table 9.7 tion scale between 20 and 50 feet was assumed. 'there Itrooks.Corey Coemtlents were insufficient data to determine the distribution of the Unit Coemclent mean and standard deviation of ks, so it was assumed that the mean and the st<mdard deviation # log ks were uni-TSw 5.9 formly distributed. The mean, standard deviation and Cliny 4.2 correlationlengthof ks were used to choose representh-Clinz 7'o tive hydraulic coefficients for each hydrogeologic unit as pp, 4O described in Section 9.3.1. PPn 52 IlFw 4.6 llFn 5.2 9.4.2.H Porosity 'lhere are probably more porosity data available from 9.4.2.10 11etardation cocmclents core taken at the Yucca Mountam site than for any other type of data used in this study. As used in this study, water Values of retardation coefficients for the matrix wcre velocity and radionuclide transport speed in the matrix chosen to reflect reported values for batch and column were tied closely to the average value of peosity for the tests performed. For the key radionuclides pluf anium and columns.The porosity ranges were chosen from available americium, values were chosen a. the low end of the data ascraged over each unit. There were insufficient range to account partially for data that indicate that these data to determine the distributions of the average proper-substances do not behave simply, tend to form colloids, ties, so the averages were assumed to be uniformly distrib-and may be difficult to predict under repository condi-uted. llepresentative values of porosity for cath hydroge-tions (ihompson,1989).110 wever,it shwld be noted that ologie unit were sampled from the distribution of much of the data in column experiments that indicated arithmetic mean porosities shown in Table 9.6. Iow retardation for some elements was collected for flow 59 NUltliG-1327

9. Analysis and itesults i

rates 3 to 4 orders of frago w ( cater than those used in 9.4.2.11 Solubilities this demonstration, anc,w w e may be pessimistic. Furthermore, the sesub . m.Jivity studies presented Values for the solubilities of radmnuchdes were taken in Section 9.5 indicate tk u the present study, retarda-primanly from DOli references, including several pre-tion coefficients for plutonium and americium are rela-hminary perfonnance assessments.They reflect those re-tively unimportant, indicating that factors such as low ported in previous performance auessments from Yucca solubility and long half life may be more important than Mountain (l.in,1986). retardation for these nuclides. Values used in this study are typical of thcsc used previously in Yucca Mountain 9.4.3 Direct.Helease (Drilling)l'atliway performance studies (ljn,1986). Althoua.h most of the retardation coefficients are sampled ty 1.11S from the The drilling program was developed to calculate the con-distributions presented in Table 9.3, the retardation coef. sequences from an expected number of borcholes inter. fients for a few of the elements were taken to be con. cepting the repository (see Appendix 11). He following stants. %ese retardation coefficients are 1.0 for iodine, values were needed; drilling rate,4/ and number of 10,000 for cesicm,1,000 for strontium, and 1.0 for ea rlxm. waste packages, arca of epository, time of drilling, and radius of the lorehole. Additionally, the following data from the liquid pathway were used: time of waste package lletardation coefficients for the fractures were taken failure, volumetric flux, water contact fraction, and solo-from the study by Lin (1986), and are orders of magnit ude bility limits (these values wcre discussed in the previous smaller than the matrix retardation coefficients. The val-section and will not be discuued here). ucs of retardation coefficients for fractures were not sam-pled, but remained fixed for all realitations. De values-liased on ecmceptual repository designs, the dimensions used are given in Table 9.8. I or both the matrix and used for the repository system were: area of repository - fractures, no distinction was made for the variation in 5.1 square km. number of waste packages - 18.tX)0, retardation between different hydrogeologie units, liow. borehole radius - 6 cm, waste. package radius - 0.34 rn, ever, those uruts that have low values of saturated hydrau-and waste package length - 4.8 m. He time for com. lic conductivity will tend to have lower effective retarda-mencement of dnthng was sct to a arbitrary value of 100 tion coefficient values because of the greater proportion years, and the drilling rate to.0003 drillings per square km of the flow to be expected in the fracture mne (as calcu. per year, based on !!PA average drilling rates (liPA, lated by the procedure presented in Section 9.3.1.5). 1985). Tubie9.8 9.5 Sensitivities and Uncertainties for itetardation Coeffidents for l'racture$ Liquid-Patliway Analysis Element Ildt 9.5.1 Introduction Curium 1.4 This section covers the sensitivity and uncertainty analy. Plutonium 1.1 vs of the liquid-pathway calculations on a scenario by scenario lusts. It does not cover either the drilling or gas Unu.ium 1.0 pathway analyses, he CCDFs presented in this section Americium 1.4 for the base-case and pluvial scenarios take into account Neptunium 1.0 the uncertainty in the values of the coefficients for each scenano, but not the scenario probabilities. The sensitiv-Horium 3, sty to variations in parameters using rank regression and lladium 2.8 ad hoc variations of single parameters (including those I cad 1.0 parameters relating to the NitC guidelines of 10 CFil Cesium 100 60.113) are also presented. Total system results, which lodine 1.0 insporate the probabilities of the scenarios, are covered in Section 4.6. l'ormal sensitivity and uncenainty analyses , fin 13 on these total system results were not performed. Technetium 1.1 zirconium 2.0 9.5.2 Statistical Uncertainly Analysis Strontiam 10.0 An important part of conductmg a performance assess-Nickel 1.2 ment for an FII.W repository is quantifying ihe uncertain-Cartmn 1.0 ties associated with the probabihties of occurrence of credible scenarios and those anociated with the offsite s NUl(!!G-1327 60

9. Analph and Resuht and onsite consequences (both radiological and nonradi-

'lhis latter state-of knowledge uncertainty may be subdi. ological). vided further into model and parameter uncertainty. Pa-rameter uncertainty is because of insufficient knowledge Many risk and environmental impact assessments apply about what the input to the code should be. Modeling single or best-estimate values for model parameters and uncertainty is because of simplifying assumptions and the assert that these valuations are reasonable and conserva-fact that the models used may not accurately model the tive (i.e., lead to overpredictions)without quantifying the true physical process. 'this stub deals primarily with pa-degree of conservatism inherent in the anessments. A rameter uncertainty. 1 variety of techniques is available, to quantify the uncer-h shown in Table 9.9. a eet of key parameters in the tainty in complex models for assessing radiologicalimpact redel under study must be identified first, For each cho-on man and the environment that may include nonlin. caritiesand time varyinE phenomena.'thescinclude: the sen variabic, a set of quantitattve information is devel. Monte Carlo mcthod (llelton,1961): fractional factorial oped regarding the range of variation and probabihty distribution, as well as correlations among the variables. design (Cochran, 1963); 1.11S (Cranwell and llelton, 1981; Iman and Conover,1979; McKay, et al.,1979); response surface (Meyers,1971); differential sensitivity analysis (e.g., adjoint (llaybutt et al.,1981, Oblow,1978, Steps to Pnform Uncert inty and Senstthity Analysis Cacuct, et al.,1980)); and I ast Probabilistic Perfortnance Assessment (CNWRA,1988) methodologies. A pre-1. Specify Maximurn-Minimum Ranges of Probabili-ferred technical approach would be flexible, economical ties to use, easy to implement, provide a capability to estimate 2. Specify Correlation Matrix an output distribution function, and rank input variables by different criteria. 3. Run latin Ilypercube Sampling Code 4. Run Source Term and flow and Transport Models 9.5.2.1 Latin liypercube Sampling (LilS) 5. htatistical Analysis (fitting distributions; repression In this study, the LilS scheme was chosen to be used on analysis; graphical display and analysis) the flow and transport rmxici in the performance assess-ment of the llLW repository,The advantages and proper-ties of the LilS techniques are: For this demonstration, no correlations between input variables were used. The data input to the 1.1IS program The full range of each input variable is sampled, and is provided in Table 9.3, which shows the distribution and o correlation coefficients between all pair-wise input range of input for each variable The basis for choosing variables can be specified. these inputs was discussed in Section 9.4. 'lhis informa-tion was used as input to the LilS code (Iman and Shor-It provides unbia,cd estimates of the parameters tencarrier,1984;1 man and Davenport,1984). LilS is used (means and variances) of cumulative distribution to generate what is called a design matrix. Specifically, if functions and means for model output, under mod-N computer runs are to be made with k parameters under erate assumptions. study, the design matrix has dtmensons N x k. Each row of The LHS method is a member of the class of sampling this matrix contains the input valuations of each of the techniques that includes Monte Carlo and stratified ran. chosen k parameters (independent variables) for the N dom sampling. Several risk assessments for nuclear waste computer runs.The sample size N is specific to the prob-repositories (Campbell, et al.,1979) have applied LilS lem being investigated (Iman, 1980). Appendix D techniques. Furthermore, LilS has been applied to the presents a sensitivity study on the sample size for one rmdel for atmospheric transport of reactor accident con-scenario. sequences and recently used for the severe reactor acci-dent calculations in NUREG-1150 (NRC,1989). 9.5.3 Ad lloc Sensitivities Different types of uncenainty associated with the model-In this section, results of the NEITRAN runs for the ing of physiochemical processes can be distinguished-in base-case and pluvial liquid-pathway scenarios are pre-particular: sented, with the intent of demonstrating the effects of individual variables on the resultant cumulative radio-nuclide releases to the accessible environment. 'Ihe statistical uncertainty due to inherent random nature of the processes, and De N ElTR AN computer code, as modified for this dem-The state uf (perhaps " lack of') knowledge uncer-onstration, calculated cumulative releases, for the base-tainty. case scenario, over either 10.000 years or 100.000 years, 61 NURE41327

9. Analysis and 11esults and for the pluvial scenario, over 10,000 years. For each potential for spurious correlations betw een parameters is simulation, a list of 47 variables was generated. using much greater.

LilS the list of variables for each simulation is called a " vector." The input constants tanges and distributions for generating the vectors are presented in,l'abic 9.3. g g,ggggg. Analysis Using Regressiott As shown in Table 9.9, following the execution of the !J IS 9.5.3.1 Senstthity to infiltration program and the source term and flow and transport Figures 9.5 and 9.6 show the resultant conditional CCDFs models, the neu step involves performing a sensitivity for the base-case scenario at 10.000 and 100.000 years, an lysis on the calculated results.*lhe aim of this analysis respectively. Also plotted on these figures are CCDFs is to determine and quantify the telative contributions of composed only from vectors having infiltration rates less a particular variable toward the output variability. Sensi-than limits set at 2.0 or 1.0 mm/yr, to demonstrate the livity analyses can be very fruitful in preliminary studies particular significance of this parameter to repository such as this one, since sensitivity analyses can help to performance. !dentify which parameters and models should be refined m future studies. In addition, sensithity analyses may 'the great sensithity to infiltration rate can be partially allow the analyst to check his intuition about the impor-crplained by the next two figures. Figures 9.7 and 9.8 tance of the parameters and phenomena of the model, show the CCDFs for the base case scenario at 10,000 and 100.000 years, respectively, comparing the contribution Sensitivity can be determined by perforrning step-wise of Column D to the contribution from all four columns-linear regression analyses on either the raw results of the Column D contains just 10 percent of the waste, but has model analysis (i.e., the !!PA ratios) or the ranks of the the shortest pathway to the water table. In addition, Col. raw results (i.e., replacing the " raw" data values with their umn D contains just two units; the Topopah Spring ranks). Ranks may be preferred when highly nonlinear welded and Calico llills zeolitic. lloth of these units have relationships are present between the model outputs and relatively low saturated hydraulic conductivities, which inputs, but the correlations obtained have less signifi-would make them prone to fracture flow for higher infil, cance than those obtained from the raw data. Iloth tration rates. Fracture flow can lead to short radionuclide graphical analyses and statistical-distribution fitting pro-travel times aloeg the liquid pathway, because the overall cedures may also be useful in identifying patterns in the radionuclide retardation is low. Figure 9.7 shows the ef. data. 'Ihe present report shows only the regret,sion analy. feet most dramatically, as vinually all the contribution to ses on raw results; i.e., the !!PA release ratios, the high-consequence portion of the curve is caused by Coluna D, alone: retarded radionuclides have not yet 'lhe sensitivity of the cumulative release was analyred fer staned to arrive from the other columns. Travel times several cases using a modified version of the 5111PWISI! through the other three columns would be too long to program from SNI (Iman,1980). *lhe Kil!PWISli pro-contribute rnuch to the CCDF within 10,000 years. Fig. gram was modified to read the data fde of input vectors ure 9.8 shows that more of the contribution to the CCDF generated by the I llS procedure and the combined re-comes from the other three columns over the sults for Columns A through D generated by NIFIR AN 100.000-year period, because the long-lived radionuclides for those input vectors. 'the regression coefficients are begin to arrive,' presented in Table 9.10 for both the base-case and pluvial scenarios. There were 500 vectors for the base-case see-Figure 9,9 shows the CCDP for the pluvial scenario, in nario, but because of excessively long computer-run this case, the water table is shallower and infdtration rates times, only 98 vectors for the pluvial scenario.'lhe paucity are higher than for the base-case scenario, so radio-of vectors led to more equivocal results for the pluvial nuclide travel times are shorter for all columns. A rela. scenario. Only the most significant regression coefficients tively larger portion of the cumulative radionuclide re, or, in some cases, those regression coefficients pointing to leases comes from Columns A, II, and C than in the an apparent lack of sensitivity to particular lurameters. base-case scenario. 'ihese two scenarios are not directly comparable, however, because long computer-run times 'lhe sensitivity analyses proved to be very revealing, both led to the necessity of reducing the number of vectors for the sensitivities to some parameters and apparent iack l from 500 to 98 for the pluvial scenario. It should also be of sensitivities to others. The parameters consistently in- ~ pointed out that the 98 vectors for the pinvial case were fluencing to the !!PA ratio were: contact fraction,infiltra-generated from a truncated run originally intended to tion rate, solulyility of the matrix, and saturated hydraulic contain 200 vectors. 'the desirable propeny of statistical conductivity of the Calico Ilills sitric unit. Of these, high independence in the LHS procedure can only be ensured infiltration rates, combined with the low ks, led to radio-when the final sample site matches the intended sample nuclide release along fast fracture flow pathways with low size. When this is not the case, as with the pluvial scenar-retardations in Column D, which contributed most of the io, statistical independence cannot be ensured, as the high consequence releases in the base-case scenario. NURl!G-1327 - 62

9. Analysis and Results Table 9.10 cept for theinfluence of fractute flowin Column D as the Regression of Llquid Pathway Cumulative Releases saturated conductivities of most of the hydrogeologic (Raw data correlations) units in the other columns was sufficient to ensure reten.

tbn M mmt d the egnkant, ht retaMed M. Variable liase. case llane. case Pluvial onuclides. 10,000 yrs 100,000 yrs 10,000 yrs - 9.5.6 Sensillvity lo NRC 1 erformance W.P. Lifetime 4.045 -0.049 Criteria 0.09 0.13 0.32 . Solubility UOf Infiltration 0.10 0.31 0.23 NRC defines a set of performance criteria for particular b rriers in 10 CFR 60.113: Contact Traction 0.18 0.44 "60.113(a)l(ii)(A) Containment of IILW within the waste Mean log ks TSw 4.11 nackages wdl be substantially complete for a period to bc Mean log ks -0.14 -0.22 --0.28 determined by,the Commission... that such penod shall i CHnz not be less than 300 years not more than 1,000 years after -0.20 permanent closure of the geologic repository;...." Retardation Coeff. (- Cm Retardation Coeff. -0.23 -G.22 "60.ll3(a)l(ii)(ll)1he release rate of arfy radionuclide Pu from the enginected barrier system folkiwmg the contam-ment perio-1 shall not creced one part in 100,000 per year Retardation Coeff. 0.18 of the inventory of that radionuclide calculated to be Ra present at 1,000 years following permanent closure....." Solubility Cm 0.17 (a) geolog repmitory shall be located so -0.27 that pre-waste emplacement groundwater travel time Solubility Pu along the fastest path of likely radionuclide travel from Correlation 0.11 Length the disturbed rone to the accessible environment shall be at least 1,000 years...." 9.5.5 Average imporlance of Radlonuclides Wsc limitations imposed by NRC have the intent of providing a set of critena for the repository independent !!he average importance by radionuclide to the cumula. of the EPA release limits specified in 40 CI'R Part 191, tive radionuclide release for the scenarios was calculated and prevent rellance on a single barrier to the release of radionuclides to the accessible environment, by taking their average contribution to the EPA ratio for all vectors. Table 9.11 shows this contribution for the base case scenario at 10,000 and 100,000 years, and the 9.5.6.1 Effects of NRC performance criteria on pluvial scenario at 10,000 years. In addition, the base-case CCDFs scenatio results are broken down by infiltration rate, in-cluding only those vectors with rates less than 1.0 mm/yr, The relationship between compliance with the NRC 2.0 mm/yr, or 5.14 mm/yr, to demonstrate the sensitivity standards and the outcome of the performance-assess-of the results to this parameter,%c isotopes Pu-239 and ' rnent calculations was examined in terms of compliance Pu-240 stand out as the most important contributors to with the cumulative release limits. This was not intended the EPA ratio because of their.large inventory in the to be a demonstration of the effectiveness or lack of source term, long half-lives and potentially low retarda-effectiveness of the NRC subsystem performance ente. - tion in the rock Nearly all tne contribution of these radi-ria, but was instead a demonstration of the usefulness of onuclides comes from the initial source-term inventory performance-assessment modeling in making future deci-rather than from the chain decay of heavier radionuclides sions on regulations.The conditional CCDF for the base-(e.g., Am-243). Other radionuclides are important in a case scenario was recalculated by using the original set of few cases. For example,1-129 appears for the 100.000-500 input vectors and output releases, but screening out year base-case scenario, for infiltration rates of less than those vectors that did not comply with the NRC cnteria -.1.0 mm/yr, because it has an exceedingly long half-life, stated previously The subset of vectors that " passed" the The isotopes 1-129 C-14, and Tc-99 would take on high criteria were then used to plot a CCDF and compared to relative importance if the groundwater flow were always the CCDF plotted from all of the vectors for the base-restricted to matrix. rather than fracture, flow.nis would case scenario, unconditionally.The screening procedure have been the situation for the base-case scenarios. ex-is described next: 63 NUREG-1327

-Ei(I

t!I!!ll' iff!'

>t ti ,I, i' !t

• ' > Ee $ 8c; g E E

) l 9 3 9 0 a 6 4 5 0 0. i 0 v 0 4 4 0u 0 0 0 01l 1< p 5 's 0 ( t i 0 m 0 n i 0 7 3 1 9 3 2 3 1 0 a L 0 ) 1 1 6 2 8 4 1 1 1 h e 0 0. a 0 0 0 2 1 4 0 0 0 t s e 0 s 0 0 0 0 0 0 0 0 0 r a 11b e < ( e t l a eR 0 e 0 rg A 0, ) 4 1 7 4 9 1 1 2 1 4 8 8 e 1 3 3 1 8 8 1 2 1 3 1 1 f P 0 0. a 0 0 0 5 1 0 0 0 0 0 0 i s 0 E 0 o 12b 0 0 0 0 0 0 0 0 0 0 0 0 d t <( l s o e B d ,0 i 0 5 5 6 9 4 6 2 8 6 4 l 0 ) 1 1 2 6 2 1 1 4 2 2 n , 4 e cu 01s 0 0 7 0 0 0 0 0 0 0 o i n 0. a 0 0 0 0 0 0 0 0 0 0 t o 15b i <( u d b i a r R 0 t f 00) 9 6 8 5 n 6 1 3 6 o o 0. e ,1s 0 0 4 4 c e 0<a c 1 b 0 0 0 0 1 ( 0 n a 0 tro 00 n 1 6 3 8 3 a p 0. ) m 02 e 6 1 1 3 6 ht I ,< s 0 0 0 4 4 0 a e 1 b 0 0 0 0 0 r ( e g a r t r y a r e y /m e r v g A m) 7 4 4 5 1 1 9 1 2 2 0 e 7 1 9 01a 0 0 0 0 0 4 0 f 0 3 0 04s i 0 0 0 0 0 0 0 1 ,. b 0 0 0 0S( d 1 1< et 9 ne e l e s b d e a n i r 1 3 7 8 9 0 1 2 0 p T o l 9 9 3 4 6 8 i c 4 4 4 3 3 3 4 4 4 3 2 9 3 3 3 3 y t u 2 2 1 2 2 2 2 2 2 2 1 2 2 2 2 l ea n c n mr o m m C p u u u u u it i A A N P P P P P T h I U U U U O T Tl d i a f R n I d TCEcL.M" I l! i t

E 1-n e 2, =go suo ~ sEE ~ o c.E _'-- ~ s % Rg goi N Demonstration Plot gij &Rn \\ see caption E 5' = \\ 3;n 10-1._ \\ \\ rs n 8 \\ = E, < g-g93 \\ g str O g c.7 a u. -= u \\ ";R \\. \\~ ~ ~ ~ ~ ~ ~ &G P o w c; 5 v, 2, a.g (' oo< N c ~,a .' ~~ ~~~ ~~.~~~. 3 a r,' N 10 % oa n e ei e _ng ~~. ~~ \\ g% I 5$3 e4 \\ } ,a n ,w-EGB 5.14 mm/yr or less c- -5a? --- 2.0 mm/yr or loss a 3$ns.

---

1.0 mm/yr or less

.c -:2 5. a i EdT 10-3 > i aiaiag

• ' ' '**'l i

>>.aug i .......l .>>>I w 10-4 10-3 10-2 10-1 1 10 ., o a ],]g z !E 17 Normalized EPA Release E. c. = k 3 O ? c t X M tJ N

S s 2 C o m d. C 2! .L M $ _R.- - --- - s g x ~ Demonstration Plot ? g ~ ' N, N s m s see capt. ion c_ -cn s gga s s

c. - c osm

\\ \\ e n -, o n,c i N w- \\ e r. - E 8.2 \\ z we9 s .s x 10-1 _ \\ T g n-i-e, < =9a \\ ~2 2 u,, i, T s a e A "Z8 u. Egg C ~ N e o o ~ l %) EC9 O g25 \\ n E-N e e, c a N-2e ew \\ 5N 10 ', T s w . ' -, " = \\ =e a m_ g ? @8@ 5 m I a =m, Vectors less than 5.14 mm/yr 3 w z, .,9E E 45E- --- Vactors less than 2.0 mm/yr 1 a 5,,e7 @,

---

Vectors less than 1.0 mm/yr W 59

..og t o ". 10.......I ......y ......i 10-4 10'-3 10-2 10-1 1 10 ., w y,2 =a:; = s: Normalized EPA Release ,o-- w qc

-dc 1_ oe a ~ -an wo ogo ~s*s o ac o om =n-, N OmC \\ me s ng8 g

2. n.

N 5'2,g \\ $.y s \\ Demonstration Plot

• A n' 10 \\

\\ see caption ggg \\ F' =e cRu cEg 1 SE< w \\ u. -g_oen o i e8* O 9 3 MR 0 cea as f,e w w m wo. -83 10-2_ 537 ~ w;.cEga 1 o no n

• s 5ma mo i

C$m c i sg= All columns l 'E3h

---

Column D only j

oa a ja = m.. e

c. o o

o .

• s' 10-2 g

g a g: ......g ... ug ......g ..ng .......j 3M 10-4 10-3 10-2 10-1 1 10 f z o._ w E M Normalized EPA Release E = F o-i C f 1 o ti E

2 ,C 9 E.. .o 3 3 b C 1 ___~~,_- w m w g o e c. 3 ,~~~g' x eb s*% a C ._= 5770 ogoo 's s eoEC . hse' 'g' '"~ 's~~_% > o m"n,,n-- o r. 58 \\ ,a a G 10 _c, Demonstrat on Plot '\\ i

3. s,, e m

g so n see capt. ion a< m g ty3 3 "Y U i ur.a m ) k @TE a " <> a g s 8 0 3qa< 0 g g r 2 x an m sman t

• ::r e o w

.o g i r e a:m. g i ?EEn ?, a o ?, %s\\ s5j 10-2_ t E a c, \\ o -- Esn a gem 2 t 3 4 g r 3," c w I a i E9s g e g e.'E = 1 L o r u -. c -.EE All columns t i mx =EE ~

---

Column D only 3

3 1 =8a o aa I Q. C. U- -c* 10-3 4 c.,, ...ii1 ......I .....I .. i i.i ......i $"4 10-4 10-3 10-2 10-1 1 10 =r Normalized EPA Release 1 i I L 1

3-^.- -A -4 ~ .a E o e ,,,, -e,, -e 1 s - c. n A A gn o

c. 3 C s

a $. c,$ 3 5x ,4 Demonstration Plot A, E. c - "E 1 mcy see capt. ion g ? n w-10-1 _ 1 cr F,2 g ~w e @2: w1 I -f.. \\ n.

c. e. e

= <r, u. 4G< a g 96R O i c85 0 I <=w 29o I ee s r, a \\ jW \\ o o._: \\ '< 2 9 10-2-- \\

c. ; -

B 4,- B \\ 2ss' \\ a v. o =r saa m v u,gr 2$ n n; c y ~o 3g khh -- Colurnn D only - Total 3 =. o e, ~e E. o 2 zyB 10-3 i I 't ,,....g ....,I ,,,1 o.s 5-- 10-2 10-' 10 1 02 v. $~ e, 7, -2 e 7. b E b. Normalized EPA Release E G 5 '"-~ g s we 1 =* Y 9>J 4

2 P C3w 1-y .L c s-M o a w 9 ~'% E. o ~ w

e

\\ O s -c*ng N E. =r

c. g w N

E o g 3IP N 8a6 i E5O N E $,s, 10_3- \\ Demonstration Plot o ? n g- \\ see capt. ion W E, a \\ =. -' E- ~ i =T_, n \\ B9n t O 1 E =. *O 1 \\ ' ~., 5.?: o \\ s.e g c n ;; R 8 S E -< g ce5 9,88 \\ 37 s,= - 1 i, >3

• aa 10-2_

o n.B r { v b b?, ~ \\ 's, ~ -- u - oa5 \\ s

! 2.,,

) w=5a cn c,c \\ s o 4l45 ~ ---Complies with 1000 yr GWTT g 5 All Vectors and 10-5/yr release } [ l l =J

---

Complies with 500 year W.P.

• a=

oW+ 7=5 10-3 ,,,,,1 i,...1 ,,,,,I ,,,,i 5 ~e5 10-3 10-2 10-1 1 10 n _5-l Sf$1 Normalized EPA Release o. m i I r b

9. Anolysis and ltesults E

e E

.9 Q. e8 .e c E= b ci. yu - e i c e 2 E 2 E o o e e l c f b _ g b-eg__-._._._.____ - e '5 D p F E 0 a: r, z g u c S 8

e m

0 o o o I L. O -e O O O o I O O ( ~ O o qr -0 0 ^ 3......... ........ry......... 8 8-8 8 8 8 8 e4 6 6 6 -4 e4 o-e e O!)e 8 V d 3 Figure 9.11 flase Case liquid Pathway Scenario: 10,000 Years. Effects of Groundwater Travel Time on liPA Release. 'ihis graph presents results from an initial demonstration of staff capability to conduct a performance assessment. The graph, like the demonstration, is l limited by the use of many simplifying assumptions and sparse data. 1 71 NUREG-1327 I a,

9. Analysis and Results T

s

-n e> U c. c' l l' soo o o o 7 o .~ e u E t .-) o x 2 2 D-e e@ g .e z = g o. .5 33 T Ee o ~ w ~ o E E a m O o e E o m ,i n, != T) 1 1r 6 H" o Ir o iS ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,i,,,,,,,,,i..... 8 8 8 8 8 8 8 H d a d e' d d e r Olleu Vd3 Base Case lj uid Pathway Scenario; 10.000 Years, liffects of Release Rate from Figure 9.12 q lingineered Harrier. 'Ihis graph presents results from an initial demonstration of staff capability to conduct a performance assessment. *lhe graph, like the demonstration, is limited by the use of many simplifying assumptions and sparse data. NUREO-1327 72

.g 12.00 - o C Demonstration Plot w = g s'g see capt. ion o .t e u. g o 5 g,in 10.00 - cc n. y co ,o e,-- NRC Lower Limit of 300 Years

• NRC Upper Limit of 1000 Years

's Dr s- $ hg 5 l (there were no vectors greater than g M cali 8.00 - , h *,I 3 l 1000 years for lifetimel I. I qw 8 ;f .o o 1 5's = I wo-c ucag=- a sa c: eoe 6.00 - o a = .e. 4 eW o""- n. 5.:3 e m 0 av2-n n'.a E < I -a 5 l ag-g i ?=4a 4.00 i I

w-=

I . y-l co ag 1 aK50 ~ l no o o l o o i M o I

c. n o en-l sam 2.00 -

e-m =5n na3 o Sc" e o i e 4 l a mo l y asa ___A___ D_.a_ _. a = _m 2 m r., ,1.......... s {9 O. 1.,r-._ 3 EE" 0 200 400 600 800 1000 1200 E ,,,,,1 a g,' D E a:. Lifetime of Engineered Berrier, Yrs V p x E 2 Y E G E w" s

9. Analysis and itesults 0

Substantially complete containment-Vectors with curve for all vectors. Seevening based ou a waste package a waste. package lifetime less than a specified time lifetime of 500 years or greater caused a shift toward were scrrened out. For this demonstration, only a lower radionuchdc releases, but only for the low. single representative cutoff time of 500 years was probability, high-consequence releases. chosen. e Screening on the basis of groundwater travel times had 'O ^" ""*#4""## '#' llelease rate limitation ~'lhe release rate inmielin NEITilANaccounts onlyforIhecongruent e! case quecan y e nuna nW ca ngto of radionuclides contained in the uranium dioxide W' #' "*" k"E" "" ' I'"" # # #"" "" W "#U "" "##"E""F"E fuct. The maximum rate is cmtrolled by the disolu-tion rate of the matrix. l'or this demonstradon, the ta onuclide transport) along (olumn D was controlled release rate was assumed equivalent to the dissolu-by the fractures, for infiltration tales higher than the tion rate of the matrix. Releases of some af the kssof the units in the column liracture flow was faster an w Huuugh the rock matrix, and, because retarda-radionuclides rnight actually be smaller than the congruent = dissolution rate because they are onue a mu mnsported more quickly. 'lherefore, solubility limited; thus, the screening criterion might be slighdy over restrictive. *lhe diswiution e m nadng e nton with Qu inMadon mtes ab we tmne only ou$ tk mats unk unsatu-rate calculated in NEFillAN is a function of ura+

• • * " * "I ' I "" E " E '"""

nium solubility, infiltration rate, and water contact fmetion. It should be noted that, for this demonstra-tion, the assumptions used do not corre$ pond pre-cisely to the rule. Specifically, the rule states a limit 9.5.6.2 Aurage contributions by radionurtide of "one part in 100,000 per year of the inventory of that radionuclide present at 1,000 years," with a Table 9.12 illustrates the average contribution by radkw limitation on those radionuclides that might have nuclide for the unrestricted veetors and for those vectors decayed to very low levels at 1,000 years, 'lhe dem-meeting with: (1) the 500 year waste-package hfetime, onstration wrs therefore only an approximate com-and (2) the groundwater travel time of 1,000 years. For i parison to the limitations of this subsystem require-the unrestricted and waste. package lifetime cases, the ment. major contributors to cumulahve radionuclide rclease are the isotopes Pu-239 and Pu-240. Ilowever, for those releases with groundwater travel times of 1,000 years or o Groundwater travel-time limitation-For this dem-greater, the radionuclides C-14 and 1-129 are the main l onstration, radionuclide transport was assumed to contributors.'lhese radionuclides wcre unretarded in this i occur along four separate liquid water pathways: demonstration and therefore could move relatively Columns A,11. C, and D, in part to simulate the spa-quickly tbrough the inatrix. tial variability inherent in the Yucca Mountain re- . pository. Column D was the shortest pathway, and contained hydrogeologie units that would saturate 9.5.6.3 Ad floe Sensitisitics to NitC Criteria more quickly at increased infiltration rates, thus - f leading to fracture flow 'lherefore, the mean travel "Ihe results of the 500 runs for the base-case scenario time along Column D was taken to be the " ground-(liquid-pathway analysis) were plotted against the values water travel time along the fastest path of likely ra. of the individual NRC subsystem performance criteria of 'dionuclide travel". In this demonstiation, ground.' groundwater travel time, waste package hfetime, and re-water travel time is defined as the average time for lease rate from the tills 'the results, presented in Fig-plug flow through the column and is a function of in-urcs 9.11,9.12, and 9,13, show that, for the assumptions filtration rate, porosity, saturated hydraulic conduc-inherent in this demonstration, imposing the NRC per. tivity, and correlation length..

formance criteria could have a favorable impact on the-total radionuclide releases to the accessible environment.

For the scenario considered, imposing the-1000 year Figure 9.10 shows the conditional CCDF for the base-groundwater travel time limitation virtually eliminated case scenano for all veetors, and for vectors limited by the any rion compliance with the FPA containment require-EDS release rate, the waste-package lifetime, or the ment in 40 CFR Part 191. Additionally, none of the vee-groundwater travel-time restrictions. No retationship was tors yielded release rates, from the Ells, greater than assumed between waste-package lifetime and the Ells 10Myr, although it was noted that cumulative release release rate, in this demonstration, all 500 vecton had increased with an increase in the release rate. Finally, release rates less than 10Wyr, so the CCDF curve for there was also a noticeable decrease in cumulative radio-vectors under that restriction was coincident with the nuclide release, with increasing waste-package hfetime. NUREG-1327 74 a. .a

i I

9. Analysis and Results l

Table 9.12 Frictional Contribution by Radionuclide to F.PA Release Itatio fcr Unrestricted Vrctors and Those Restricted by NRC Pc formance Criteria Hestricted to Unrestricted A 500 r, W.P? Restekted to 3 Rudt nudlde settors 1.ifetime A 1000 yr, GWI'I** e - = - P,.-240 0.41 0.40 0.0 Pu-239 0.39 0.37 0.0 C-14 0.094 0.13 0.94 Am-241 0.077 0.062 0.0 Am-243 0.014 0.014 0.0 1-129 0.005 0.007 ' O 05 'W.P. is waste package. L l "GW'IT is groundwater travel time. 9.6 Total System Results determined by multiplying the prolmbilities of their inde. pendent constituent events.'lhe hkelihood of each event 9.6.1 Introduellon was based upon staff judgment in the case of the pluvial / nonpluvial events, and 40 t 1 R Part 191. Appendix !!. for 'Ihe results presented here can only be considered as a the human. intrusion events. 40 CFR Part 191 assumes a demonstration of a performance assessment capability likelihood of drilling at the site r.s a set number of and should not be taken as representative of Ihe perf orm. boreholes per unit area over 10,000 years, based on the ance of a repository at Yucca Mountain, Nevada Among geologic formations in which the repository is located. the most important limitations of the study were: ,lhere are two,mportant points to note in Figure 9.14. i l.- the lack of sufficient site data, First, given the scenarios and probabilities chosen for this demonstration, the case m w hich conditions at the reposi-2. the large uncertainties in the data now in use, tory over the next 10,000 years remain as they are today appears highly unlikely. Secondly. the two human. 3. the use of only four scenarios to characterize intrusion scenario classes have probabilities orders-of-future states at the site, magnitude greater than the base-case and pluvial scenar. ios, which do not incorporate the possibility of drilling 4. uncertainties in the site conceptual model, and events.'lhis difference is due to the high probability of drilling, as opposed to not drilling, as shown in the figure. 5. uncertainties in modeling the physicochemical processes leading to radionuclide release and 'The consequences of each scenario and of all scenarios migration in the geosphere, combined can be expressed in terms of normaliicd cumu. lative releases of radionuclides to the accessible environ-For this demonstration, four sce nario classes were consid. ment over a specified period of time, These results, dis-cred: played as curves of consequences versus the probability that such consequences will be exceeded (i.e. a compte. .l. Undisturbed or base-case conditions, mentary cumulative distribution function, or CCDF), can in turn be compared with the curve of the EPA contain. 2. P1uvial c<mditions, ment requirements in 40 CFR Part 191 (Figure 9.15).flhe EPA standard requires sumnied normalized cumulative 3. Drilling under undisturbed conditions, and releases to the environment equal to:(1) 1.0 not to exceed a probability of 0.1, and (2) 10.0 not to exceed a likelihood 4. Drilling under pluvial conditions. of 0.001. As shown in Figure 9.14, these particular scenarios arose Compliance with the containment requirements cannot from the possible combinations of two fundamental be determined solely on the basis of the strict numeniced events: a pluvial period (or not) and drilling at the site (or results of a performance assessment. As recognized in 40 l not). Probabilities for each of the scenario classes were CI<l Part 191, substantial uncertainties are inherent in l 75 NUREG-1327 ...e-, y v_ rw-w ,._._..,,,.~.w,-_,_,,,w,,,.., .,~n,..

9. Analpis and itesuhs projecting future de gwd-spiern performance, and thus a subset of W sectors might hase led to'punous concla-the bias towards this uncertainty in component perform-tions and an inadequate reptesentation of the p.uttal ante must also be taken into account. For example, in CCD1 s.

l'ipure 9.16. a portion of the empnical CCDi' hes aime the I:PA contamment requirement and is therefore la-9.6.2.3 lhilling under undisturla d omditions beled a "possible violation." If the bias in component The CCDl for dnthnr under undaturbed conditions performance was consistently toward more resumisue performance, then the results expressed in the CCDp (l'igure 9.14) shows a shght step m the h>w consequence! may be too conservatn e, and thus thit. paroon of the curve Mgh probMuhty end of the curm 'lh:s is attributalde to may be judged te be not a violation. If. on the other hand, consequences f rom the dolling. The rest of the CCDF component perimmance was viewed optirnistically, the curve is dommated by releases ua the lhimd pathway. CCDF may well be found to be in nolation of the liPA standard.$ince defmitive proof of future mtem perform-Mon: importantly though, with the addition of dnihng ance cannot he provided,40 Ci l( Part 191'only rcquires a evenn to the base-case, the overall probabihty of the wenano is increased to 0.9. Thus, for this scenario, al. "r easonable expectathm" that compliance wdl be thmph the sum probabihty of an !!PA ratio of 1.0 is bclow achieved. the 1:PA standard at 0.022, the I;PA ratio /cumulatne The partial CCDFs for each of the scenario classes are pnWabihty pair of MO and ODOM fab abow the stand-shown in Figures 9.17 through 9.20. These curves ddier ard, which appears as a step ftmetion m this and the from the distribution of consequence figures shown car-foHowing bruta tier in Section 9.5, in that these partial CCDFs incorpo. 9.6.2 A Drluing under plustal conditions rate the probabihties of t he scenarios themsch es. For this reason, the cumulative probabihty of any single scenario .lhe shape of the partial CCDU for dnihng under pluual presented here never reaches 1.00, as it wdl f or the total conditioWFiguic 9.20)does not exhibit the ef fects of the L C Dl,, which is a composite of all four scenario classes. drilhng.This is because the consequences due to dulhng are m the range of.000), and are therefore nephpble 9.6.2 Partial CCl)F 1(esults when factored into oserall stenano consequences 01.01 to 100. 9.6.2.1 Undisturbed or base case wnditions With the overall hkchhood of dothog under plunal condi. The log log plot of summed normalized I!PA release tions equal to 0.1, the 1.0/0.0S2 consequence!argregate versus cumulative probability for undisturbed conditions probabihty pair falls just below the liPA contammem (1 igur c 9.17) shows the characteristic concave downw ards requirement, whereas an liPA ratio of 10.0 f or this scc-shape for a CCDF. As will be the case for each CCDF, the natio has a cumulative probability of approxunately 0.06, curve intersects the paxis at the Ukclihood of the scenar-which lies abose the staadard. io; here, the hkelihood is equal to 2.3 x 10 7. For this scenario, liPA ratios of 1.0 and 10.0 have conesponding 9M lisib h* tb Tomi CCl)l' cumulative probat<ilities of approumately 4.9 x 103 and 4.1 x ION, both 01 w hich are well below the !!PA critical-Figut e 9.21 demonstrates how each of the four mdiviJual point hkelihoods of 0,1 and 0.001 for these sarne liPA scenarios contributes to the total CCDF. It is readily ratios. apparent that releases from the two human-intrusion sce-natio classes dommate the CCDF, under the given prob-9.6.2.2 Pluslat conditions abihties and condmons. Contrbutions to the total conse-quences from the undisturbed and plunal scenanos are Consequences from the pluvial case scenario (1igure negliptble, because their respective scenario probabihties 9.18) equal to an EPA ratio of 1.0 hase an aggregate are too low. probabihty of 1.9 x 10A while an I!PA ratio of 10.0 has a cumulative likelihood of 1.3 x 10A These results, com-The total CCDU for.he four scenano classes modeled is bined with an overall scenario probability of 2.0 x 10A plotted agamst the 1:PA sta tard in 1 igure 9.22. Tlus leave the CCDF for the pluvial case orders-of magnitude comparison shows that, for this demonstration, the em-below the !!PA containment requirement. pirical Ct Ol' lies above the i t A standard at both containment requnement break points, with cumulative 'Ihe reader should be aware that an inordinate amount of probabihties of approximatcly 0.104 and 0.06 for liPA computer time limited the pluvial and the drilhng-under-ranos of I 0 and 10.0, respectively. pluvial-conditions scenarios to only 98 mput vectors. I:ur-thermore, because a sample of 200 inpat sectors was The results of this demonstration should not be taken as planned and generated with the 1.IIS samphng routine to representative of the performance of a repositor) at represent this release pathway and these scenano classes, Yucca Mountain, Nevada.1(ather, they should be used to NUl(ILl.127 76

~. _ _ _. _. _ _.,.... _ _ _ _ _... _. _ _ _. _ _ _ _. _ _.... _ _ _. _..

9. Analysis and Results indicate the importance of: (1) the assumptions in model-tration rate; and (3) the consistency of the bias, whether l

ing phenomena, such as fracture / matrix interactions; pessimistic or optimistic, toward the performance of the 1 (2) the data used in the total system modeling, e.g., infil-various disposal-system components. I i 1 l i i I ( t t 3 4 77 NUREG-1327 -. -c .....:. u..

a..u.

. u - ..=.

eure?Kr; am -3.........

9. Analysis and llesults DETERMINATION OF SCENARIO PROBABILITIES FROM THE PROBABILITIES OF FUNDAMENTAL EVENTS P

P 0.9 0.1 Sconario Scenario class # 0 class # 1 D 2.3 x 10-7 ^ Probab!!lty Probability = 2.0 x 10-7 = 2.3 x 10-8 Scenario Scenario class # 2 class # 3 D ~ 1.0 Probability Probability N 0.9 ~ 0.1 --P is not pluvial D is no drilling P is pluvial D is drilling Scenario class # 0 is no drilling, not pluvial Scenario class # 1 is no drilling, with pluvial Scenario class # 2 is drilling, not pluvial Scenario class # 3 is drilling and pluvial Note: Probability combinations assume that fundamental events have independent probabilities of occurrence; this is not a general restriction. Figure 9,14 Determination of scenario probabilities from probabilities of fundamental events. ~1his figure presents results from an mitial demonstration of staff capability to conduct a performance assessment. The figure,like the demonstration is limited by the use of many simplifying assumptions and sparse data. Null 110-1327 78

l

9. Analysis and itesults i

EPA BOUND 1.0 LIKELtHOOD 10i! OF EXCEEDING VALUES ON THE l

---

~l HORIZONTAL l

l EPA BOUND AXIS 104l b--- 1 1.0 10 MULTIPLES OF EPA RELEASE UMITS (M) Figure 9.15 Graphical rep (resentation of hypothetical CCDP with the liPA containment requirements DOE,1988). I I m 1.0 g Q , EPACONTAINMENT Z<e g 8, REQUIREMENT e3 10*1 -

• 2 4-wm cw 10 2 _

l ,l POSSIBLE u) a 1 V10LATION gg j s do es 10 3 - j m_ Op I tsa hO l'0*4 - <a Sw Os E2D 10 5-m I i i 101 10+0 101 102 g03 SUMMED NORMALIZED RELEASES (M) Figure 9.16 Plot of an empirical CCDF against the liPA containment requiremen's (DOli,1988). 79 NUltliG-1327

l

9. Analysis and llesults CCDF FOR UNDISTURBED CONDITIONS (10,000 years) 10 N 10 7-

>2 25j Demonstration Plot 2 sco caption n.e 10 -8.; .2 lii 3 E 3 0 10 1 1 Y ] T TY] T T YT y Wgik y TTF y Y gvq'y T*v y T T Y Y Y] 1TT T T T T 'T 5 1 W T11 W F WT q W 10 ") 10-8 10-s 10-4 10-2 1 1 02 Summed Normalized EPA Release Figure 9.17 Partial CCDi' for Undisturbed Conditions. Phase 1 of the Iterative Perfoima.., assessment. Itesults based on 500 vectors, yielding 398 values af ter duplication. 'this graph presents results from an initial demonstration of staff capability to conduct a performance assessment. The graph, like the demonstration, is limited by the use of rnany simplifying assumptions and sparse data. i N Ult!!G-1327 80

-. ~,. -.. - -.. - - . - -. ~.... -. -. _... - _. -

9. Analysis and itesults CCDF FOR PLUVIAL CONDITIONS (10,000 years) 104-4 a

l .I j ~ t = 10 - 2i e .o Demonstration Pim l o see caption e >w .!!!s - Es 10 -.O n. \\ I e 10 : 10 .......i .....y i -. 2 - 1 1 10 1 02 Summed Normalized EPA Release i Figure 9.18 Partial CCDF for Pluvial Conditions. Phase 1 of the Iterative Performance Assessment. Ilesults based on 98 vectors, yielding 98 values after duplication. The graph presents results from an initial demonstration of staff capability to conduct a performance assessment.111e graph. like the demonstration. is limited by the use of many simplifying assumptions and sparse data. 81 NU11EG-1327 ..~

9. Analysis and Results CCDF FOR DRILLING UNDER UNDISTURBED CONDITIONS (10,000 years) 1-

{ EPA Standard 10-1 : 5 .e

=

.O .O o5 Demonstration Plot

• 10-2~i soo caption

.z .e u ~ Es O 10-3: 10 4 ,ii,1, i,,,,,,,, 10-5 10-4 10-3 10-2 10-1 1 to 102 Summed Normalized EPA Release Figure 9.19 Partial CCDF for Drilling Under Undisturbed Conditions. Phase 1 of the Iterative Performance Assessment. Results based on 500 vectors, yielding 5(X) values after duplication. 'this graph presents results from an initial demonstration of staff capability to conduct a performance assessment. The graph,like the demonstration. is limited by the use of many simplifying assumptions and sparse data. NURiiG-1327 82

9. Analysis and itesults l

l CCDF FOR DRILLING UNDER PLUVIAL CONDITIONS (10,000 years) 1: ~ EPA Standard 10-1 .k_ 3 .E 2 Demonstration Plot o 10-2.; see caption 3 .9 o E o0 10-3 M 10 'y ......rr .- r,, r n,, ii 10-2 10-1 1 10 102 Summed Normalized EPA Ralease Figure 9.20 Partial CCDl' for Drilling Under I'luvial Conditions. Phase 1 of the Iterative Performance Assessment, licsults based on 98 vectors, yiciding 98 values after duplication. '1his graph presents results from an initial demonstration of staff capability to conduct a performance assessment. 'Ihe graph,like the demonstration, is limited by the use of many simplifying assumptions and sparse data. 83 NUllllG-1327 __.____._________.__.______m _m_. --._.m.

9. Analysis and itesults TOTAL CCDF (10,000 years) 1

! Total CCDF ~ EPA Standard s 1 Pluvial /DrillinD Case-N -g - --.. - 4 %~~, 102] Undisturbod/ Drilling % Case l I 3 1 j 10-4 1 Demonstration Plot E see caption l

• a 3 10-6 l E

Undisturbed Case au -~~~~.. l

• N Pluvial Caso -- b-

-.'N ( -.~~ 10-e, %'s b \\ \\ l ) \\ l I 1 0 - 10 ..,...,....,....,..m....,..,.,,,,m..1, .i..,...,- 1 0 - 10 10-s 10-6 10-4 10-2 1 1 02 Summed Normali ed EPA Release Figure 9,21 Composite CCDF Curve for the Scenario Classes Considered in Phase 1 of the Iterative Performance Assessment. llesults based on 596 vectors, yielding 1196 values.1094 after duplication. This graph presents results from an initial demonstration of staff capability to conduct a performance assessrnent. 'the graph. like the demonstration, is limited by the use of many simplifying assumptions and sparse data I NUlll!.G-1327 84

9. Analysis and Results TOTAL CCDF (10,000 years) 1.

EPA Standard 10 = 2 y iE Demonstration Plot see caption cE 10-2= .2 E 3 E { = O in : 10-4 1 0 -10 10-8 10-s 10-4 10-2 1 10:- Summed Normalized EPA Release Figure 9.22 Total CCDF for Phase 1 of the Iterative Performance Assessment. Results based on 598 vectors, yiciding 1196 values,1094 after duplication. His graph presents results from an initial demonstration of staff capability to conduct a performance assessment. De graph, like the demonstration, is limited by the use of many simplifying assumptions and sparse data. I 85 NUREG-1327 l

10. PRELIMINARY SUGGESTIONS FOR FURTilER WORK Dased on this preliminary analysis and the limitations radionuclides directly to the environment through notd some preliminary recommendations regarding the exploratory drilling. Ilowever, there did not appear directions for further technical work can be made. These to be any readily available models and computer suggestions are based on insights gained during the Phase codes to estimate consequences from volcanism, 1 effort, Although these suggestions were derived from faulting, subsidence, upttft, and other icctonic this work, they are not necessarily unique to this work, are events and processes and other types of scenarios.

generally consistent with scientific intuition, and are The consequences from these scenarios do not ap-largely consistent with planning documents such as the pear to be readily treatable by extensions of models DOE SCP. He suggestions relate to this report and are currently in use (such as the way the pluvial case was not intended to indicate an evaluation of the DOE pro-treated by extending the base case "catment). gram outhned in the SCP.These recommendations have Therefore, the capability for modeling the conse-all the limitations inherent in the analyses on which they quences of additional scenario classes must be added are based.These suggestions presented are in the spirit of to the methwlology, if such seenario classes are to be providing some ideas to guide further work and are not treated explicitly in the CCDF. intended to be definitive. Some of this suggested work is clearly ihe responsibility of DOE; other items could be 2. Test the system code, using the consequence codes performed by NRC, DOE, or a third party. Most of the as subroutines, instead of generating data sets exter-s recommendations of this type reflect the general lack of nal to the system code. data available for executing an analysis of this type. The suggestions for technical improvements can be grouped in the Phase 1 effort, the consequence modules into three categories-were run separately from the system code, and the resulting files were manipulated to generate the to-1. Suggestions to improve or extend the modeling used tal system CCDF. An attempt to run the conse-to obtain preliminary estimates of performance; quence modules as subroutines of the system code was not made. Such an attempt would indicate 2. SuggesJons for refmmg or adding auxiliary analyses whether such an approach may be practicable and to help better evaluate the performance estimates would provide an important insight into the direc-obtained; and tion for further development of the NRC independ-ent performance-assessment capability. 3. Suggestions for refinements or additions to the sci-entific bases, including t he methodologies available, 3. Acquire, test, and evaluate codes that SNL devel-for arriving at estimates of repository performance, oped for a repository in the unsaturated zone. SNL, under contract to NRC, has been developing 16.1 Improvements and Extensions to an extension of the SNL performance-assessment Modeling methodology, to treat an HLW repository in par-tially saturated tuff. At the beginning of the Phase I he following are recommended improvements to mod-demonstration,it was recognized that the SNL tuff eling of performimcc. These ate considered to be ideas methodology would not be ready to use in the Phase for further work that could improve the current prelimi-1 effort.This tuff methodology also willincorporate nary assessment and might be suitab!c first steps in gener-the ability to treat transient conditions by a multiple ally upgrading the methodology. steady-state approximation, llecause this methodol-ogy was developed specifically for the NRC waste General management program, it has the potential to greatly improve the accuracy and adequacy of the perform-1. Add the capability for modeling additional scenario ance assessment capability. By acquiring and evalu-

classes, ating this methodology, the NRC staff can deter-mine what improvements o: additions, if any, may be in this Phase 1 demonstration, the staff used a read-

needed, ily available computer code, NEFFR AN, to model the release of radionuclides by the groundwater 4.

Evaluate additional con.puter codes, which could pathway.His code was able to treat both the " base-not be acquired and evaluated during the Phase 1 case' for current climate conditions and the " pluvial effort, to determine whether existing codes can meet case" for a wetter climate. A simpic model and com-the NRC modeling needs, or whether additional puter code were developed to treat the release of code development is needed. l 1 87 NURl!G-1327

10. Preliminary Suggestions Several computer codes, which appeared to be variables may not have been revealed by the sensitiv-promising in terms of providing missmg parts of the ity analysis, aruuysis or which might offer improved treatment of certain aspects of modeling, were not asailable Flow atid Transport for use in the Phase i demonstration. Several of these codes should be evalcated in subsequent itera.

1. Refine groundwater modeling (e.g., by considering tions to determine how relevant and useful they are more dimensions). for the NRC iterative performance assessments. .lhe assumIitions used as the bases for flow nmdel-Some of the codes that might be worthwhile inves.. ing, w hich are then the bases for transport modeling, -ting are: TOSPAC, AR11ST NEFl'RAN 2, and greatly simplify the complexity of the structure. "AC* toundary conditions, and physical processes consid-cred in modeling flow at Yucca Mountain. Among with the CNWRA, the adaptation of the the more significant simplifying assumptions used in I

• babilistic Performance Assessment (FPPA) the Phase i effort were that: (1) flow was one-al^gy to generate the total system CCDF.

dimensional and vertically downward; (2) flow was steady-state;(3) surface infiltration was assumed to , the Phase i demonstration, questions arose be constant over time; and (4) fracture flow was initi-Jng the number of vectors required for an ade-ated when the infiltration' rate exceeded the satu-pate representation of the distribution of conse-rated hydraulic conductivity of the matrix. A more quences for a given scenario class. This question usu-precise and coinplete treatment of the hydrology at ally arises in studics of this type where performance the site could treat some of these aspects by using is estimated using an " empirical distribution" de-two-or three-dimensional models, incorporating a rived from models of system performance using mul-better treatment of fracture flow, considering the tiple samples of input data. Appendix 11 discusses coupling ;o regional hydrology, and removing addi-some of the concerns almut ensuring that enough tional simplifying assumptions. Additional site by-samples are used to obtain a sufficiently accurate drologie data could be incorporated, if available, representation of performance. A concern in this study is that several vectors yielded zero cumulative 2. Incorporate a model of gas-pathway transport in the releases; although this outcome increased confi-calculation of the CCDF. dence in the probability estimates of the low-consequence /high-probability end of the CCDF, in Phase 1, the only release pathways implemented less confidence was available for the high. in the model used to generate estimates of perform-consequence / low-probability end of the CCDF, ance were the liquid pathway and the direct release which may be the critical region for assessing regula. pathway (i.e., release by exhumation of waste or con-tory compliance. The:cfore, the use of an impor. taminated rock). A more complete treatment would tance sampling technique, such as the FPPA meth, explicitly use the concepts discussed in Appendix D, odology, if made applicable for the total CCDF, may " Gaseous Releases of Cartmn-14 " to formulate a combine an increase in accuracy and confidence in model that quantitatively estimates releases by the results with a saving in computational cost and time. gas pathway and then incorporates these estimates into the total system CCDF, as discussed in Sec-tion 4, " System Code." In addition, it might be nec-6, Perform a sensitivity analysis, using both drilling and essary to couple the hquid and direct-release path-groundwatec transport parameters. way calculations of releases to those from the gaseous pathway to ensure conservation of mass During the Phase 1 analys.ts, the sensitivity analys.is (currently, the models used assumed all C-14 was was performed only on the liquid pathway model released in dissolved groundwater) and to character-(because the drilling model and code were not avad. ize correctly the interactions between the various able at the time the sensitivity analysis was done), us-pathways. ing the variables and distributions germane to that model. Some of these same variables were impor-3. Include flow and transport through the saturated tant for the model of direct releases by drilling-zone. However, some of the variables that could have a sig-nificant effect on the consequences of drilling were in the Phase i demonstration, flow and transport of not included in the sampling procedure used to per-radionuclides in the saturated zone was not incor-form the sensitivity analysis, but were fixed in the porated into the estimations of total system model. As a consequence, the variability in the out-performance. Instead, the radionuclide releases put of the drilling model inappropriately may have were calculated at the water table (i.e., the boundary been kept small, and the importance of some of the between the unsaturated and saturated zones), NU Rl!G-1327 88

10. Preliminary Suggestions although estimating consequences in this manner In the Phase 1 demonstration, the coupling between was probably conservative, because retardation and groundwater flow in the fractures and the matrix ns travel time in the saturated zone were neglected.

modeled by assuming that:(1) flow was entirely m Adding consideration of transport in the saturated the matrix, if the infiltration was less than or equal to zone is recommended because: (1) a more realistic the ks of that segment; and (2) flow greater than the model of system performance will result and (2) syn-ks was carried by the fractures, if the infiltration ergistic effects will be portrayed with increased con-through the segment was greater than the ks. Al-fidence. For example, the impact of releases from though the NEFIRAN code has the capability to the vertical columns, used in the Phase 1 effort to treat matrix diffusion, this capability was not exer-describe the geometry of the repository, may be sub-cised for the Phase 1 demonstration. A more com-stantially different when the effect of transport plete, precise treatment of the coupling between the through the saturated zone on' those releases is in-rock matrix and the fractures, for both groundwater cluded in the model, flow and radionuclide transport, would improve the completeness of the model and would provide fur-4; Use a more sophisticated computational model for ther insight into the importance of these couplings transport through partially saturated, fractured and the parameters influencing the couplings. rock. ~ The NEFIRAN code was used to calculate trans-port in the Phase 1 demonstration, it was developed 1. Attempt to develop or use a previously developed ) to simulate radionuclide migration in saturated rock. mechanistic model of waste-peckage failure. (The following analytical steps were used to simulate ' radionuclide migration in partially saturated rock us-In the Phase 1 demonstration, a distribution was as-ing the NEFIRAN code: sumed to describe the time of waste-package failure, and all waste packages were assumed to fail at the

i. fl'he saturated flow solver incorporated in the

, same time. The assumed distribution was not related NEFIRAN code was bypassed, and the flow to any of the parameters that are usually thought to - was calculated assuming partially saturated influence waste-package failure, such as: repository flow in four one-dimensional columns. _ temperature as a function of time, the rate and man-ner of water contacting the waste packages, the geo-Lii. If the calculated conductivity of any segment of chemistry of the groundwater, and the stress field to a column was less than the saturated hydraulic which the packages are subject. These factors can be ~ conductivity for that segment, then the porosity a function of the repository design, the evolution of was multiplied by the degree of saturation (to repository conditions with time (primarily thermal account for partially saturated conditions), and. . and hydrologic conditions), and the_ occurrence of this modified porosity was used in the substantially changed conditions produced by vari-NEFIRAN code to calculate radionuclide mi-ous scenarios. A mechanistic model = of waste-gration. package failure would relate the source term to these factors. Incorporation of such a mechanistic iii. If the calculated conductivity of any segment of model can help to reveal the interactions between a co!umn was greater than the satarated hy-the source-term behavior and the behavior of other draulic conductivity for that segment, then all parts of the repository system. l the transport was assumed to occur in the frac- _ ture, and the properties of the fracture were ' 2. Develop a mechanistic model of contact between - used in the NEFIR AN code, to calculate radi-groundwater and the waste. onuclide migration. In the Phase 1 demonstration, the fraction of gro-

Improvement in the transparency, accuracy; and ro-undwater contacting the waste (and thereby brought

~ bustness of the modeling of transport through un-up to the appmpriate limiting concentration for saturated, fractured rock could be achieved by taking each radionuclide) was assumed Io be a random vari- . a' more direct approach =to modeling phenomena able, selected from an assumed - distribution. ' A ~ - such as: (1) flow in the partially. saturated rock; mechanistic model for the fraction of groundwater (2)the transition from matrix flow to matrix plus raised to the limiting concentration of radionuclides fracture flow:(3) transport in the partially saturated could relate this fraction to parameters generally matrix: and (4) the exchange of mass between the thought to influence such mass transfer, e.g., the na-fractures and ma rix. ture of flow near the repository (including, the flow rate, the degree of saturation, and the flow profile 5, Explicitly model fracture / matrix coupling. near the _ waste packages), the degree of mixing 89 NURIE1327

~~ - -

10. Preliminary Suggestions induced by the repository design, the thermal condi-mation is and under w hat conditions it is more tions in the repository and the potential for ther-or less accurate would be tiseful.

mally driven flow. An even more direct approach would dispense with the concept of the fraction of h. The effects of spatially varying saturation on groundwater contacting the waste and instead, radionuclide migration, would calculate m iss transfer fre n the ensemble of in the Phase I demonstration, the effects of waste packages to the geosphere, based on the ap. propriate physical and geometrical parameters, spatially varying saturation were assumed to be limited to changing the amount of groundwater available for advection and dispersion, as the 3. Treat the repositi,ry as a source of radionuclides distributed in time and space. E*".ndwater moved through vanous hydroge-ologic uruts ne possibility of a more complex influence of the variation in saturation along In the Phase 1 den onstration, all the waste packages the migration path on radionuclide transport were assemed to fail at a single time, rather than the was not considered. For example, some reac-more realistic assumption of waste-package failures tions, such as those resulting in precipitation, distributed m time, and therefore, space, ne re-may depend'on the amount of water available. _ lease rate of the inventory from the containers was An auxiliary analysis to determine how well - assumed also to be limited by the solubility of the approximations useful in fully-saturated flow UO matrix.Someof thespatiallydistributednature can be extended to model partially-saturated 2 of the repository was treated in Phase 1 by partition-flow would be useful. ing the waste into four groups of packages overlying four columns for radionuclide transport. Ilowever, c. The waste form / groundwater / tuff interactions. because all waste packages were assumed to fail si-multaneously, the variance in radionuclide releases in the Phase 1 demonstration, the dissolution may have been underestimated. A more inclusive of the waste form was based on a simpic model and mechanistic model of the repository distributed of the solubility of a particular radionuclide in in space and time should provide a more realistic pic-the groundwater, A more complex, compre-ture of the dependence of repository performance hensive, realistic treatment of the dissolution - on various parameters and on various components. of the waste form, that considers the complex -Improved modeling co;ld be accomplished by ex-interactions of the waste form, the host rock, tending some of the methods used in the Phase 1 and the groundwater, would help to determine demonstration, the accumcy of the simpler modeling ap-proaches. 10.2 Improvements and Extensions to d. The degradation of the waste package. Auxiliary Analyses In the Phase 1 demonstration, a non- - The following are recommended improvements to and rnechanistic model of waste-pqckage degrada-tion was used. An essential mgredient of a extensions of the auxiliary analyses. These appear to be m re reahstte treatment would be to consider

important. aspects of a performance assessment, requir-the geochemical intert.ctions among the canis-ing more detailed study, which were not within the scope ter, host rock, and groundwater. An auxiliary of Phase 1' analysis of this type could indicate important par mets outstandng psdons mgaMng T.

Perform detailed gecchemical anal 8es to investi-phenomenology, and the directions for addi-Y i. . gate tional work to take, I a.' The use of Kos (distribution coefficients) in The oxidation of the spent fuel matrix. c estimating radionuclide transport. l In the Phase I demonstration, oxidation of the in the Phase 1 demonstration, Kos were used spent fuel matrix was a phenomenon important in estimating the transport of radionuclides. in determining the behavior of he source term, t llecause of the complex and time-e(msuming especially the gaseous phase releases of C-14. nature of detailed geochemical analyses, which Various empirical and semi-empirical ap. are an alternative to the Ko approximation, proaches were employed to dascribe this phe-l additional modeling efforts are likely to use the nomenon. Detailed geochemnal analyses of l. Ko approximation. Therefore, an auxiliary the rate of spent fuel matrix o>idation and its p analysis to show how appropriate this approxi-dependence on temperature and geochemical NURI!G-1327 90

10. Preliminary Suggestions conditions would help to determine how well ternative hypotheses regarding hydrology at Yucca this phenomenon is understood and whether Mountain.

the modeling should be improved. In the Phase 1 kmastration, the hydrologic analy-f. The geochemical behavior of plutonium. sis consisted of a one-dimensional, steady-state ap-proximation of the unsaturated flow conditions at In the Phase 1 demonstration, plutonium ap-Yucca Mountain. Detailed hydrologic analyses that peared to be a major contributor to the total evaluate the applicability of these and other assump-system performance measure, the CCDF. An tions (e.g., vertical flow downward, a fixed water-auxiliary analysis, to evaluate the adequacy of table location) and the effects of regional flow condi-the modeling of plutonium transport and to tions could suggest ways to improve modeling of determine whether the geochemical data base repository performance. for plutonium interactions with tuff is ade-quate, would be useful. The geochemical be-havior of plutonium in th e near field would also 10.3 Recommendations for Additional l be a useful subject of study. Scientific Input 2. Evaluate heat effects at early times and estimate the The following are recommendations for additional scien-thermal, hydrologic, and geochemical environment tific input, some of which could be performed by either of the repository at early times. DOE or NRC, whereas others are clearly the responsibil-ity of DOE). Rese suggestions were clearly beyond the In the Phase 1 denionstration, the calculated per-scope of the Phase 1 effort, but were identified as gaps in formance did not explicitly take into account the knowledge, as the work in Phase 1 progressed. thermal, hydrologic, and geochemical conditions of the repository at early times, and how such condi-1. Develop and demonstrate a mathematically rigor- ~ tions might affect performance. Consequently, the ous, scientifically robust method for scenario analy-design, environmental, and site conditions that in-sis. fluence these conditions were not explicitly mod-cled. An auxiliary analysis of these complex interac-In the Phase 1 demonstration, an attempt was made tions could help to determine which phemomena to follew the methodology for scenario analysis de-and parameters should be included in improved veloped by SNL Conceptual and logical problems models of repository performance. were encountered when attempting to define, enu-merate, and screen scenarios. A more mathemati-3. Evaluate the importance of thermally-and cally rigorous, scientifically robust approach to sec-barometrically-driven air flow on repository per-natio analysis would streamline the interactions formance at Yucca Mountain, between modelers and various scientific disciplines and would permit a more transparent, direct deriva-In the Phase 1 demonstration, the flow of ground-tion and presentation of results. water - was calculated using a simple, one-dimensional flow approximation that did not include 2. Obtain geoscience input foi modeling volcanism. interaction with fluids in the gaseous phase. The SCP, NRC's Site Characterization Analysis (NRC, During the Phase 1 demonstration, some considera-1989), and other documents (including several re-tion and evaluation was given to the scientific bases viewed as part of the Phase 1 effort) indicate that the available to model the morrence and manifestation barometrically-and thermally-driven flow of air and of volcanism. Altb.,ogh some information was iden-water vapor at Yucca Mountain may have a signifi-tified regarGg previous occurrences of volcanism cant impact on the movement of groundwater and, at Yucca Mountain, the physical mechanisms for therefore, may have a potential impact on repository predicting site-specific volcanism at Yucca Moun-performance, An auxiliary analysis on the nature of tain appear to be poorly understood. Additionalin-such gas flows and their impact on the movement of formation was identified regarding how different groundwater at Yucca Mountain could indicate types of volcanic events might be manifested within whether these effects should be included explicitly or near to a repository. It would be useful to perform in models of repository performance-a comprehensive review of potentially valuable lit-erature, as well m to consider what additional gen-4. Perform detailed hydrologic analyses for T ucca eral and site-sp.fic information and original re-Mountain, to provide a better input to the transport search are needed to estimate the likelihood and analysis and to examine, in more detail, various al-consequences of volcanism at Yucca Mountain. 91 NUR!!G-1327

10. - Preliminary Suggestions 3.

Obtain geoscience and hydrologic input for model-which phemomena and parameters to include in im-ing faulting, uplift, and subsidence at Yucca Moun-proved models of repository performance, execution tain. of such auxiliary analyses appears to be limited by the lack of phenomenologicalinformation and data During the Phase 1 demonstration, tectonic events available for tuff. Additional field and laboratory ex-and processes such as faulting, uplift, and subsi-periments could provide needed data. dence were identified as potentially important fun-damental events that should be considered in defin-6. Obtain more data on plutonium geochemistry, ing 'and selecting scenarios for a performance assessment of a Yucca Mountain repository. Al-In the Phase 1 demonstration, plutonium was a ma-though some substantial information has been com-jor contributor to the total system CCDF. An expan-piled (e.g., in the SCP) on~ these processes and sion of the geochemical data base for plutonium in- - events in the tectonic province and in the iminediate teractions with tuff may be useful. vicinity of Yucca Mountain, additional field data and

other original research may be needed A more com-7.

Obtain a better understanding of waste-package cor-prehensive review of applicable literature and the rosion in the unsaturated zone. identification of additional data needs would be use-ful. In the Phase 1 demonstration, a selected distribution of waste-package failure was used, in large part be-4. Obtain laboratory chemimi analyses to determine cause few analyses and data exist that treat the cor-the partitioning of radionuclides in various compart-rosion of waste packages in a partially saturated re-ments of the spent fuel waste form. pository. On-the basis of the literature teview performed as part of Phase 1, it appears that addi-During the Phase 1 demonstraJon, an important is-tional phenomenological data are needed before

sue regarding the behavior of spet fuel as a waste waste-package corrosion in the unsaturated zone form was the quantity of various ra$onuclides in can be modeled.

various compartments of this complex etc form. Spent fuel can be considered to consist of at Mast 8. Obtain field and laboratory data and perform analy-five different compartments (proceeding from out ses to investigate the issue of non-vertical flow at side in):(1) crud adhering to the outer surface of the Yucca Mountain. . cladding. (2) the cladding, (3) the gap between the E cladding and fuel' pellets, (4) the intergranular An assumption used in the Phase 1 demonstration . spaces in the fuel matrix, and (5) the fuel matrix it-transport calculations was that flow moved vertically self.The rate of release of a particular radionuclide downward in four columns underlying the reposi- ' depends on the compartment in which it is located, tory. An auxiliary analysis, performed in Phase 1 to

because of the physical and chemical form it may be evaluate the potential for non-vertical flow (Appen-in and because compartments closer to the geo-dix G), indicated that non vertical flow might occur, sphere may release their radionuclide inventory

- tmder certain conditions. Nonvertical flow could af-first. Ris consideration appears to be important in - fect radionuclide transport and groundwater travel determining the rate and quantity of C-14 release, times. Therefore, it appears that additional field and .However, very little data on the inventory of various laboratory data and additional analyses on the po-radionuclides in these different compartments were tential for non vertical flow would be useful. identified. His lack of data limited the Phase 1 analysisJ 9. Obtain field and laboratory Pta on the transport of gaseous radionuclides, especially C-14, at Yucca 5. Obtain field and laboratory data on phenomena im-Mountain. L portant to the near-field behavior of the repository, = especially the effects of heat. ~ In the Phase 1 dememstration, the release of C-14 and other gaseous radionuclides along the gas path. _Although the Phase 1 demonstration explicitly took way was not explicitly incorporated into the total sys-into account the thermal, hydrologic, and geochem-tem CCDF An auxiliary analysis executed in Phase ical conditions in the near-fictd of the repository and 1 (Appendix D) indicated that the release of these how such conditions might affect performance, con-radionuclides in the gas phase may be important. An siderations of such complex, near-field interactions obstacle to the realistic modeling of such releases is was limited to rudimentary, frequently nonmech-the lack of general and site-specific data on gaseous anistic, modelingc Although an auxiliary analysis of radionuclide transport. Additional data would be these complex interactions could help to determine useful. NUREG-1327 92

R. REFERENCES Amter, S., et al.,1988,"Carbsm-14 Traveltime at Yucca Campbell, J.E., R.L Iman, and M. Reeses,1979, " Risk Mountain," Appendix to "An Example Postclosure Risk Methodology for Geologic Disposal of Radioactive Assessment Using the Yucca Mountain Candidate Site," Waste: Transport Model Sensitivity Analysis." P. Doctor, et al.,1992, PNL-SOS), Pacific Northwest Laboratories, Richland, Washington. Campbell, J.E. and D.12mpsine,1989, " Application of Generic Risk Assessment Software to Radioactive Waste Anderson, LA.,1981a," Rock Property Analysis of Core Disposal (Draft)," SAND 89-1419C, Sandia National Samples From the Calico Hills UE25a-1 Borehole, Ne. laboratories. Albuquerque, New Mexico. vada Test Site, Nevada." USGS OFR 81-1337, U.S. Geo-logical Survey, Reston, Virginia. Carlos, B.1985. " Minerals in Fractures of the Unsatu-rated Z(me from Drill Core USW-G-4, Yucca Mountain, Anderson, LA.,1981b," Rock Property Analysis of Core Nye County, Nevada," LA-10415-S, UC-70, Ims Alamos Laboratories, Los Alamos, New Mexico. Samples From the Calico Hills UE25a-1 Borchole, Ne-vada Test Site, Nevada " USGS OFR 81-1338, U.S. Geo-logical Survey, Reston, Virginia. C, enter for Nuclear 'Vaste Regulatory Analyses,1988, " Fast Probabilistic Performance Assessment (FPPA) Anderson, LA.,1984," Rock Property Measurements on "I"' polog Waluadon," CM AMN, San Artm t I '*" *' large-Volume Core Samples From Yucca Mountain USW GU-3/G3 and USW G-4 Boreholes, Nevada Test Cochran, W. G.,1963, Samphng Techmques, John Wiley Site. Nevada" USGS OFR 84-0552, U.S. Geological and Sons. Survey, Reston, Virgmia. Cole, C.R., et aL 1987, " Lessons Learned from the Baybutt, P., D.C. Cox, and R.E. Kurth,1981, "Methodol-liYDROCOIN Experience," Proceedings of GEOVAL ogy for Uncertainty Analysis of Light Water Reactor 1987 Symposium, Swedish Nuclear Power Inspedorate, Meltdown - Accident Consequences," Topical Report Stockholm, Sweden, from Battelle Columbus Laboratories to U.S. Nuclear Regulatory Commission. Cox, D.C. and P. Baybutt,1981, Risk Analysis, Vol. I No. 4, p. 251, Beahm, E.C., W.E. Shockley and O.LCulberson,1985, " Organic lodide Formation Following Nuclear Reactor Cranwell, R.M. and Helton,J.C., March 9-13,1981,"Un-Accidents," NUREG/CR-4327 U.S. Nuclear Regula. certainty Analysis for Geologic Disposal of Radioactive tory Commission, Washington, D.C. Waste." Proceedings of:ymposium on Uncertainties Asscci-ated with the Regidation of the Geoksic Disposal of High-l Bear, J.,1979, Hydraulics of Groundwater, McGraw-Hill, Lercl Radioactive IVaste, Gatlinburg. Tennessee, l New York, New York. March 9-13,1981, Conf-810372, ed. D.C. Kocher, pp. 131-144. Bixler, N.E.,1985, "NORIA-A Finite Element Com. puter Program for Analyzing Water, Vapor, and Energy Cranwell, R.M, et al.,1982," Risk Methodology for Geo-Transport in Porous Media," SAND 84-2057 Sandia Na-logic Disposal of Radioactive Waste: Scenario Selection Procedure," NUREG/LR-1667, U.S. Nuclear Regula-tional laboratories, Albuquerque, New Mexico. tory Commission, Washington,. D.C. Bonano, E.J., et al.,1989," Demonstration of a Perform' Czarnecki, J.B.,1995, " Simulated Effects of increased ance Assessment Methodology for liigh 1 evel Radioac-Recharge on the Ground-Water Flow System ot Yucca tive Waste Disposal m Basalt Formations," NUREG/ Mountain and Vicinity, Nevada-California," Water-CR-4759, U.S. Nuclear Regulatory Commission, Resources investigations Report, USGS WRI S4-4344, Washington, D.C. U.S. Geological Survey, Reston. Virginia. Boyars, C., et al.fl985," Demonstration of Methodology - Doctor, P., et al.,1992. "An Example Postclosure Risk. for Waste Package Performance Assessment," WPR-86 Assessment Using the Potential Yucca Mountain Site," (6810-01)-81, Aerospace Corporation. PNL-8081, Pacific Northwest laboratories, Richland, Cacuci, D.G., et aL 1980, " Sensitivity Heory for GeneraI System of Nonthear Equations." Nuc/ car Science and En-Dudley. A.L. et al.,19S8, " Total System Performance gineering 75,88. Assessment Code (TOSPAC), Volume 1: Physical and 93 NUREG-1327

References Mathematical Bases" SAND 85-0002, Sandia National 'through Porous Material," S AND84-OS78, Sandia Na-Laboratories, Albuquerque, New Mexico. tional laboratories, Albuquerque, New Mexico. Einsiger, R.E. and R.E. Woodley,1985a, " Evaluation of flayden, N.K.,1985, " Benchmarking NNWSI Flow and the Potential for Spent Fuel Oxidation under Tuff Re-Transport Codes: Cose 1 Results." Sandia National pository Conditions". HEDL-74$2, Hanford Engineer, laboratories, SANDS 4-0996, Albuquerque, New Mex. ing Develonmem laboratory, Richland Washington. ico. Einsiger, R.E. and R.E. Woodley,1985, *Iow Ternpera. Helton, J.H.,1961, SIAM Rcriew,12, I. ture Spent Fuel Oxidation underTuff Repository Condi-tions " HEDL-S A-3271, llanford Engineering Develop. Hodgkinson, D.P., and T.J. Sumerling,1989, "A Review ment laboratory, Richland Washington. of Approaches to Scenario Analysis for Repository Safety Assessment," paper presented at NEA Symposium on Einsigr, R.E., and R.E. Woodley,1987, " Predicting Safety Assessments for Reposnories. Pam, I rance. Spent Fuel Oxidation States in a Tuff Repository," HEDL-SA-3267, Westinghouse Hanford Company, Huyakom, P.S.,1989 "VAM2D-Vanably Saturated Richland, Washington. Analysis Model in Two Dimen sion s" N U R EG /CR-5351, U.S. Nuclear Regulatory Com miWon, Washmgton, D.C. Einziger, R., and R Kohli,1984, " Low Temperature Rupture Behavior of Zirealloy-Clad Pressurized Water Iman, R.I..,19SO," Stepwise Regrewn with PRESS and Reactor Spent Fuel Rods under Dry Storage Conditions,. Rank Repression, S AN DMf 147., Sandia national Nuclear Technology, vol. 67, pp.107-123. Iab rat ries, Albuquerque, New Mexico. Engel, D.W., et al.,1989, "The AREST Code: User's f*^* """? Guide for the Analytical Repository Source Term 9) Model," PNL-6645, Pacific Nonhwest laboratories, Richland, Washington. Iman, R., W. Conover, and J. Campbell.1960, " Risk Methodology for Geoloric Disposal of Radioactive Englund, E. and A. Spnks,1988, ' GEO-EAS (Geosta-Waste: Small Sample Sensltinty Anahsis Techniques for tistical Environmental Assessment Software) User's Computer Models, wah an Ap' plication to Risk Assess-Guide," EPA /600/4-88/-33a, U.S. Environmental Pro-tection Ageng, Washington, D.C. Inent," NUREG/CR-1397. U.S. Nuclear Regulatory Commission, Washington, D.C. ~ Fischer, F, and J. Ehrhardt,1985, " Uncertainty Analysis Iman, R., J. Davenport, and D. Zeiglet, 19N), "I atin [ with a View Towards Applications in Accident Conse-Hypercube Samplino (Program User's G uide)," quence Assessments " Kernforschungszentrum Karlsruhe, SkND79-1473, Sandb National laboratories, Albu. kfk 3906. querque, New Mexico. Freeze, R.A. and J.A. Cherry, 1979, Groundwater, Iman, R.L and J.M. Davenport,19M, "An Iterative Al-Prentice-Hall, Englewood Cliffs, New Jersey. gorithm to Produce a Positive Definite Correlation Ma-trix from an Approximate Correlation Matrix (with a Pnw

Gelhar, L.W.

and C.L Axness, 1983, "'lhree-gram User's Guide)," SANDS L-1376. Sandia National Dimensional Stochastic Analysis of Macrodispersion in labon. tories Albuquerque, New Mexico. Aquifers," Water Resources Research, Vol.19, no.1, pp.161-180. Iman, R.L-. and MJ. Shortencarier,1984 "A I ORTRAN 77 Program and User's Guide for the Generation of12 tin Gelhar, LW. and A. Mantoglou,1987, " Stochastic Mod-Hypercube and Random Samples for Use with Computcr eling of large. Scale Transient Unsaturated Flow Sys-Models," NUREG/CR-3624 (SANDN3-2365), U.S. Nu-tems," Water Resources Research, Vol. 23. no. 1, clear Regulatory Commission Washington, D.C. i pp.37 46. International Formulation Committee,1467. A Formula-Grambow. U.,1989," Spent Fuel Dissolution and Oxida-tion of the 7hermodynamic Properties of Ordinary Water tion-An Evaluation of Literature Data," SRB 89-13, Substance, YC Secretariat. Dusseldort'. Germany. Swedish Nuclear Fuel and Waste Management Com-pany, Stockholm, Sweden. Katz, JJ, and E. Rabinowitz,1951, The Chemistry of Uru-nium-[he Element. Its Smary and Related Compounds, Hadley. G.R.,1985, "PIsl ROS-A Program for Calcu-Naticaal Nuclear Energy Series Dmsion VIII, Vol.5, lating Traasport of Heat, Water, Water Vapor, and Air Do' er Publications inc, New York NUREG-1327 94

n References Killough, G.G., and J.E. Till,1978, " Scenarios of "C Montazer, P., et al.,1985, ' MonitoringThe Vadose Zone i Releases from the World Nuclear Power Industry and in Fractured Tuff, Yucca Mountain, Nevada " Proceed-Estimated Radiological Input," Nuclear Satety, Vol.18, ings, National Water IVell Association Conference on Char- = no.5, pp. 602-17. acterization and Monitoring of the Vadose (Uracturated) Zone. Denver, Colorado, November 19-21,1985. Knapp, R.,1987, "An Approximate Calculation of Ad-vective Gas Transport of C-14 at Yucca Mountain, g Oblow, E.W.,1978, Nuclear Science and Engineering, Vol. vada," UCRlr97805 (preprint), Lawrec= Livermore 65, p.187.- - National Laboratory, Liver'noc, california. 0prd, A., et al.,1983, "Are Solubility Limits of Impor-tance to Leaching?," Scientific Basis for Nuc/ car Waste 12houd, D.H., et ali,19F4, "Geohydrolog: of Volcanic anagment, ateria escarch Sxktyheedmgs, M 3, Tuff Penetrated by Test Well UE25b#1, Lea Moun. pp.3 1-331 tain, Nye County, Nev.da," USGS WRI 84-4253, U.S. Geological Survey, Reston, Virginia. Parkhurst, DL, D.C. 'lhorstenson, and L.N. Plummer, 1980, "PHREEQE-A Computer Program for Geo- - Lee, W;. April 6,1989, Letter to Dr. N. Eisenberg. detail-chemical Calculations," Water. Resources Investigations ing codes deveivped v UCB on file at the National En-Report 80 96, U.S. Geological Survey, Reston, Virginia, ergy Software Center, Parsons, A.M., et al.,1990, "conceptuali/ation of a Hy-Liebetrau,' A., et al.,1987, "The Analytic Repository pothetical liigh-Level Nucicu Waste Repository Site in g Source Term Code for Waste Package Performance," Unsaturated, Fractured Tuff," NUREG/CR-5495, U.S. PNL-6346, Pacific Northwest Iaboratories, Richland, Nuclear Regulatory Commission, Washington, D.C. Washington. Peters, R.R., et al.,1984, " Fracture and Matrix liydro- ' Lin, Y,T.,1985,1" SPARTAN-A Simple Performance logic Characteristics of Tuffaccous Materials From Yucca Assessment Code for the Nevada Nuclear Waste Storage Mountain, Nye County, Nevada," S AND84-1471, Sandia Investigations Project," S AND854602, Sandia National National laboratories, Albuquerque, New Mexico. --I2boratories, Albuquerque, New Mexico. Plummer, l_N., B.F. Jones, and A.H. Truesdell,1976, "WATEQF-A Fortran IV Version of WNITiQ, A Com-L. Y,T and M.S.Tierney,1986,"PreliminaryEstimates puter Program for Calculating Chemical Equilibrium of in, of Groundwater Travel 'Itme and Radionuclide Trans; Natural Waters," Water-Resources investigations Re-port at: the Yucca Mountam Repository Site, port 76-13, U.S. Geological Survey, Reston, Virginia. . SAND 85-2701, Sandia National laboratories, Albu- _ querque, New Mexico. Prindle, R.W.,1987," Specification of aTest Problem for HYDROCOIN lxvci 3 Case 2: Sensitivity Analysis for lengsine, D.E... EJ. Bonano, and C.P. Harlan,1987, Deep Disposal in Partially Saturated. Fra'etured Tuff " " User's. Manual for the NEFI'RAN Computer Code," S AND86--1264, Sandia National I.atxm tories, Albu- - NUREG/CR-4766, U.S. Nuclear Regulatory Commis-querque, New Mexico. - sion,~ Washington, D.C. Pruess, K.,1987, " TOUGH User's Guide," NUREG/ McKay, M.D., WJ. Conover, and RJ. Beckman,1979, CR-4645, U.S. Nuclear Regulatory Commission Wash- "A Comparison oflhree Methods for Selecting Values of ington, D.C. Input Variables in the Analysis of Output from a Com-puter Code," Technometrics '21, pp. 239-245. Rasmussen, T,C., and D.D. Ev;ms,1987, " Unsaturated Flow and Transport Through Fractured Rock Related to ~ Meyers, R.H.1971, Response Smface Methodology, Allen High-1 evel Waste Repositories, Final Report-Phase . and Bacon, Inc., Boston, Massachusetts. N'" NUREG/CR-4655, U.S Nuclear Regulatory Com-mission, Washington, D.C.

Miller, C.W.,

1983, "CHEA1TRN User's Manual," Reda, D.C.,1987, " Influence of Transverse Microfrac-Pub-3031, _ Lawrence Berkeley Laboratory Berl<cley, tures on the Imbibition of Water into Initially Dry Tuf-California; faccous Rock," Geophysical Monograph 42, American L Montazer, P.and W. Wilson,1984," Conceptual Model of Flow in the Unsaturated Zone, Yucca Mountain, Ne-Reed, D.T., et al.,198~7,"liffects of lonizing Radiation on vada," USGS WRI 84-4345, U.S. Geological Survey, Moist Air Systems" UCRL-97936, lawrence l.ivermore Reston, Virginia. National laboratory, Livermore, California. i. 95 NURiiG-1327

__ References ' Reed, D.T., and R.A. van Konynenburg,1987, " Effects oi Travis, BJ.,1984, "TRACR3D: A Model of Flow and Ionizing Radiation 'on Moist Air. Systems," Afaterial Transport in Porous / Fractured Media," LA-9667-MS, Research Society Symposium on the Scientific Basis for Nu-Ins Alamos National Laboratory,1xs Alamos, New Mex-clear IVaste Afanagement XI, Boston, Massachusetts. ico. Rice, W.A.,1984, " Preliminary Two-Dimensional Re-Tsang, Y,W. and K. Preuss,1987, "A Study c' Thermally gional Hydrologic Model of the Nevada Test Site and Induced Convection Near a Ifigh-Level Nuclear Waste Vicinity," SAND 83-7466, Pacific Northwest I aboratory. Repository in Partially Saturated Fractured Tuff " Water Robinson, P.C. and D.P Hodgkinson,1986, " Exact Solu-tions for Radionuclide Transport in the Presence of Pa. Updegraff, C.D.,1989, " Comparison of Strongly Heat-rameter Uncertainty," AERE R 12125, United Kingdom Driven Flow Codes for Unsaturated Media," NUREG/ - Atomic Energy Authority (Harwell 12boratory), Ox-CR-5367, U.S. Nuclear Regulatory Commission, Wash-fordshire, England, ington, D.C. Ross, B.,1988, " Gas-Phase Transport of Carbon-14 Re-U.S. Code of Federal Regulations, " Environmental Ra-leased from Nuclear Waste into the Unsaturated Zone," diation Protection Standards for Management and Dis-Afaterial Research Society, Symposium Proceedings, vol. posal o(Spent Nuclear Fuci, High-Level and Transuranic 112. Radioactive Wastes," Part 191, Subchapter F, Title 40. Rush, F.E.,1970,'

• Regional Ground-Water Systems in U.S. Code of Federal Regulations, " Disposal of High-the Nevada Test Site Area, Nye, Lincoln,- and Clark Level Radioactive Wastes in Geologic Repositories" Counties, Nevada," Water Resources Reconnaissance Part 60, Chapter I, Title 10, " Energy."

Series Report $4, Department of Consenation and Natu-ral Resources, State of Nevada, Carson City, Nevada. U.S. Department of Energy,1986, " Environmental As-sessment, Yucca Mountain Site, Nevada Research and Sass,J,and A.Lachenbruch,1982,"PreliminaryInterpre-Development Area Nevada " DOE /RW-0073. - tation of Thermal Data From the Nevada Test Site," USGS OFR 82-973, U.S. Geological Survey, Reston, U.S. Department of Energy,1988, " Site Characterization Virginia; . Plan, Overview, Yucca Mountain Site, Nevada Research and Development Area, Nevada," DOE /RW-0198. Sinnock; S., Y,T. Lin, and J.P. Ilrannen,1984, "Prelimi- ! nary Bounds on the Expected Postclosure Performance of U.S. Department of Energy,1988 " Site Characterization the Yucca Mountain Repository Site " Southern Ne-Plan: Yucca Mountain Site, Nevada Research and Devel-vada," S AND84-1492, Sandia National Laboratories, Al-opment Area, Nevada " DOE /RW-0199. - buquerque, New Mexico. U.S. Environmental Protection

Agency, 1983, Sinnock, S., et al.,1986, " Preliminary Estimates of "REPRISK Code Documentation," prepared by Arthur GroundwaterTravel Time and Radionuclide Transport at D. Little, Inc., Cambridge, Massachusetts.

the Yucca Mountain Repository Site," SANDS 5-2701, Sandia National Laboratories, Albuquerque, New Mex-U.S. Nuc! car Regulatory Commission, Draft Generic ico. Technical Position " Interpretation and Identification of - the Extent of the Disturbed Zone in the High-Level Stumm. W., and J. Morgan,1970, Aquatic Chemistry, Waste Rule (10 CFR 60)," Notice of Availability in Fed-Wiley-interscience, New York. - eral Register. Vol. 51, Na 132, 25125 (J uly 10,1986). --Thomas, K.,1987, " Summary of Sorption Measurements U.S. Nuclear Regulatory Commission,1989. " Severe Ac. LPerformed _with Yucca Mountain Tuff Samples and cident Risks: An Assessment for Five U.S. Nuclear Power Water from Well J-13," L.ASL LA10960MS, Los Alamos Plants," NUREG-1150 Washington. D.C. National Laboratory, los Alamos, New Mexico. van Genuchten, M., and PJ. Wierenga,1976, " Mass Transfer Studies in Sorbing Porous Media," Soil Science -'Ihomas, S.D., et al.,1982,-"A Summary of Repository Siting Modeis, NUREG/CR-2782, U.S. Nuclear Regu-Society of,4merica Joumal, 40 (4), pp. 473-480. latory Commission, Washington, D.C. van Konynenberg, R., et al,1984," Behavior of C-14 in Thompson, J.L.1989, " Actinide Behavior on Crushed Waste Packages for Spent Fuel in a Repository in Tuff," Rock Columns " Joumal of Radioamdyrical and Nuclear UCRL-90855, Lawrence Livermore National Labora-Chendstry, Vol.130, no.2, pp. 353-364. tory, Livermore, California. NUREG-1327 96 ,.. ~.

m References van Konynenberg, R., et al.,1987. " Carbon-14 in Waste Tuff Test Well USW 11-1, Yucca Mountain, Nevada " - Packages for Spent Fuel in a Tuff Repository," in Scien. USGS WRI 84-4193, U.S. Geological Survey, Reston, tific Basisfor Nuclear Waste Afangement AfaterialResearch Virginia. - Society Proceedings, ed. J.K. Bates and W.B. Seefcidt, Vol. 84,pp.185-196, Wilson, W.E.,1985, Letter from W.E. Wilson (USGS) to D.l_ Vieth (DOE /NVO), Decernber 24,1985; regarding unsaturated zone flux,

Voss, C.I.,1984,"S'JI'RA Saturated-Unsaturated Trans-pon-A Finite Element Simulation Model for Saturated-Winograd, I.,1981," Radioactive Waste Disposal in Thick Groundwater Unsaturated, Fluid-Density-Dependent,

Unsatumted Zones," Science, Vol. 212, No. 4502, Flow with Energy Transport or Chemically-Reactive June 26,1981* . Single-Species Solute Transport," USGS WR184-4369, - U.S. Geological Survey, Reston, Virginia. Wolery,TJ.,1979, " Calculation of Chemical Equilibrium between Aqueous Solution and Minerals: The 1103/6 Weeks, E.P. and W.E. Wilson,1984, " Preliminary Evalu-Software Package," UCRIe52658, lawrence Livermore ation of Hydrologic Properties of Cores of Unsaturated National 12boratory, Livermore, California. 97 NUREG-1327

_= APPENDIX A -SYSTEM CODE REVIEW A.1 Introduction standard. The code does not take into account various scenanos. The following discussion provides summaries of several programs evaluated by the staff to determine their suit-A.2.3 TOSPAC ability, as a whole or in part, for use as a system code for - the Phase 1 demonstra* ion. Not all the programs pre. Sandia National laboratorics developed TOSPAC (Dud-sented are system codes per se, but each contained cle. ley, et al.,1988) for DOE, specifically for the Yucca ments considered necessary to the approach used in this Mountain, Nevada site. The model considers the one-

effort, dimensional, transient, unsaturated flow and transport of solub!c waste materials with coupling between the matrix and fractures.

A.2 Program Summaries The code is a FORTRAN 77 program that uses various A.2.1 AREST modules to manage the input and output tasks and to model the differential equations governing water flow, The ARF 50 code (Engel, et al.,1989)was developed by radionuclide transport, and liquid-phase mass transport. Pacific Northwest laboratory for the U.S. Department of A management driver oversees the interactions between ' Energy (DOE). The program takes a modular approach these modules. Input to the code covers the material to the problem of making preliminary, quantitative per. properties of the geologic strata, the radionuclide proper-formance assessments of the engineered barrier and ties, and different boundary conditions. Output consists near-field systems. Input variables to the code include of release over time, radionuclide concentrations in the values assigned to the spent fuel waste package, as well as matrix and fractures versus time, and three-dimensional to variables describing the physical and chemical environ-plots of concentration versus time versus distance. ments of the repository /near-field system and the waste package. A.2.4 REPRISK AREST models the performance of the assemblage of REPRISK (EPA,1983)is an hi.* orogram that models individual waste packages from repository closure to the the long-ternt radionuclide release angpulation health failure of the canister, the release of radionuclide-s from effects associated with the disposal of highevel radioac-the failed packages, and the subsequent movement of the tive wastes in mined geologic repositories,it as ongi-radionuclides away from the waste packages. Average nally developed for a repository located m a saturated, release rates and cumulative releases over time can be p rous salt media and can address variations in geologic - calculated from successive waste-package simulations. setting, radionuclide inventories, radionuclide-release mechanisms and pathways, time frames, and dose-uptake l The code cannot be considered as a total system code, pathways. I since it treats only various failure mechanisms for the The code handles four designated " release mechanisms:" waste packages and t ot the possible scenario classes cre-(1) direct impact of a waste package with release to air and ating the conditions for failure. land; (2) direct impact of a waste package with release to an aquifer;(3) disruption of the repository with release to - A.2.2 SPARTAN land; and (4) disruption of the repository with release to an aquifer. REPRISK does not treat radionuclide decay SPARTAN (Lin, 1935) is a simple performance" chains and does not incorporate a random sampling pro-assessment model developed by Sandia National Labora-gram (like I atin Hypercube Samp'ing (1.HS)) or any sen-tories to support DOE's Environmental Assessment of a sitivity and uncertainty analyses, potential repository at Yucca Mountain, Nevada. Input, consisting of repository, hydrogeologic, waste-package, Consequences of a release to the accessible emironment and spent fuel characte istics, is used to simulate the can in expressed as somatic or genetic health effects, a one-dimensional, dispersionless transport of radio-ratio of release amount to limits set in 40 CFR Part 191, nuclides in both a porous matrix and a fractured media. and/or total curies released per radionuclide. Radionuclide release rates and cumulative curies re-A.2.5 SUNS leased are calculated. From this, the performance of the repository can be measured relative to the U.S. Nuclear The Sensithity and Uncertainty Analysis Shell (SUNS) Regulatory Commission's (NRC) performance objectives (Campbell and longsine,1989) is a Sandia National - and to the U.S. En ironmental Protection Agency (EPA) Laboratories generic software shell created to perform-A-1 NUREG-1327

- =. Appendix A Monte Carlo and LHS analyses. It is a modular, menu. pler programs (Bonano, et al.,1989) to provide linkage driven code with a flexible input editor that can incorpo-between a suite of Sandia codes for a total system per-rate a variety of application models suitable for such. formance assessment.This linkage is given on two scales: analyses. The user provides replacement statements to (1) between the regional and local flow rnodels; and equate model variable names to kutions in the various (2)between the local flow model and the radionuclide SUNS arrays. The program is designed for parametric transport model. analyses and correlation studies. SUNS performs all file-management operations. Output LilS is used to create a common database for input in is available in both statistical and graphical formats. ' order to maintain a consistent description of the system for each of the models. Programs are available to plot A.2,6 Code Coupler Programs estimated flow paths, discharge rates (in ' curies per day) versus time (in years), and complementary cumulative Sandia National Laboratories developed the Code Cou-distribution functions (CCDFs). i l NURiiG-1327 A-2

i APPENDIX B-SOURCE TERM CODE REVIEW B.1 Introduction assumed to be from the matrix also, neglecting contribu-tions from the cladding and gap compartments. Interest-This appendix presents reviews of source term models ingly, a DOE screening analysis indicated that most of the used in previous U.S. Department of Energy (DOli) radionuclides would never reach the accessible environ-analyses of the Yucca Mountain repository, and of other ment, except for carbon, technetium and iodine, Since models related to source-term considerations, in general. these are the very elements that tend to collect outside of it covers both dedicated source-term codes such as the UO, matrix, neglecting the other compartments may AREA r, as well as sou rce-term routines in systems codes. be a weakness in this approach. This model appears to be This is not a comprehensive list, but represents a sam-virtually identical to that presented by Lin and Tierney _ pling of codes whose references were available to the (1986). staff. H.2.1.2 The TOSPAC model B.2 Review of Available Source-Term The Total System Performance Assessment Code Models Used for Assessing the (TOSPAC)(Dudley, et al.,1985)is a more sophisticated ne dim nsi n 1 model d y loped by Sandia National Yucca Mountain Proj'ect (YMP) IAmratories that considers transient unsaturated flow Sile and radionuclide transport, with coupling between the 4 matrix and fracturcs The source-term model considers D,2J Early DOE Assessment Models for either the complete dissolution of the UO, matrix, with - Yucca Mountain release of all radionuclides (an extremely conservative assumption) or the more realistic congruent release DOE performed several preliminary, simplified, scoping model. The congruent release model assumes that: (1) assessmeats of the Yucca Mountam site performance. the fractional release rate of radicauclides from the spent Two models used in these assessments were the Environ-fuel inventory is equal to the fractional leach rate of the mental Assessment model and the TOSPAC model. uranium dioxide matrix (2) the rate of the matrix dissolu-tion is a function of the solubility limit of uranium dioxide 11.2.1.1 The Environmental Assessment model and the availability of water; and (3) the transport of dissolved species to the source boundary is instantaneous, The Environmental Assessment (DOE,1986)modelcon-and the transport behavior in the near-field region of the sidered the repository to be composed of three compo-waste package (where the rock is thermally and mechani-i nents: the waste package, the engineered barrier, and the cally disturbed) is similar in the adjacent undisturbed geological barrier.The lifetime of the waste package was rock, The authors neglect radionuclide releases from 3000 to 30,000 years, before which time there would be no compartments other than the UO, matrix, but acknowl-liquid releases of radionuclides. The 3000-year lower-edge the potential imponance of these compartments. bound lifetime was adopted to achieve "some degree of The amount of radionuclide release is limited to being conservatism." The source-term model used in this as-less than or equal to the nuclide's solubility in the water sesenent assumes that there would be congruent dissola-contacting the waste. The authors claim that, in most tion of the uranium dioxide (UO ) matrix, and that the cases, the solubility limit would be greater than the con-2 radionuclide release rate is proportional to the water flow centration, therefore, the release is truly congruent. This past the fuel and the solubility of the matrix, DOE esti-model would not appear to treat daughter products for mated that for an infiltration rate of 0,5 mm per year, a chain decay unless alldaughters had the same solubility, fuel matrix solubility of 0.05 kilograms per cubic meter, and an infiltration area per canister of 0.33 square meters, Dudley, et al. recognize that the assumptions about how there would be a fractional release rate by congruent liquid water contacts the waste to begin the release proc-dissolution of 2.56-9 per year, The model does not take ess is not well understood. They assume that all the water into account solubility limits for released radionuclides, intercepted by a container (which is equal to the product but assumes that with the exception of carbon, cesium, of the infiltration rate and the cross-sectional area of the technetium and iodine, all solubility values would be less canister) becomes saturated with waste. They also recog-than or comparable to the value of the UO matrix.The nize that additional mechanisms may limit the dissolution 2 authors recognize that there are ether sources of radi-of the matrix, e.g., diffusion out of the waste container, onuclides, e.g., in the pellet-cladding gap, the hardware, and that the advection-only model may be pessimistic. and the cladding, but that except for C-14, they argue that the radionuclide inventories wauld not significantly affect Waste canisters arc assumed to fail at a uniform rate, for their results for cumulative release. All C-14 releases are lack of any data on actual failures. 11-1 NUlWG-1327 ~

Appendix 11 I I IL2.2 More Recent Modeling of YMP Site diffusive mass transfer.1 or unsaturated media, the model Performance assumes that the environment is oxidi/ing, and that trans-port is likely to be convective rather than dif f usive. Radi-Doctor, et al. (1992) conducted a preliminary risk assess-onuclides are relea ci from the waste matrix congruently ment for the Yucca Mountain site,in which releases from at a tale given by the forwar+ matrix dissolution and the the enginected barrier were evaluated using the AIESI' fractional inventory of the nuchde in the matttx. The code (Liebetrau, et al.1987)Jihe ARiiST model consists model chooses the larger of the dtffusiselsolubihty re-of three major components: the Engineered System Re-lease rate or the convective release rate.The release rate lease (ESR) model, the Waste Packape Containment may be solubility hmited if the rate of congruent release is high, and the solubility of the released species is low. (WPC) model, and the Waste Package Release OVPR) The model also looks at thc non matnx components of model. The code treats waste packages individually, with no interactions between adjacent waste packages. The the source term, and treats those radumuclides accumu-WPC model simulates corrosion and degradation leadmg lated in the interstices and cladding gap as solubility / to waste-package failure.'the WPR model simulates re-transport-hmited, until the msentory is depleted. The lease of radionuclides and their.nigration outward modelers secognue that um UO rnatrix dissolution may 2 through the waste-packape narriersfl he ESR modelinte-not be truly controlled by solubdity, but, rather, by the grates the simulated releases from individual waste pa& instabthty m an oxidizing environment, so that the rate ages with respect to their failure-time distributionJlhere could remain non zero even wben the solution becomes is also a geochemical model to provide inputs to the three saturated with respect to the matrix. The modelers hm-ited the release of the matrix radionuclides on the basis of major component models. The authors used the concept an oxidized and more-soluble uranium s6heate mineral. of support models external to the AIEST code to per-form site-specific calculations that are too time-consuming or dif ficult to include in the overall simulation. Even if release rate is not controlled by thw solubthty of the matrix, and the radionuclide in question is not The ARl!ST code uses detailed site-specific information solubihty-controlled, the rate of reh'ase might still be about the physical and cNmical environment of the waste controlled by diffusion away from the waste form rather package and the repository.The code describes the ther-than by convection, if the latter is very small. The models mal, geochemical, and hydrological environments of the allows for certain of the radionuclides to form colloidal simulated waste package.The geochemical model deter-species. Ihlfusion of colloids might also limit their release mines the chemical environment of the waste package. for very low flow rates. Since collmds has e much smaller The hydrologic model for the unsaturated case deter-diffusion coelficients than molecular spec:es, this rate mmes the time that the waste packages might be rewetted must be very small when the release is diffusion limited. af ter they cool, although it appears that the authors only It is not likely that both diffusion and convection netd to considered porous media and not the possibility of frac-be considered simultaneously for the Yucca Mountain ture flow near the waste package. For saturated condi-case for any single species. tions, the hydrologie model calculates the time to achieve resaturation after repository closure. For unsaturated The WPR model makes no special provision for release 01 media, the thermal model calculates the time for the gaseous radionuclides, such as "CO. It assumes that all 2 waste package to cool to a point where liquid water can this inventory is released upon fmlute of the canister.The come into contact with it. non volatile' radionuclides that are not contained in the ~ matrix generally have high solubilities and do not form Several mechanistic models of uniform and pittmg corro-colloids in oxidizing environments. sion, as well as empirical models derised from site-specific testing, are assumed in the WPC model. The lhe reochetnical model is used to determine the chemical model does not differentiate between canister and clad-environment of the waste packagesflhe modelcalculates ding containment. In thia risk assessment for the Yucca the steady-state equibhnum concentration of J-13 water Mountain site, the authors did not use a mechanistic code in equilibrium with the tuff at different tempcratures and f or waste-package containment. Instead, they chose arbi-m a 3aturated condition. It does not treat radiolysis reac-trarily a normal distribution of failure times, with a mean tion between the water and the corroding ctmister mate-of 1000 years and a standard de-on of 200 years, and rial, the sorption of radionuclides, and water vaporization the lower tail of the distnbution truncated at 300 years. or rew etting.These may be senous omissions that should be tested with support models. In particular, the conse-The WPR model takes separate approaches for the satu-quences of corroston products of the canister and other rated and unsaturated cases. The saturated model as-materials on the rate of corrosion and dissolution of radi-sunies hw oxygen levels (leading to low dissoitition rate onuctides, and the effects of mineral concentrations m for the UO matnx), low radionuclide solubihties, and low the near field resulting from the eftects of heat and 2 groundwater flow rates, so that releases are based on drying, should be tested. NURl!G-1327 11-2

Appendix 11 B.2.3 Other Models Not Developed UCHNE-106-A time-dependent version of Specifically for the YMP Site UCHNE-107. UCHNE-106D-Calculates the time history of the 11.2 3.1 NEITRAN diffusion coefficient. NEl'rRAN (lengsine, et al.,1987) implements a Net-UCHNE-106N-Calculates the species concentra-work Flow and Transpe-1 Model developed by Sandra tion in the void water, as a function of time. National Laboratones pnmanly for the modelmg of re-pository performance at saturated sites. NEITRAN ccm-UCHNE-106F-Calculates the fractional release tains taodels for solubility-limited and leach limited rate of the species at the void / rock interface, as a cases. If so desired, the program will determine whether a function of time, particular release is limited by leaching or solubility. A third model, mixing cell, assumes that the radionuclides UCHNE-108-Calculates the mass flux rate and are released into a well-mixed cell.The concentrations of the fractional release rate at the interface between ~ the radionuclides in the cell are governed by flowrate the first layer of porous matenal and the next layer through the cell, volume of the cell, and solubility of the of porous material of solub!c species released in radionuclide spec;es. water-saturated porous media. UCHNE-102-Calculates the mass flux of the non-The source-tenn model follows three radionuclide inven-decaying contaminant outward from a spherical tories. 'Ihe first tracks the total mass of radionuclides waste form, when there is stationary precipitation at remaining in the wast e and is called the "unleached inven' a prescribed distance from the waste separating an tory." The sectmd inventory is "uudissolved." and it is that inner region of higher solubility and an outer region portion which has been released fcom the matrix by icach-of lower solubihty.

ng, but whose release to the geosphere is limited by solubihty.The third inventory represents dissolved radi-In addition to these codes that are specifically for near-onuclides. Releases of radionuchdes from the matrix de-field phenomena, there are a set of U, H codes that pend on the leach rate of the matrix, that is, upon congru-integrate the source-term and the transport models. To ent dissolution Releases become part of the soluble get analytical solutions, the source tenn pan of these compartment if their solubility is greater the the concen-models must be simple, either an impulse (i.e., instanta-tration, or part of the undissolved compartment if vice neous release), a step function in concentration or flux versa. Concentrations of different isotopes of the same (band release), or a concentration boundary. None of element are taken into consideration for solubility limits these models can handle solubility limits, because these by specifying the fraction of the inventories for each iso-are inherently non-linear and cannot usually be solved in tope.

nlosed form. The models can treat the releases of chal-decaying radionuclides in the source, provided that their B.23.2 Exact and asymptotic solutions c ncentrations can be expressed by the Bateman equa-tions and are not distorted by preferential removal of The University of California, Berkeley (UCH), Earth Sci-daughters. ences Division has published a number of computer codes dealing mainly with the closed-form solution of flow and IL233 CONVO diffusive transport from waste packages and through the CONVO is a code developed for the U.S. Nuclear Regu-geosphere (Lee,1989) Some of these solutions have fatory Commission (NRC) to model the performance of been incorporated into the AREST code and the Pacific the waste canisters and engineered barrier system Northwest laboratory (PNL) assessment of L)octor, et al-(Boyars, et al.,1985). The code was primarily developed (1992). Some of the codes that may prove to be useful for for demonstrating compliance with the NRC annual re-defining the source term releases are: lease criteria in 10 CFR Part 60.113, rather than the cumulative release enteria of EPA, as embodied in UCHNE-101 -Calculates the concentration of 10 CFR Part 60.112. CONVO has three models for re-solubility-limited species as a function of space and lease of radionuclides: 4 time, and the mass flux rate from a waste sphere bur-ied in a nuclear waste repository in water-saturated

1) a one-dimensional, duc.1-media model,

2) a three-dimensional. single-medium model, and UCHNE-107-Calculates the fractional release

3) a two-dimensional. cylindrical, dual-media model.

rate of soluble radionuclides that are released from nuclear waste emplaced in water-saturated porous These models assume that the radionuclides are released media. at the surface of the waste package through a porous B-3 NUREG '327

Appendix 11 packing material, and that releases are limited by diffu-

2) The cascade approach, in which the sequential fail.

sion and by solubility. There is no consideration given to urcs of the canister lucking are considered. radioactive decay and the rate at wh;ch the radionuchdes are released from the UO matrix, or to compartments in The code was targeted mostly to a saturated, 7ero-2 the fuel other than the matnx. Release events are consid. schx ity, low solubility groundwater system, in which cred by two approaches: diffusion rather than advection was assumed to be the dominant transport mechanism for the release of radi-

1) The convolution approach, in which the time of peak onuclides from the waste packages. It appears that, in its releases is considered to be independent of the time present form, CONVO would be of httle use as a source-of canister failures.

term model for the Yucca Mountain site. l l NURl!G-1327 11 - 4

i APPENDIX C-FLOW AND TRANSPORT CODE SUMMARIES C.1 Regional Flow Program De current version of VAM2D has ne capability to han. Suntiiiaries die fracture-matrix problems. Future development (i.e., in IM) of the program will include a capability to ac-count for fractures via a composite characteristic curve. g. SUTRA (Voss,1984) solves the equations for fluid C.I.3 TRACER 3D density-dependent saturated or unsaturated groundwater flow, and either transport of a solute m the groundwater 'Dm TR ACER3D program (I ravis,1984) simulates two-pnase mr.ss flow and transport in a three-dimensional, or tmnsport of thermal energy in the groundwater and deformable, heterogeneous, reactive porous medium, solid matrix. Solute transport m groundwater mcludes equilibrium adsorption on the solid matrix, and radio-The program solves the equations for mass conservation I of the liquid and gas and a reduced form of the momen-nuclide production and decay. Additionally, SlJFRA may tum equation. The program has the flexibility to solve be used to examine variable-density leachate movement onc<limensional, single-phase flow problems, or to in-and salt water mtrusion. Although energy-transport ciude features such as additional dimensions (up to three simulations can be performed with SUTRA, the program only simulates the hquid phase, without any consideration dimensions), the gas phase, and solute transport. for phase changes. The partial differential equations are approximated using 'N" P The program uses an integrated finit -difference method "I"

• E
• U" "E " ""

to approximate the governing equations. The finite ele- "*C ment mesh can accommodate arbitrary geometries em-ploying quadrilateral finite elements in Cartesian (one or TRACER 3D does not explicitly account for fractures, two dimensions) or radtal-cylindrical (quast-three dimen-although the geometrically flexible, integrated finite dif-sions) coordinate dimensions. ference approach would allow for discretizing very small elements that would tend to simulate fractures. However, Explicit treatment of fractures is not accounted for,m the the program represents the relative conductivity with the model. However, a dual porosity type of treatment for llrooks and Corey expression that is reasonable for po-simulating fracture-matnx interactions would be possible rous media, but may be unacceptable for fractures, through th: use of a composite characteristic curve. C.1.2 VAM2D C.2 Two-Phase Flow and IIcat Transport Program Summaries VAM2D (Huyakorn,1989) is a two-dimensional, finite element program developed to simulate moisture move-C.2.1 TOUGil ment and solute transport in variably saturated porous media. in solving the goveming equations for ground-TOUGH (Prcess,1987) solves the equations for two. water flow, the 9togram can take into account hysteretic phase flow of air and water in the vapor and liquid phases, moisture characteristics and variable (due to moisture and heat transport in a fully coupled way The formula-content) anisotropy in the hydraulic conductivity of the tion used in TOUGH is analogous to that used in multi-unsaturated media. The program is capable of simulating phase, multi-component, geothermal or steam-flooded the transport of chains of radionuclides that account for hydorcarbon reservoir problems. The governing fluid. retardation phenomena, via a linear equilibrium iso-flow equations account for gaseous diffusion, Darcy flow, therm. capillary pressure, vaporization, and condensation with latent heat effects, and conduction and convention of VAM2D uses a finite element method to solve the flow heat are included in the energy equation. Water, air, and and transport equations. Time integration is performed rack are assumed to be in thermodynamic equilibrium at using implicit finite difference approximations, with non-all times. He flow domain can include liquid, gas, and linearities being handled with either Picard or Newton-two-phase regions, indicating that the code handles both Raphsen iteration schemes. Additionally, the iterative saturated and unsaturated flow problems, eitherindividu-methods employs the Preconditioned Conjugate Gradi-ally or simultaneously. The thermophysical properties of ent (pCG) for solving the matrix equations (The PCG liquid and vaporized water are repremnted by the Inter-method has recently emerged as a very promising tech-national Formulation Committee's (1967) steam-tables. nique for handling the numerical difficulties of ground-Air is approximated as an ideal gas, and additivity of water modeling). partial pressures is assumed for air-vapor mixtures. C-1 NUREG-1327

Appendix C TOUGli solves three non-linear partial-differential dimensional problems, either in linear, radial, or spheri-equations simultaneously. These are the conservation cal coordinates, and solves the equations with the equations for air, water, and heat, Air and water can be finite-difference method. 'lhere are also some differ-transported in either the liquid phase, the gas phase, or ences between the codes in the way the time mtegrations both. The dissolution of air in water is represented by are performed. PETROS uses a modified version of tLe llenry's law and flow (gas and liquid) by Darcy's law, time integrator in NORIA. The code can simulate flow in one, two, or thice dimen. PETROS solves three mass-conservation equations and a sions, because the method of solmion is based on a gen-heat-conservation equation, just as NORIA. Ilowever, eral integrated finite-difference method. Time-stepping the liquid conservation c ;uation in PETROS is forrou-is accomplished by a fully implicit procedure. The result-lated with respect to saturation rather than pressure, as in ing non-lineat difference equations are linearized by the NORIA. 'Ihe characteristic curves and the thermal con-Newton-Raphson technique, ductivity, as a function of saturation and temperature, are supplied to PETROS through user-written function sub-The linearized equations are solved by the liarw ell matrix programs. Other parameters such as diffusion coeffi-solver, that stores only the nonvero elements of a mainx, cients, water viscosity, saturation vapor pressure of water, thus reducing core storage requirements for the code. and default values of the characteristic curyn r d there mal conductivity are supplied internalb i-tw Kle as ',scosity, C.2.2 NORIA function subprograms. Constants sue: specific heats, and water density can be.

ter at de-NORI A (Bixler,1985) is designed to simulate liquid, va-fault values or supplied by the user. The er can also

. por, air, and energy transport in partially saturated and choose between equihbrium and nonequilibrium vapor-fully saturated porous media,1he following mechanisms pressure models. are included in NORIA: (1) transport of water, vapor, and air due to pressure gradients: (2) transport of water, The aforementioned equations are solved numerically by vapor, and air due to density gradients; (3) binary diffu-a finite-difference method.The equations are differenti-sion of vapcir and air; (4) Knudsen diffusion of vapor and nted in both space and time. Differentiating in time airt (5) thermo-diffusion of vapor and air; (6) conduction results in fully implicit equations. The saturation and of sensible heat: (7) convection of sensible heat; temperature equations are solved with a trediagonal al-(8) evaporation and condensation; (9) a nonequilibrium porithm. llecause the vapor and air-pressure equations and equilibrium vapor pressure model: and (10) capillary are strongly coupled, they are solved with a block tri-pressure. Nearly all the thermodynamic and constitutive diagonal algorithm. properties in the code can be defined nonlinearly in terms of the remaining dependent or independent variables by C.3 Geochemical ProEram Summaries - the user. C.3.1 PilREEQE NORIA solves four non-linear partial differential equa. tions governing the flow of water vapor, air, and energy. PHREEQE (Parkhurst,1980) was developed to model These equaGons consist of a water-pressure equation, a geochemical reactions between water and rock material, 4 vapor partial-pressure equation, an air partial-pressure Based on an ion-pairing aqueous model, the program equation, and a heat equation. The equations are solved calculates pil, redox potential, and mass transfer as a by the Galerkin finite-element method. Time-stepping is function of reaction progress. The program performs a accomplished by a two-step time mtegrator wah auto-mass balance of elements in terms ofIheir concentrations - matic time step selection. The non-linear difference in the aqueous phase and uses electrical neutrality and equations formed by application of the fmite-element electron balance relations to complete the set of equa-method are solved simultaneously by Newton Raphson tions needed to solve a given problem. iteration. Normally, a one-step iteration is used; however, a multistep iteration is used if the correction on the first Tbc program solves a set of non-linear algebraic equa. iteration is larger than a specified amount. tions, using a combination of a continued-fraction ap-proach for mass balance and a Newton-Raphson iteration C.23 PETROS technique. PETROS (Hadley,1985) is designed to simulate pmb-C3.2 EQ3/6 lems simd, ar to those handled by NORIA. PETROS solves the same numberand types of non-linear equations EQ3/6 (Wolery,1979) was developed to compute equilib-and handles the same physical processes as NORI A. but rium models of aqueous geochemical systems. EQ3 per. in a slightly different manner, The main difference be. forms distnbution-of-species calculations for natural tween the two codes is that PETROS solves only one-water compositions. EQ6 uses the results of EQ3 to NUREG-1327 C-2 ~,

.~ ~ Appendix C . predict the consequences of heating and cooling aqueous port pioperties typical of fractures. He transport model solutions ar.d of irreversible reactions in rock-water sys-takes radioactive decay and a linear sorption (Kd) into tems. Reaction path modeling is usefulin analysing com-acmunt. It allews different retardation factors for daugh- ' plex systems where analytical data do not permit the ters and parents. ' definition of reactions by mass balarce alone. The SPARTAN code was used for some very preliminary The_ program uses a Newton-Raphscm method to solve a sessments of a proposed tcpository at Yucca Mountain. the algebraic governing equations of chemical equilib-In the test cases that the authors demonstrated, there were either two or three pathways for radionuclide trw-rium.. port, which was supposed to represent the different lengths from the repository to the water tablec For an . C.3.3 WATEQF infiltration rate of 0.5 mm/yr, there were two pathways for WATEQF (Plummer,1976) simulates the thermody-matrix flow. In this case, only I-129, C-14, and Tc-99 namic speciation of inorganic ions and complex species in reached the accessible environment within 100,000 years. solution for a given water analysis. The program provides For a rate of 5.0 mm/yr,it was assumed that the water in - a general capability to calculate chemical equilibria in excess of what the matrix could carry would travel through natural waters at low ter peratures. a third pathway as fracture flow. Many more of the radi-onuclides were released to the accessible enviror nent for ' WATEQF uses a successive approximation method to this case, solve the mass action and mass balance equations C.4.2 TOSPAC C.3.4 CIIEMTRN TOSPAC (Dudley,1987), the Total System Performance CHEMTRN (Miller,1983) was developed to simulate Assessment Code, is a computer progra n designed to one. dimensional transport of chemical species in ground. simulate water flow and transport of soluble waste in vater. Equilibrium is assumed in all chemical reactions, fractured, porous unsaturated rock. The groundwater flow module solves either the transient or steady-state and thermodynamic aethities of all reacting species are related by mass-action expressions. The program includes partial-differential equations for an equivalent porous, fractured medium, in which the properties of the matrix the effects of dispersion and diffusion. advection, sorp. and fractwres are combined into one constitutive relation-tion via ion exchange or surface complexation, aqueous complexation, precipitation and dissolution of solids, and ship for saturation versus hydraulic conductivity (or ma- . the dissociation of water. trix potential versus hydraulic conductivity). The site is represented as a series of one-dimensional flow tubes The governing equations are approximated using a finite. with no lateral interchange. Within any single flow tube, difference approach. A Newton.Raphson iteration tech. either the steady 4 tate or transient flow equation for the nique is used to to selve the system of equations. equivalent matrix-fracture relationship is solved. For the ~ steady-state situation, the solution is iterative, to allow for the self-adjustment of the hydraulic ccnductivity and CA Transport Program Summaries saturation values to correspond to the constitutive rela-tionships for each layer. Once the solution reaches C.4.1 SPARTAN steady-state, the hydraulic conducthity is known, and con-sequently so is the net downward flux and groundwater

The SPARTAN code (Lin,1985)is a simple performance veh, city that can then be used in the transport calcula-assessment t ode developed at Sandia National Laborato-tions. The transient solution solves for pressure head, ries. The model employs a simplistic hydraulic model Sr with a numerical solution of Richard's equation using flow of water infiltratmg the surface. and reaching the Pickard iteration.

water tabte.This model has little in the way of a mechanis-tic explanation for the way water would flow at Yucca he module for radionuclide transport uses the vehicities Mountain he rate of infiltration in the matrix is as-calculated from the flow module. First, the code esti-sumed to follow Darcy's law, with a graoient of unity, a mates the fraction of flow in the matrix and fracture flow fixed permeability and fixed effective porosity. For infil-paths Concentrationsof each radionuclideare calculated tration rates less than 1.0 mm/ year, the speed of ground-for the matrix and fracture compartments, with a dynamic water movement is proportional strictly to the infiltration coupling between them. rate and does not take into account the change of hydrau-lic conductivity with moisture content. For infiltration C.4.3 NEFTRAN rates greater than 1.0 mm/yr, the model assumes that a fraction of the water infiltrating will move through the NEFFRAN (Inngsine, at al.,1937)is a network flow and fracture zone faster than through the matrix, with trans-transport code developed by Sandia National C-3 NUREG-1327

l Appendrx C 12boratories primarily for modeling the repository per. should be overcorne. Sandia is deseloping a multi-formance at saturated sites. The flow rnodel in dimensional, fmiteshfference model to calculate steady-NE171RAN consists of an arbitrary network of one-state and transient, unsaturated flow in porous media. dimensional pipes, connected at nodes. Boundary condi-Output from this moJet will be led ducelly into a tions of pressure are set on some of these nodes, and the NEITIRAN modified to accept flux boundaries and tran-hydraulic properties of transmissivity and porosity are set sient flow conditions, llowever, the soarce term will con-within the pipes.The network model then solves for the tinue to be represented by a singk leg. and L.crefore will steady-state velocity anu flux within the network. Radi-not be able to simulate a highly-distributed source. onuclides are transported in the network by the calcu-lated ilux.The model uses retardation factors to express C.4.4 University of California, Berkeley the speed at which a pa ticular species is transported. It also allows for transport between the actively flowing legs (UCB) Programs and immabile water adjacent to the leg, in order to simu- 'lhere are a large numbc r of analy ncal ma s (i.e., closed. late matrix diffusion. 'Ihe program r1so can simulate the form solutions) available that could serve to calculate ~ transport cf radionuclide chains. There are two models flow and transport particularly for one-dimensional, for cham transportJ11e first assumes equal retardation steady state flow,in v hich there are really few considera. coefficients and allows chains of up to three members. tions'as to whether the flow is saturated or unsaturated. The other model allows arbut ry retardation coefficients 'the UCB codes combme simple source-term models with for chains of up to six daughters. Calculations using t' e analytical solutions for one. dimensional, stead -state 3 former model are much less time-consuming. N EFIR AN flow, and radionuclide transport. These codes have been simulates dispersion along the legs using the Distributed used in a mimber of imptant U.S. studies (e.g., WISP Velocity Method (DVM), which assuraes that dispersion report and AREST code deselopment). The programs is caused by the distribution of velocities in the flow field. are unique analytic solut ons, because they provide ex-E In its Present form, NEl' IRAN ts not ideall suited for Y coefficients for each daueh ter radionuchde. How ever, the performance assessment of a repository located m. un-solution technique raay'not perm" the meorporation of saturated, fractured tuff for the followmg reasons; more than one hydrologic layer: this would rnake dif ficult 1. The model is set up for boundary conditions that are e appMon hemh & Yuaa Wuntane, tW am mal hnu laym we hen! me w appropriate for a saturated site, terial properties. 2. The flow model is for steady-state conditions.Tran-sient recharge may be an important consideration in C.4.5 Laplace Transform Solutions unsaturated, fractured tuff. Another class of analytical codes is laplace Transform 3. 'Ihe model assumes that the source term is concen-domain solutions (Robinson and Hodgkinson,1986). The trated in one leg only, and cannot represent source source term, transport model, and even stochastic setu-terms highly distributed in time and space. This limi-tion c:m be set up using this method, solving the linear-tat:an may not be as important for saturated sites differential equations in the I aplace domain and obtain-where flow is more horizontal than vertical, but it ing the time-domain solution by numerical evaluation of could be a limitation in situations whee multiple the contour integral in the complex niane. This solution travel paths are needed to adequately account for technique should be relatively easy to apply to the prob-system performance. lem of transport through multiple layers.The recent de. velopmer.t and progress of this solution technique in the Presently, NE171RAN is being modified specifical!j for United Kingdom needs to be followed for later use in the Yucca Mountain case, and some of these limitations performance assessment. NURiiG-1327 C-4

APPENDIX D-GASEOUS RELEASES OF CARBON-14 D.1 IntrOduClion urements for boiling-water reactor (BWR) fuel are not available (van Konynenberg,1987). It is expected that Carbon-14 (C-14)is produced in nuclear reactors by the measurements for llWR fuel may be different because activation of nitrogen impurities in the fuel cladding, and the oxidation potential is different in these reactors. The by the activation of oxygen-17 (0-17), particriarly in the remainder of the C-14 is dispersed in the fuel, cladding uranium dioxide fuel and in the circulating water of light-gap, and the intergrain boundaries. There is little or no water reactors. The release of C-14 from the waste pack-information on C-14 inventories for other non-fuel parts ages may be of concern because there is at-least the of the reactor, possibility of a fast gas pathway to the accessible environ-ment through the unsaturated fractured rock, excava. The two mechanisms for producing C-14 in the reactor tions, and tunnels. Although in this Phase 1 demonstra. are important to understanding its availability. Presum-tion, the release of C-14 was treated m the liquid pathway ably, C-14 created by activation of nitrogen would be analysis by inclusion with the total release into the liquid dispersed in the cladding, because the nitrogen may also phase, this treatment would not be conservative from the be dispersed. Much of the C-14 appears to be from the standpoint of the gaseous release pathway. This appendix oxygen-activation mechanism, and is adsor a onto the presents models for the gaseous release of C-14 from the cladding, fairly close to rhe surface. Sus fact may be waste package ar d its transport to the accessible en iron, important because it allows the C--14 to be more readily

ment, accessible to the environment than if it were uniformly dispersed in the cladding (SCP, Sections 8.3.5.9,8.3.5.10)

D.2 Source Term D.2.1 Possible Release Modes from Spent C-14 is found in quantities ar) order of magnitude greater Fuel than would be allowed under 40 CFR Part 191, if all were released. For the 70,000 metric tons heavy metal as-Upon failure of the canister, gaseous C-14 could be re-sumed, the initial inventory of C--14 for tnis study was leased to the geologic setting. Most of the C-14 in the 98,000 curies (Doctor, et al.,1992), whereas the allowed canister is apparently in the form of elemental carbon, release under 40 CFR Part 191 is only 7,000 curies. C-14 metal carbides, or oxycarbides (van Konynenberg,1987). has a half life of 5720 years. The majority of erwiron-In inert nitrogen and helium atmospheres, spent fuel mental C-14 comes from interaction of cosmic ray neu-does not readily release its C-14. Upon exposure to air, trons and nitrogen, although it is also created by activa-however, some of the C-14 oxidizes and is usually re-tion of the rare 0-17 isotope in the atmosphere (van leased to theimmediate atmosphere as " cop. About I to Konynenberg,1987). It is produced in great quantities in 2 percent of the available C-14 inventory, mainly that atmospheric nuclear explosions through neutron activa-portion deposited on the surface of the canisters as crud tion. Once in the atmosphere, C-14 is removed from the or collected on the intergrain boundaries of the fuel, environment mainly byabsorption in the bicartmnate ions could be released quickly by this mechanism (van in seawater with an spparent relaxation time (i.e., time for Konyn enberg,1987). For elemental carbon, release could half to disappear from the biosphere)of approximately 9 depend on oxidation to carbon dioxide and carbon monox-to 15 years (Killough and Till,1978). A portion of the - ide, the rate of which ts extremely slow at low tempera-C-14 recycles through the food chain and is very biologi-tt'res. Elemental carbon is known to be extremely stable j' cally active. The combination of biological actnity and under normal conditions, as is evidenced by the presence long half-life leads to relatively large population doses of graphite in schists exposed for thousands of years at the per curic released. earth's surface. Here is some experimental evidence to suggest however, that carbon will oxidize to carbon diox-In reactor fuel, C-14 is produced by the same mechanisms ide at a temperature of 275'C within a radiation field of as in the atmosphere.The main routes to production are: 10,000 rLis per hour (van Konynenberg,1987). Tempera-(1) activation of nitrogen impurities in the metallic strue-tures of the fuel may be in this range for the first few ture of the reactor and the fuel cladding, and (2) activa-decades after storage, but are likely to be considerably tion of O-17 in the uranium dioxide fuel and in the circu-cooler near the time required for minimum canister life. lating water of the reactor, with subsequent deposition Radiation levels of 10,000 rad /hr are likely to be present onto the cladding and other structural material. for up to about 100 years. Although the radiation field diminishes with time, no experimental evidence is avail-Measurements indicate that about one-third of the total able to indicate that there is a threshold below which no C-14 inientory resides in or on the cladding of oxidation would occur. For the sake of conservatism in pressurized-water reactor (PWR) fuel, but similar meas-this analysis, it was assumed that a mechanism is available D-1 NURl!G-1327

Appendix D - l l I to oxidize available carbon to gaseous carbon dioxide for the UO will spall, becoming more porous and less dense. 2 the lifetime of the repository. The increase in volume could promote continued crack-ing of the cladding, allowing more pellets to be exposed. The more likely C-14 release mechanisms from spent fuel are: Spallation is an indication that significant oxidation has occurred and may also provide for increased exposure of o Dissuution of the cladding and oxidation of the the C-14 to oxidants. Iloth UO and C will be competing 2 C-14 attached to the :ladding, e.g., crud. for the oxid;mts. From thermodynamic considerations alone, cartxm would be oxidized first at low oxygen activi-o Quick release of a small percentage of carbon diox-ties, followed by oxidation of UO at higher oxygen activi-2 ide gas from the cladding-pellet gap, upon failure of ties. However, the relative rates of the competing reac-the cladding. tions probably govern how the components of the spent fuel will be oxidized. Einsiger and Woodley (1985a) state o Diffusion of oygen im ihe waste form, particdarly that for irradiated fuel, the uranium dioxide crystalline the matrix, reaction of the carbon with the oxygen, structure is damaged and pellets are fragmented, thereby and the subsequent diffusion of carbon dioxide out opening more surface area to oxidation. In addition, gas of the matrix-bubbles and fusim rnducts may migrate to the grain boundarice d ; it mterconnected paths can form, Other possible mechanisms might also release C-14 (al-making grain boundanes preferential sites for oxidation. though little or no direct evidence is available to show that The radiation field can ioniec or excite atmospheric oxy-they apply): gen or water, possibly enhancmg the oxidation rate. o. Galvanic reaction between elemental carbon in the Einsiger and Strain (1984) present two curves bounding cladding or metal carbides and the surrounding the time to spallation, ts, as a function of temperature T: metalin the waste form. o Reaction of metal carbides on the zirconium or ura-nium with water to form acetylene gas (Katz and log ts - (1.03 x 10NT)- 15.9 (D.2) Rabinowitz,1951). w here ts is in given in hours, T in degrees Kelvin, and log o ,licrobial action on carbon or carbon compounds in denotes the base 10 logarithm. liquations D.1 and D.2 are the waste. not directly useable to determine the rate of release of C-14, because they are formulated with steady tempera-0 Release of methyl iodide created from the reaction tures in mind. Since the fuel temperature changes with of carbon and iodine present in the fuel.This could time, it is more convenient to convert spallation time to be a potential release mechanism for I-129. Methyl an oxidation rate. If it is assumed that the rate of oxida-tion,1,is the reciprocal of the spallation time,ts, then a iodide would be volatile around 200*C, tempera. 3 tures expected in the repository during the first few time-dependent rate of C-14 release can be defined: decades after site closure. (D.3) A = 1/t3 3 Little direct evidente is available to support a particular model for C-14 gaseous release, but a reasonable set of This model may be conservative from the standpoint that release mechanisms was chosen, based on the limited the time for the onset of spallation does not signal the information, and applied in a release model for this auxil-total oxidation of the fuel pellets. On the other hand, the j iary analysis. These mechanisms are discussed next in rate of relcase of C-14 may not be as low as indicated for theit order of expected importance. In addition, conserva-long spallation times that occur at lower temperatures. tive ranges of parameter values were chosen for use in the models for these release mechanisms. However, it cannot 1).2.1.2 Oxidation of zirconium be stated that the overall model is conservative. A large fraction of the C-14 inventory may be in or on the !M.l.1 Releases due to oxidation of uranium dioxide claddmg. caused by neutron activation of O-17 picked up from the circulating water, particularly in ItWRs, or from For this study, the C-14 trapped within the uranium diox. nitrogen impurities in the metal itself. Corrosion of the ide fuel was released at a rate coupled to the rate of zirconium may be the first step in releasmg the C-14 to oxidation of the fuel. Uranium dioxide is unstathe in an the atmosphere, although it is possible that this corrosion oxidizing environment, and oxidizes to other forms, with may not be necessary to initiate release. In addition. clad. corresponding increases in volume, porosity, and fractur-ding corrosion leading to perforation could expose the ing,in many cases. If the reaction proceeds fast enough, UO to oxidation. 2 NUREG-1327 D-2

l Appendix D Oxidation of the cladding has been studied for the case of an existing oxide coating. since it might have been picked dry-cask storage of spent fuel. liinziger and Kohli(1984) up externally from the circulating water. The fact that present a rate equation for zirconium cladding in terms of little if any oxidation of the zirconium alloy occurs at temperature: temperatures lower than about 230*C leads to a tentative conclusion, for the purposes of this study, that there will L - 3.6SE8 x t x exp (-15810/F) (D.4) be little additional zirconium oxidation after canister fail-o ure. it was assumed, therefore, that only the readi!y avail-where L is the oxide thickness in millimeters, t is the time able portion of the C-14, about one percent, will be in years, and T is the absolute temperature, degrees Kel-driven off duting the pre-cimister-failure period, and that vin. To find the growth of the zirconium oxide layer with there will be no additional releases from 'ne rirconium time, we first convert Equation D.4 to a rate, and inte-compartment. Corrosion of the cladding might be rela-grate from the time of faile 2 tg, using the expected tively more important if it causes perforation, allowing temperature of the waste: oxygen to reach and oxidi7e the spent fuel matrix. 5 D,2.2 Proposed Source Term Model L= 3.68E8 exp(-15810/T(t)) dt (D.5) I ae s y, the Mwing mW was cMsen W f the release of gaseous carbon from the waste package. The calculated oxide thicknesses are presented in Table incorporating the mechanisms discussed previously: D.1 for a typical fuel temperature ranging from a high of 320*C to a low of Il0*C over 10.000 years, and an as-Canisters failed at a rate predicted by a normal prob-sumed failure time of tg - 0. ability distribution.Two different distrDutions were chosen to demonstrate the sensitivity of the C-14 re-Table D 1 Calculated Zirconium Oxide Thickness lease to the waste-package lifetime. At the time of canister failure, oxygen cntered the Temperature Time Thickness ('C) (years) (mm) canister and became available immediately to react with the uranium dioxide in the fuel rods.He model 320 5 4.90E-3 assumed that release rate was tied closely to the 300 25 1.25E-2 spall tion rate of the fuel, and that there was suffi-275 50 152E, cient oxygen available upon canister failure for the '50 75 l$59E ; fuel xidation to proceed. Although most fuel rods b30 100 1.61E-2 will have additional protection from oxidation based 200-110 10000 1.62E-2 on resistance to corrosion of the zirconium alloy cladding, it was assumed for the purpose of conser-vatism, that all fuel rods are available for release of their C-14 inventories. A typical cladding thickness is 0.61 mm. so the niaximum oxide growth is only about 3 percent of the total thickness. On failure of the canisters, a small fraction of the Most of the oxidation takes place when temperatures are C-14 inventory was released rapidly. This fraction highest, with virtually none after about 100 years. represented the C-14 inventory of the cladding-pellet gap and the C-14 close to the outside surface Einsiger and Woodicy (1985a) also describe a possible of the cladding or crud that would be readily oxi-condition that might affect the rate of cladding failure in dized. the absence of oxygen. Canisters might contain a few tens of milliliters of water from rods stored in cooling pools. The average fractional release rate of C-14, f(t), was The radiolysis of the water could provide oxidants that calculated based on the random failures and exidations of could oxidize the cladding (Reed,1987). a large number of canisters, to which was added the frac-tional prompt release, f, from the etmisters at the time of p The ramifications of zirronium oxidation are not entirely failure, tg: clear. It appears that there would be relatively littie oxida-tion of the zirconium in the repository. lf the fuel is kept N cool; e.g., in wct storage casks before being placed in the f(t) = 1 I [ H((t - tri) f + As (t) dtl (D.6) camsters, the reaction would not proceed very far. It N p

=g would be more oxidized in dry storage, bu t might be inhib-n i

ited by the presence of inert atmosphere in the canisters. Although the percent oxidation may be small, most of the where N is the number of canisters, and il(t -tg;) is the C-14 might be close to the surface as crud, or attached to Eleaviside mit step function at time t = tri. D-3 NLiREG-1327

Appadix D D.2.3 Results of Source Term l'redictions and bicarbonate ions. Part of the dead carbon will be available to exchange with the C-14 along the transport Figure D.1 illustrates the fraction of the total C-14 inven-pathwapihe effect of this exchange may be to retard the tory teleased up to l0,000 years for two different assumed speed at which the C-14 could be transported to the canister failure models.The higher release curve (sohd) accessible environment. A potentially important reaction corresponds to canister failure with a naan failure time of is the preapitation of calcite (calcium carbonate) by the 550 years, a standard deviation of 150 years, and an upper reaction of c deium ions and carbon dioxide to form a low and lower limit of 100 and 1000 ) cars, rcspectively. The solubihty precipitate (Ross,14SS). The significance of lower (broken) curve corresponds to a mean failure time retardation of C-14, or its removal by precipitation, will of 1000 years, with a standard deviation of 300 ) cars, and depend on the relative rates of exchange betwcen the an upper and lower limit of 200 and 1800 years, respec-CO pas, the bicarixmate, and calcite, and the schicity of 2 tively.The maximum cumulative releases were approxi-air flow through the rock, mately 13.2 percent and 2.5 percent, respectively, thus illustrating the strong dependence of C-14 release on Several reports propose C-14 transport models. Knapp waste package lifetime. (1987) describes a one-dimensional model for C-14 t rans. port by advection, with exchange between the gas phase D.2.4 Limitations of the C-14 Source Term and the bicarbonate m ttje groundwater. The results of 'qg this study show that for T ucca Mountain C-14 released as a pulse from the repository horvon at 2000 years after The C-14 release model was based on the following hmit-repository closure would reach the surface within 6000 ing assumptions: years. A non-mechanistic failure of all canisters in a time Amter, et al.. (1988) expand on the concept of a C-14 short relative to the half-hfe of C-14 and the 10,000 transport model with more computational detail. Their year period of interest, model accounts for two dimensional pas advection, with diffusion and exchange between liquid and gas compart-An influx, upon canister failure, of sufficient oxygen ments.They assume that pas and water are in equilibrium to cause unimpeded fuel oxidation. Oxygen was not for carbonate species because of the rapid diffusion of cathon dioxide. Dissolved bicarbonate ions in the rock are consumed by other reducing agents, such as the can. ister walls and metal components of the fuel assem. considered to be essentially immobile because of the rela-

blics, th ely high vehicity of gas flow, as compared to liquid flow.

Liquid-phase diffusion is also ignored because hquid-The highly corrosion-resistant cladding on the fuel phase diffusion constants are much smaller than gas-offered no protection from oxidation. phase diffusion constants.The presence of the C-14 com-ponents m the hquid will have the effect of reducing the A prompt release from the claddmg and pellet-gap speed of transport by a retardation factor, o inventories for 100 percent of the fuel rods. (In actu. ality, the prompt release might occur only fron D.3.1 Chemical Modeling failed fuel rods.) The chemical retardation of C-14 depends on equinh-rium between carbon dioxide, bicarbonate ion, and solid The rate of oxidation was equal to the recip:ocid of the spallation time. In fact, spallation time may be carbonate. The equilibnum between C-14 as CO and 2 morc representative of the oxidation of onh a irac-bicadxmate is tied to several possible chemic d factors, ~ tion of the fuel. Based on this assumption,' the re-including the presence of calcite, CO partial pressure, 2 lease rate of C-14 might be conservative at high and pH. Ross(1%8) assumes that CO d:ssolved in water 2 temperatures, and may not be conservative at low a tmmobile, a conservative asmmption for atmospheric temperatures. releases, since there is likely to be a net movement of groundwater to the water table from the ground surface. CO is produced naturally in plant roots. The decline in 2 D.3 Gaseous Transport Model the concentration or CO with depth seems to indicate 2 that it is being removed by some mechanism, possibly Once released from the fuel, the C-14 would probably be calcite precipitation. Ross speculates that this would carbon dioxide or another gas, such as methane or acety-require a source of calcium ions infiltrating the site. lene. Van Konynenberg (1987) estimates that there Although the groundwater is not saturated with calcium would be no more than 22 kilograms of C-14 in the re-carbonate, and it does not appear that the cMette n pository. as contsasted to greater than 300.000 kilograms precipitating naturally, an increase in repository tempera. of dead carbon in the immediate vicinity of the repository, ture inight cause precipitation both by evaporation and in the form of carbon dioxide and even more as carbonate the decrease in calcite solubihty Calate solubility is NURIEl327 44

m a o i 10 ~ r;:: x 5bh W D c? _5 ggs =a =0 L / o g s. /. / $gF 10-2__ f o 36 / "EE o / 5ao e / d*a / c~ e c m? H f $0h b { ~3X e / Demonstration Plot &. B =2 v- / see caption c g g ~o g

• 8,*

10 - I e m o I E@2 t 8 c 9 KE o i ?B* I u 5' o R o I o E. = n w c m / N= w u E., ", / W.P. Lifetime - mean 550 150 clipped 100,1000 I iti $ 10 -

---

W.P. Lifetime - mean 1000 i 300 clipped 200,1800 I

a 4 w=. I o= - l a. 9 = ?. ~ ,i nb,' ~ e O6 = 0O g i a a i a i i g l gg iiy i i i i, @5 103 1 08 I =a @ E-Time - Years y zc

== a y co g M ?" y CL C n' C O a

Appendix D retrograde, decreasing with increasing temperature. This tory, and therefore the data used in the analyses are trend leads to some interesting possibilities of how the somewhat subjective. bicarbonate ions in the heated area would react and whether there would be an irreversible deposition of The results of the PHRI!IiQli calculations were func-C-14 in precipitating calcite. tions expressing the dependence of retardation on tem-perature for the hydrogeologic units The expected retar-Differences in the atomic weights of C-14 and C-12 may dation coefficients ranged from about 20 to 90 over the lead to fractionation because of the slightly different rates expected temperatures and concentrations of carbonate in the rock. of reaction, evaporation, condensation, etystallization, adsorption, and diffusion. Fractionation was estimated by comparison to enrichment factors for stable C-13, and D.3.2 Gas Phase Transport Modeling found to be negligible (Amter, et al.,1988). Several mechanisms potentially dnve the gas flow, but Amter, et al. (1988) present the results of geochemical Amter, et al. (1988) consider two mechanisms to be domi-modeling to determine the comp;icated equilibrium nant: among the C-14 gas, liquid, and solid phases. 'Ihc concep-Temperature-driven circulation caused by reposi-tual model of the geochemical system had three principal e assumptions: tory heat and the geothermal gradient: 'the dif ference in density between the moist, warm "1. Sufficient calcium carbonate is present in airin the rock and the cool dry air in the atmosphere. the unsaturated zone to dominate the aque. ous chemistry and buffer the pH of the 'the authors considered and climmated the following po-water. tential mechanisms for transport of C-14: 2. A relatively minor amount of calcium is de-1.iquid phase advection--The downward flux of liq-e rived from silicate weathering reactions. As uid water is likely to be about one tenth the gas flux a first approximation,it can be assumed that during the period of repository heating that is most calcium concentrations are the result of important to ill.W performance assessment, equilibration with calcium cartxmate. lhffusion-Using a travel distance of 350 meters e 3. Fractionation plays a negligible role in re-and a retardation factor for C-14 of 70 gives a travel moving carbon-14 from the gas phase, and time for diffusion of 43,000 yrs, w hich is much larger concentrations of carbon-14 are propor. than either the ambient of heat-driven travel times tional to those of carix n-12.. for the repository. The effect 4 isotopic equilibrium between phases is to B nary diffusion-A mass flow of air from higher to e reduce the speed of transport by a factor B, defmed: lower temperatures in the rock will be driven by dtf-fusion, but this flow was shown to be much smaller o CT than the temperature-driven flow. B=1+ (D.7) 60 CT Mixing by seasonally alternating flow-Under ambi-e ent conditions, gas within Yucca Mountain will move where OT= total porosity upward in winter and to a lesser extent downward in summer, but would move C-14 molecules only a few dra.med porosity centimeters per season. much smaller than even the OD = Cr* = concentration of carbon ion in the liquid phase at equilibrium The authors' C-14 transport model relied on a tempera-ture field developed by Tstmg and Preuss (1987) that concentration of carbon ion in the gas showed a gas phase velocity of meters to thousands of CT = phase at equilibnum meters per year resulting from the repository heat, as shown in Figure D.2. The model of Amter, et al. (1988) Ampter, et al. (1988) determine the equihbrium concen-predicted travel times for C-14 of several hundred years trations needed for liquation D.7 using the PilRiiliQi! to several tens of thousands of years, depending on the reaction path model.There are few data available on the hication in the reposnory and the depth of the overhur-chemistry of water in the unsaturated rocks of the reposi-den, as shown in Ihure D.3. NUR11G-1327 lb6

Appendix D I i l l 1 -100 - 1, I 1 1 I I -200 -

g i

3 I g ie O W l 5 -300 - ig O "I O s ie O a i to O s Q i

1. w O

9 O y 0 O .-400 - e,O O e i I l -Legend O e -1 ~500 - I e Without Diffusion i O wit'i Ditfusion i I I Estimate i I I -600 "i i "i " i O 10 100 100n 50,000 Convection Gas Velocity (treters/ year) Figure D-2 Gas convcction veke along the repository center IL 4 at 100 years after waste emplacement. Plus t nd Minus signs refer to upward anu downward fluxes at the repository depth for the case with binary diffusion (modified from Tsang and Preuss, 1987). D-7 N Ult!!G-13'.7

Appendix D 1 11 a 10 - de s. s as488 p, o-e 8-Y 7_

• E>-

^* j% 6-n. -. }. ^* { 5-3

1. 8 ao 2

4~ he 45 + 5 64 3-A#a g +++++++++, ++++++++ 2 agg+ +++++++++.e.+++++++++++ ERNNNggy,uu 3 + I O EENNNNNm gg kNNagg 8ENNumannummE MA 0 8 i 0 200 400 600 800 1000 Distanco From Repository West End (mcters) A Pre emplacement; Ambient + 10,000 Yrs T=314 M 2,000 Yrs Ta330 9 50,000 Yrs T=303 Figure D-3 Carbon-14 travel time from the repository to the surface for ambient conditions, 2,0(X1,10,000, and $0,000 years (modified from Ampter, et al.,1988). NURIIG-1327 1)- 8

Appendix D i D33 C-H Transport Mo&l Table 1).2 nelease reaulon as fundion of Hilcase Time ir. the Phase I demonstration, the staff used the cral-Time of sticase frattion Reaching mated travel times calculated by Amter, et al. (1988), as pars Sur face shown in l'igure DJ, to develop a scoping model that accounts fr t ranstmrt of carbon dioxide from the relvsi-500 0.86 3 tory to the surface of the earth. 'the rnodel considered 1500 0.91 radioactive decay using the average travel tirne for C-14 2500 0.66 from the repository to the surface. Arnter, et al. (1968) 3500 0.82 calculated the trasci times along a transect of the reposi-4500 0.7 ' tory at zero,2000 years,10,000 years and 50,000 years. 5500 0.7 'lhe fractional release, f. at the carth's surface for a 6500 0.A g parcel of C-14 released at time tq,ic repository to the was octermined by beyond 6500 none 2.10 000 yrs integratm, g along the path from t surface, assuming that the vehicity of the parcct would be cverywlycre equal, but varying with time according to a DJJJ IJmNima of C. M gas transpmt model h.near mterpolation of the four times calculated by Ampter. Some of the lirnitations of the transport model are: 'lhete is the posi,ibility that gaseous releases from 'lhis is not necessarily a good assumption, because the the repository level rnay follow the shortest path, velocity is known to v'uy in space within the complicated and that there.nay be ample ground transport be-convection currents predicted by 'lkmg and Preuss, tween one part of the relmsitory and another be, 1987): cause of the network of drifts, shafts, and fractures. The effective travel Cme for C-14 released any-where in the repository may therefore be more char. acteristic of the shortest travel time adculated. L(t) = v(t) dt GAS) 'lhete is evidence that in natural waters, CO is not f M in equihbrium with the atmosphere, partially be-cmse of unfavorable mixing conditions and the slow-where 1.(t)is the normaliicd distance that the parcel has ness of the pas transfer reaction (Stumm and Mor-traveled relative to the distance to the surface, v(t)is the gan,1970). The chemical model for C-14 behavior normatiecd velocity, defined here as the reciprocal of the was based on the assumption of equilibrium. Failure travel time at time * "nd to is the time of release. The to attain equilibriurn would have the effect of reduc-integral was evaluateu graphically to find the time t when ing the retardation of C-14. 14t) = 1. The object of the integration was to find the-In their transport and chemical models, Ampter, et trave! time of the parcel and to determine whether it could reach the surface within the stated time limit,i.e., al. (1989) assume intimate contact between the gas l 10,000 years. Once the travel time, t, was determined and v.ater phases, Such contact is unlikely at Yucca Wuntain, because, tmder unsaturated conditions, for each parcel with release time toi, the fmetim f. g released at the tarth's surface was deterrnined by radiom:. water would be present primarily in the smallest live decay: rock pores, and the flow of air would be most preva-lent in the largest rock pores and fractures. 'lhere-fore, the potential for close air water contact would f = exp (- A t ) (D.9) be diminished, having the effett of reducing the re-g g tardation of C-14. where A = In 2/tm.The results of tbese calculations are D.4 ConcitlSionS nnil summarized in Table D.2 for releases at $00 to 6500 years. Ilecolunlentifitions The fractional release ranges from a manmum of 0.91 to a minimum of 0.65. It is interesting to note that because the The resuhs of Amter, et al. (1988) and Knapp (1987) for upward velocity approachu a maximum value, the earli-transport of C-14 from the Yucca Mountain repository to est release shown in Table D.2 does not result m the the surface of the carth predict travel times ranging from highest fraction escaping. Helcases afterabout 6500 years a few hundred to a few thouvmd years, and are shortest do not arrive at the surface of the carth before 10.000 during the period where there is significtml heating from

years, the radioactive decay. 'this period of short travel times D-4 N URiiG-1327

Appendix D I coincides roughly with the period when the demonstra-of site characterization. Ihcre n wnsidcrable seat-tion model predicted most of the C-14 releases to occur; ter m the data on spallation of the uramum dioxide however, any release depends on the failure of the waste fuel, and this could be a potential source of uncer-canisters. 'lhe release of C-14 is very sensitive to the tainty. Direct measurements of C-14 relcases from waste-packare lifetime in the modeling approach chosen the variouscompartments of the fuel would be more for this study, particularly because early fadute times lead reliable than models bascJ indacelly on elIcets such to faster and more complete oxidation of the uranium as fuel spallation. dioxide. Considering the $720 year half hfe of C-14, Investigate the peuthermstly of calate precipitation there would be relatively little attenuation of the cumula-tive release of C-14 at the carth's surface because of at the Yucca Mountain site, undcr repositoryeondo holdup in the geologic battier. tions, to determine w hether the released C.14 is re-moved cf fectively before reaching the acc essilyle en- 'the telease and transport models for this auxiliary analy-vironment. 'lhere are several counteracting factors sis were formulated using assumptions believed to be involved in the ef fectneness of this mechannm for conservative, although there is little direct evidence to removing C-14. K< app (1987) states that " Water-support these assumptions. '!he following areas were rock interaction is probably insigmficant due to the identified as the most fruitful for the collection of addi-low abundance of calcite at the Nevada site anJ due tional data: to the prediction that calcite will not precipitate? Iloweser calcite solubihty dimmishes with increas-e Investigate the mechanisms 'or C-14 release, in-ing temperature, leading to the possibiht) that cluding the available information on dry-cask stor-repository induced heatmg would cause cakite pic-are, and the investigations to be performed as part cipitatian. NUlti;O-1327 D - 10

APPENDIX E-TESTING STATISTICAL CONVERGENCE E.1 Discussloli random seed for the IJIS sampling. 'the results are shown in Figure !!.l. Only one of the five CCDF curves 'Ihe latin Ilypercube Sampling (Ills) method is an effi-generated with the 100-point samples was close to the cient method for performing Monte Carlo analyses $00-point CCDF eurve, Convergence was best in the low. (Iman,1980a). As with all Monte Carlo analyses, increas-consequence region, and pencrally poor in the high-ing the number of samples increases the convergence of consequence region.'the 100-point case led to a spread in the statistical results. Minimitmg the number of repeti-the release in the high-consequence portion of the curve tions is usually of interest to the analyst, particularly for of about two orders of magnitude. 'this result indicates complicated, time consuming calculations. A rough " rule that the " rule of thumb" in this case is inadequate, and of thumb" for Ills analysis is that the rainimum number many more samples would be required. 'lhis analysis, of samples should be 4/3 the number of independent however, used only a single scenano, and the statistical variables for gud statistical eonvergenec (llonano, et al., convergence treating all scenarios, along with their re-1989). It is not clear however whether the rule of thumb is spective scenario probabilities, might behave differently. meant to apply both to the generation of the Complemen. tary Cumulative Distribution Function (CCDF) curve 'the probable explanation for the inadequacy of the " rule and the sensitivity analysis, or lust to the latter. 'lhe fol-of thumb" in this case is that there were relatively few lowing example was des;grn:Co test whether this " rule of samples giving high radionuclide releases, and many cases thumb" applies to highly tralmeat problerns, such as the in which these was no release at all within 10,000 years, present calculation. For this analysis, the low consequence sarnples were far more prevalent, as demonstrated by the pencrally good 47 independent variables were sampled in the present agreement in that portion of the curve.'lhe result of this analysis. The rule would predict therefore, that about 63 exercise points to the need for care in using the 1.llS samples would be sufficient to generate an acceptable method to ensure that enough samples are generated for output distribution; i.e., the CCDF of radionuclide re statistical convergence. Iman, in fact, warns that the sam-lease. To test this hypothesis, the 10,000-year CCDF for ple si/c is highly problem specific (Iman,1980b). In fu-the base case scenario was genenated from $00 lJ IS sam-ture demonstrations, sorne of the more sophisticated ples, to provide a smooth benchmark curve representing a sampling methods, such as Fast Probabill.. tic Perform-statistically converged distribution. Next,5 CCDP curves ance Assessnmnt methods and importance Sampling were generated for the same distribution, but using only (Center for Nuclear Waste llegulatory Analyses,1988), 1001.IIS samples each, with each case emphrying a new should be pursted. l li-1 NUlti!G-1327

o# ,'s', O" s' s'e' / p'

1. a' f

M p o r E eg I s' i .9 'O I I I / a i I -w "E m / I i / ,I p' I i co I 'I / / o o I ,I t y/ Em y / / O i r / / O I t/ I / o I e/ e s * ~ * !,a ",$ y/ a o r -o ,o ~h', e$,- -e m ,e"'s E o 4 .r k A W Le D o .D N '"g" o E s o s' ~ 2 ,o \\ O O 'S l' UU a o e eo i e-l og g t I v h iI I I E a 1 I i i 1 i I I 4 s 5 I i gi3 gI gI c-n e, l t i o o o e-e-

1033 Figure E.1 Statistical Convergence; 100 vs. 500 Vcciars,10,000 Years, llase Case. This graph presents results from an imtial demonstration of stati coability to conduct a performance assessment.

'Ihe graph,like the demonstration, is limited by ti.- use of many simplifying assumptions and sparse data. NUREG-0000 li-2

APPENDIX F-ANALYSIS OF llYDROLOGIC DATA F.1 liitroductioil-skewed to higher po osity values and are bimodal (Figures F.1 and F.2). These results may reflect the difference in An auxiliary analysis of hydrologic data was conducted to

n. atrix parosity values between the nonwelded CMko determine if spatial correlations could be identified for Ilills unit and he welded Topopah Spring unit.

porosity and hydraulic conductivity paraincters. This analysis did not identify any spatial correlation with depth Scatter and variogram plots were generated only for holes for Saturated hydraulic conductivity data or for Calico w hich had enough data to conduct these types of analyses. Ilills unit. porosity data. A large-scale trend of decreasing Data from five holes (USW GU-3, USW 0-1, USW G-4, porosity with increasing depth was identified for the USW 11-1, and UE25a-1)were used in the analysis (Fig-Topopah Sprmg unit, and a small scale corr elation length ure F.3). Separate plots of saturated hydraulic c(mductiv-of less than 40 meters was identified in data fiom two ity, porosity, and distancc we e made for each hole for1he holes in the Topopah Spring unit. Calico Hills and Topopah Spring units. No corTelations with depth could be identified in scatter F.2 Discussion plots and variogram plots of saturated hydiaulic conduc-tivity from either the Tepopah Spring or Calico llills l the identification of spatial correlation is important to units. In addition, no cturelations with depth cou d be perfewance-assessment modeling, because longer cor-identified in plots of porosity data from the Calico Ilills relation lengths increase the probability that contami-unit. nated groundwater pathways will be encountered that might provide quicker transport of radionuclides to the llowever, a trend of decreasing porosity with depth was water tabic. To look for correlation lengths, the program identified in scatter plots of some of the holes in the GEO-HAS (Geostatistical Environmental Assess.nent Topopah Spring unit (UE25a-1, USW GU-3, and USW Software)(Englund,1988) was used to genernic scatter G-4)(1 igure!, F.4, F.5, and F.6). This trend may be the plots, histogram plots, cumulative distribution plots result of increasing welding with depth, resulting in de-(probabi4ty plots), and variograms of depth, pmosity, and created porosity with depth. I orosity vanogram plots of saturated hydraulle conducthity data for the Calico liills the Topopah Spring unit for two holes USW GU-3 and r and Topopah Spring units. A variogram is a means of USW G-1) contained a pattern, which could be due to the quantifying the commonly observed relationship that trend noticed in the scatter plots. When the trend was samples close together will tend to have more similar removed, there nppeared to be spatial correlation dis-values than saraples far apart. in this analysis, the scatter played in variograms for holes USW 0-4 and Uli25a-1 plots were used tolook for trends with depth, whereas the (Figures F,7 and F.8), in both cases, the variogram has a . variograms were used to look for spatial cortclation in the sill of 40 meters or less, indicating that beyond a 40-meter vertical distance between pairs of measurements. separation distance, there is no spatial correlation for porosity. Since unsaturated zone hydrology parameters were of interest, this study only used core data. The data input in summary, a large-scale trend of decreasing po osity files were generated from Department of Energy data with increasing depth was identified in data from three (Anderson,1981a,1981b,1984; llayden,1985; I ahoud, holes drilled into the Topopah Spring unit, and a small-1984; Peters,1984; Sinnock,1986; Weeks,1984). In all scale correlation length of less than 40 meters was identi-runs, depth was in meters, porosity was unitiess, and satu-fied in data fram two holes drilled into the Topopah rated hydraulic conductivity was in meters per second. Springs unit. llowever, this analysis did not identify any spatial correlation with depth for Calico Hills porosity - Histogram plots of Topopah Spring and CaHco 11 ills po-data or for saturated hydraulic conducthity in either the rosity data were prepared using all the poro-ity data from Calico Ilills or the Topopah Sprmg units. This result was these units in the database, The histogram plot of the relevant to the flow and transport modeling, because long Topopah Spring porosity values was made fr:m six wells correlation lengths lead to a broad travel-time distnbu-and 200 samples.The histogram plot of the Calico 11 ills tion for each column (Section 9.3.1.4). Very short correh-porosity values was made from six wells and 174 samples. tion lengths lead to the conclusion there is a sin;le From the histograms, it was concluded tha". (1) the groundwater travel time per column and little likelihe 3d Topopah Spring unit tends to have lower porosity values of long, fast, groundwater flow paths. In the flow v id than the Calico Ilills unit; (2) the disttibuuens of transport modeling, it was assumed that there was no Topopah Spring porosity data are skewed to the lower apparent spatial correlation for saturated hydraulic ciin-porosity values; and (3) the Calico liills porosity data are ductivity beyond 10 meters separation (Section 9.3.1.: ). F-1 NUlWG-1317 I

Appendix F o ? bu4 D _- u 8 z =w c 3 g C z L.. ; oc b' 4 E RIITRTRBM-$ $ O O MIM M MRWN .C Z BRE m o, o 1 I I I I o o o o o o o CD LO 4 M N r" (%) AON3003Hd Figure F-1 Histogram of Topopah Springs Porosity Values . NUREO-1327 F-2

_ _ _. _ = _. Appendix l' li - s TC-. u mus-, I'EF ifhhs-N o R 5 ~ Mt24RNE H E

1. NWRR_s_t_iRR&itsWrFi.

8 D 3 k h hh b O J kk5NYkkkhMhMYkNhkMbNkIhkNkkkhNhIhb ~ ,hw emhmmnew&' w m45hMMMmNaM51sste9sMMy: bbkhk ~ U5 O W h khkkkk- - - O* ~ m n. hkhhk a O RE ~ O B g in IkkN. Ikh._k.h. - E. ~ M BBE!E -, to ~ l i I I I I I C o to o tn o in o in o 4 M M N-N v-- v-(1N3DU3d) AON3nO3HJ Figure F-2 Histogram of Calico Hills Porosity Values F-3 NUREG-1327

Appendix F i il .G-1ti_1 North Peiimeter Drift Boundary .G-4 UE25a-1 Scale I I 3563 ft. l GU-3 Figure F-3 location map of holes used in Scatter and Variogram Plots. NURI!G-1327 IL4 -+ -w^+ =+-+*+es + - v w aw m. m mm am. m

i ) .i+! l' %201. 8 ee 2 e o e O g e T IN ee e ) U T 3 N T - S E OU C G LG e R P N E WR ge e,e* P I R ( E SP TUS $e e4 YT T I ALEH e e e 8#e 1 O S A C O P ee 'e R S H O PO P

• e e

OT e e I o-0 0 . m 0 0 4 2 5 :a s m 2 z 7 w o ms5f !8;te r9;m5 nEs p7f 5?ai 9 =- r ,a z5=Y C - s 4

1

{1 ,\\jill l Ii %25 " 0 32 e 3 O G T E5 1 INU T4OS GG 99 Y L N T PWI O I S RS R w g EU P R T S O TE

• O e

P ALH C OA S HP 4 10 1 O P

• O 9

5 OT e e y y g50 g 0 0 0 0 0 2 2 .n1mD 5 xFn.wO .e m_E8 'L mE5 3cr =k $4 OcL,2 '~s# %a.5mc2' 7 2Em?ci 7& t I/ll1llllll

-. -... _ =. - - . - - ~.. -. -. ~.. -. - Appendix F e o - ~. e eoe bz3 e H,m 3 o-C E 5 e ese e m 2 mDE .e e g $amW o. es e z 4 0 gxg e o ee o Ao e ee -H e e e 'r0 i i i = l 8 8 8 8 v m N-

e-l (SH313W) Hid30 Figure F-6 Scatter Plot, liole Ull25a-1, Topopah Springs Unit.

F-7 NUREG-1327 ... ~.

Appendix F ~ G -8 N. O -EO-L Rw E -o g Z b w E s 4 et: -8 O O E: 4-e e O d WVWDOlBVA Figure F-7 Porosity Variogram, llole USW G-4, Topopah Springs Unit, NUREG-1327 F-8_ _

_.m Appendix F l 6 l h i N r-O 9 -N e-W O g E E O o 2 ~8h g, E 0 oc 0O -E 9 ( L 0 9 6 i L l g y j: e. 8 g WVHDOlWVA t Figure F-8 Porosity Variogram, Ifole UP.25a-1,Topopah Springs Unit. F-9 NURIIG-1327 ~.. -, _. - _ ~ _ - -. -.. _. _ _ _. -. _ - _ =.. _... _.....

APPENDIX G-TWO-DIMENSIONAL CitOSS SECTIONAL, FLOW MODEl, G.1 Introduction values presented in iiguie o.1 and Table o.l respec. tively. The boundary conditions w cre: a con (tant infdtra- 'lhe IlYDitOCOIN unsaturated fractured tuff test case tion rate on the upper surface, a constant head at the described a hydrogeologic system comprised of layers low er (water table) boundary, and no-flow conditions on whose matrix properties varied over many orders of mag-the sides. Additionally, all the layers wcre tdted 6 de-nitude. Due to contrasts in properties at unit interfaces grecs. and a dip (average dip of 6 degrees) of the units, it could be expected that water would perch or preferentially rnove down gradient with a horizontal velocity compo. G.4 IlestillS RHd ConcluSiOUS nent rather than move only vertically. '!he diversion of flow at the intet faces was investigated by '!he degree of horizontal flow is an important considera. simulating the llYDROColN test case with different tion because: (1) above the repository, flow diversion infdtration rates of 0.1. 0 2, and 0.5 rnm/yr (for the 0.5 could lead to a reduction in flux through the repository; mm/yr simulation, the low conductinty apper layer was and (2) below the repository, horizontal flow could lead to not included, because the infiltration rate was greater a sho1er path to the water table. ilydrologie modeling than the saturated conductivity of the layer). The results can be usefulin identifying the conditions (e.g., flux rate) of the simulations, presented as the ratio of honiontal to that lead to horizontal flow and the influcoce these condi. vertical flow immediatcly above an interface, are pre-tions have on flux through a repository and the geometry sented in Table G.2. Vettical flow dominated in all units of tmvel paths. when the infiltration rate was 0.1 mm/yr. When infdtra-tion was 0.2 rnm/yr or more, horizontal flow was at least an order of magnitude higher than vertical flow above the G 2 Purpose low conductivity unit (1ayer c). The horizontal gradient, a result of the tilted bedding of the layers, and the low Ilydrologie modeling of unsaturated fractured tuff cur-hydraulic conductivity of the umt underlymg Iayer C are rently is generally limited to a single-continuum comlq-tlie primary reasons that a significant component of vehic-ite porous approach for the treatment of fracture /matnx ity was in the horizontal direction in the lower portion of mteraction and is computationally intensive, llowever. Iayer C. Although the nonwelded unit (I ayer li) shows a some relatively simple modeling of layered systems can be large com;wment of horizontal flow, this esult was due to donc to:(1) gain insights into the flow diversion issue and the imnosed boundary condition and to the tdt of the how this affects the fluid flux through the repository and layers, rather than to increased infiltration. the validity of the vertical flow path assumption, (2) un-derstand the numerical limitations better, and (3) provide 'lhese simulations indicate that infdtration rates greater simple mitial simulation studies as the basis for under-than 0.2 mm/yr, combined with the 6-degree slope in unit standmg the effects of further modeling refinements. bedding, could produce a significant amount of hoiitontal flow. If similar conditions existed at Yucca Mountain, G.3 Problem Set up these flows could result in perched zones or hicali/ed fracture flow. Ilow ever, a conclusion on what effect hori- 'ihis initial analysis assumed matrix flow only and used the zontal flow would have on overall performance cannot be description of the tuff site as defined by Department of made by this analysis. It is very important to note that this linergy in the llYDitOCOIN Project (Prindle,1987). In anal) sis did not account for the presence of fractures, addition, the VAM2D computer program was used to heterogeneities, or anisotropy in hydraulic parameters. simulate the matrix flow problem. The analysis involved a l'uture modeling efforts should examine the influence of steady-state simulation using the layering and parametric these additional complications. G-1 Nl!RliG-1327

Appendix G INFILTRATION Layer A* Layer B _= Layer C NO NO FLOW FLOW Layer D Layer E WATER TABLE I I F 5 0 200 meters (vertical exa90eration of 2) liigure G.1 flydrogeologic unin and boundary conditions used in the cross setional simulation using the VAM2D computer, opram (note: the figure does not show the six degree incline that was incluced in the sirnulation).

• see Table G.1 for 1.ayer descriptions.

NUltliG-1327 O-2

l Appendix G I Table G.1 Il3draulie l'roperties Used in the Two. Dimensional himulation of a layered Tuft Site (l'rindle,1987) I.ayer. I'mperiles A 11 C D l'. Saturated Conductivity (mm/yr) 0.3 10.000 06 0.6 8.000 Poresity 0.08 0.40 0.11

0. I 1 0.28

'Ihickness (m) 26.8 38.1 130.1 20'. 1 130.3 van Genochten Parameters: Alpha (1/m) 0.00S21 0.015 0.00567 0.00567 0.016 Deta 1.55S 6.872 1.798 1.79x 3.872 ' layer A : hkxicrately to densely welded, devitidied tuff Iayer 11 : Partially welded to nonwelded vitric tuff layer C-hhderately to densely welJed, devitrified tuff Iayer D : hhderately to densely welded, devitrified tuff layer 11 : Nonwelded vitric and zeolitic tuffs c Table G.2 Itatio of Ilorimntal to Vertical l'ims at the Interfaces littween Different flydrologie Units eter Differing infiltratiot. ?8'ts Itatio of Ilorimntal to Ve rtical l'iow Infiltration llatch Interface 0.1 mm/3r 0.2 mm/yr 0.5 mm/3r I ayer A (K,. - 0.3 mm/yr) 0.05 0.07 layer 11(Ks.i - !0.000 mm/yr) layer 11(K,i - 10,0(K) mmlyr) 0 07 l$.0 19.0 3 1ayer C (Ksat - 0.6 mm/yr) layer C (Ks.i = 0.6 mm/yr) 0.005 0.005 0.005 layer D (Kui = 0.6 mm/yr) layer D (K,ai - 0.6 mm/yr) 0.01 0.01 0.01 Iayer I!(Ksai - 8J)00 mm/yr) layer li(K - 8,000 mm/yr) 0.14 0.19 0.19 3 Water tab *e G-3 NUltliG -1327

Al'PENDIX II-ANAL 3 SIS l'Olt DitlLI.ING SCENAltlO II.1 IllirodtlClioll if it is assumed that tias dnihng occurs in a random f ash. inn, with no memory of previous drilhng, then a Poisson The analysis for the drdling scenario largely followed the distnbution may be uv;d to desenbe the probabihty of N concepts discussed in the U.S.1)epanment of Energy's hucholes being dnlied into the repository over a period OX)l0 Sitt Charactentation Plan (SCP) OX)li, lW) of time,01, $9ch that: which were used to make an estimate of the expected partial perIonnance menure O!PvM)in that document. (Rot)N esp F Ro t ) (Il.2) Ilow ever, the analysis in the SCP was expanded upon and PtN) u NI modified in some significant ways. %e drilhng scenmo was, in some ways, the archetypical direct release scenar-lyencrat, under this set of assumptions, there can be any io, and it was anticipated that many of the appro:u,hes to number of hircholes over a particular period of time; ~ analping both the probalulities and the consequences of however, the form of the datnbution given in Equation this scenario wul ' be exten'Jed to sinalar scenanos, with 11.2 ensures that the expected value of the number of appropnate mothfications. boreholes, s, is equal to Rtt. Without too much diffi-culty at this stage of the analysis, a somewhat more gen-1 ll.2 SCCHario Prohnhility nal approach to drilhng probabihty could be taken by assurmng a Weibull distnbution, rather than a Poisson To analyte this scenario, many of the coraepts used m distt:bution. l'or the Weihull dist-ibution, formulating Appenda 11 of 40 CI'R Part 191 were used. p [} t>f { 0 {(t -Il/6) up( - [f t - O/d Although Appendix 11 is offered ai guidance and is not bindmg on either the I oli or the U.S. Nuclear Regula-p(t) = tory Commission, the concepts expreued were a usef ul O t < l, stanting pomt for this initial analpis. Two fundamental ideas behind the dniling analysis werc:(1)Ihat the institu-tional memmy and control prevc ving disturbrnee of the i or this analysis, the hication parameter, I, would be repository fail alter some period of time, and (2) that the taken to be the time at which drilling is assumed to com-permanent markers at the site failin their function after mence, To. The scale parameter,5, would be taken to be 6 some time. After the greater of these times, it was as. 1/RcT. The shape parameter, P., could he chosen to rep-sumed that drilhng for econome resources commenced. resent a gradual change frorn a /cro rate of dnlling to the it was assumed that thU drilhng occured at the same late constant value R used in the "oissan distnbution (I!qua-of drilling as today, for the type of rock involved. Ilecause hon i1.2). Of co~rse, the diu.bution would need to be of these assumptions, a natural approach to the analysis suitably modifica to accou,, for the finite number of was to assume that drilling occured randomly in space and hucholes, N. Ilowever, this refinemen' was not used in time, and that it could be effectively described as a pois-the dn!hng analysis for Phase 1. son proecss. A!! hough these are rather sweeping assump-tions, siable competing hypotheses appear to be at least I or the purpose of this scenario class, the time penod of as speculative or arbitrary. l'unhermore, the purpose of intuest was the time between the commencement of the analysis was to revaal any weaknesses in the design or dothng, Ta, and the time duration of the simulation (ii., siting of the repository, so if these assumptions preserved the period of time over which the perfonratnce of the important relationships between the important variables repository was to be estimated, e.g.,10,000 )canL Tp. aftecting the perfonnance of 'he repository, then their That is: inherent truthfulness may not be important, OI 3) at = Tp-Td II.2.1 l'robability Analysis Ity combining I!quations 11.2 ad 11.3, one obtains the i To begm the analysis, one msumes that exploratory dn!'. probability of N boreholes p(r,:trating the site as: ing takes place at a constant rate, t (in uruts of per 3 car per square kilometer); therefore, the rate of drilling over the (R(Tp - T ))N o p F R(T -- Tg) d p (II 4) repository area is: PIN) = g R = rAr 011) and the probabihty that no boreholes penetrate the site from Equation 11.4, by setting N a 0, where R is the rate of drilling mio the reposiMr), and Ar exp( ~ R(Tp - T )) 01.5) is the area af th( repository. P(0) = d 11-1 NURI.G-1327

1 Appendix 11 l lloth of these probabilities ansume that Tp > T : that is, d or that drilling started sotnetime before the end of the pc-riod of consideratien. If drilling started at a later time, Nr A then I!.quations 11.4 and 11.5 must be tcplaced by, I,e* T ; d I'7I P(N) = 0 O I.4 a) where Ap is the projected mtercept area of a waste pack-age on a horizontal plane and N t a the total number of and waste packages. P(0) = 1 (ll.$a) i:or vertical emplacement of the waste packages m the repositor), the projected intercept area is a cucle, with a ,lhe probability of the drilling scenario, Ps, that is, that radius equal to the som of the package radius anJ the drill some drilling into the s epository occuted, is gicen by: radius (1 igure i1. !r.). Thus, lor s er tical emplacement. Ps = 1 - P(0) Ap = ct (rp + rd) (III") if where r is the radius of the waste packare and ra isthe Ps = 1 - exp (~ 11(Tp-TU WO d 1 or horizontal emplaecment o the waste camsicis, the f For the assumptions used in the SCP, (r = 0.0003 projected intercept area is a r, eta,. '1e, with height equal boreholu, per square km per year, Ar = 5.1 square km, to the sum of the package diametcr o sd the drill diameter Tp 10,000 years, and Td = 0) (DOli,1988, P. and width equal to the sum of the pael are length and drill = 8.3.5.13-83), P(0) = 2.2711-07, P. " 1, and n = 15.3. diameter (ligure II.lb). Thus. fm aorunntal emplace- 'lhis means that the 'ikelihood of no drilling is very small inent: and that, on average, at least 15 horcholcs will be drilled at the site over the 10,000-year peric J of consideration. A c {2(rp + rdH [1,4 2rdl O p The previous discussion csublishes the probability for the where 1. is the length o the waste package. r drilling scenario ovemil a d for rne probability of N borcholes being drilled on ne repository site. Ilowever, 1:or the current repositor) and waste package design tr p the analyus of this scenario e more complex, because a = 0.34 rn, l. = 4.3 m, NT = 1S.000. A, = $.1 squat e krn, borehole could either pene;r.'c tSe emplaced waste or and assuming the dnli diameter is 6.0 cm), the possihihty merely excavate some 0: the su.munding host rock. In of waste excavation for vertica' emplacement is: effect, embedded in this scenario is a two-branch cient tree: Pe - 0.001518 l drill excavates wave and for horizontal emplacement: and host rock drilling occurs l Pc = 0.01139 on ute I rill excavates only 11.3 Consequence Analysis d host rock if only host rock, and therefore, no waste, is excavated, With such small target. strike probabilitics, the usual out-then some radiological consequences may occur, because, come would be to excavate contammated host rock rather in general, the host rock will be contaminated to some than waste. 'Iherefore, it is important to estimate the level by the movement of contaminated groundwater consequences of excavating contaminated host rock. from the repository. The probabilities and consequences of these two event-tree branches need to be considered in 11.3.1 Volume of Waste Excurated this analysis. First, one considers the volume of waste that would be First, one considers the probability of the excavation of excavated if a borehole penetrated a waste package. Tak-waste, given that drilling occurs on the repository site, f e. ing into account the manner in which the probabihty of Assuming that the interception area of drilling is small excavation was calculated, a conservatn e assumption can compared to the repository area, then one finds: be used;i.e.,if the waste package is touched by the drill, then the entire cylinder of material excavated from the p, intercept area hori/on of the waste-package top to the horizon of the 0 repository area waste-package bottom is assumed to be solid waste. l'or NUluiG-1327 11-2

Aprendix 11 rd + W-11 1I Borehole --Waste Canister L , ;.". Q, y j e' r ( I Aintercept Area Ii 11 I i 17igure Il-la Intercept geometry for vertical emplacement. II II i l Borehole Wastc Canister re Waste Canister U / Y l f p k + + r# M o ' Borehole !I ii SIDE VIEW TOP VIEW liigure li-lb Intercept geometry for horizontal emplacement. 11-3 NUlll!G-1327

Appendix 11 boreholes just touching the perimeter of the waste pack. 1hc concentration of waste in the host rock denends on a age, or only partially overlapping it, this is cicarly a con-number of factors includmp, the solubility of the waste, servative assumption. For this assumption, the volume of the length of time the waste has been dtssolving, the excavated waste, V, for vertical emplacement is given speed at which the waste is being dispersed in the ground-e by: watcr system, the porosity of the rock, and the degrc of sturation of the rock. As an upper limit on the concen. Vur g (igg) tration of waste in the rock (closely following the assump. 2 d tions for waste teach:ng used ir: the groundwater release reenario classes), it can be assumed that the water is at the For horizontal emplacement, the situation is somewhat saturanon limit for the uranium matrix and that the rock more complicated, beca.use the length of the column of was fully saturated. Then, neglecting sorption en the rock excavated waste depends on the hication on the package and accounting for waste only dissob ed in groundwater, at which the drill impinges (Figure 11.2). If h is the height the concentration et waste in' the host rock is given by: of the column of waste and x the distan e from the center of the drill to the midpoint of the waste package, then r=ACeF (ll.10b) C s 3 h = 2 (r * - x)1/*' where C, is the concentration of radionuclidesin the host E rock (Ci/m3), A, is the specific activity of the emplaced where x = < 0, rp > and uniformly distnbuted over the waste (Ci/MTilM), C is the solubility limit for the ura-3 indicated range. niurn matrix in water (gm UOg/gm 110), e is the porosity 2 of the mck, and F is a conversion factor, where The average height of the waste column,15, is: F= (Llid)6 MT/gm) x (1.lH 06 cc!m3) x . nr (1.0 gm/cc of 110). 2 h = - ;p Then, the ratio of concentration in the host rock to con-centration in the waste will be: whereasthemaximumvalucof nis 2rp, Since the ratio of the maximum value of height to the average value is 4/n, a slightly conservative assumption would be to assume C /C (11.11) w r w that the m:uamum height should be used in calculating (Ow / NT V) p the waste volume. Ilut since the values are so close, either choice is reasonable. For a very detaded analysis, in w hich For the values assumed in this analysis (c = 0.36 - a high a great many simulations would be run, the location pa. representative value, and C = 0.001 - the upper limit of 3 rameter, x, could be treated as r4 random variable selected the range sampledh the ratio indicated in Fquation iI.11 from a uniform distribution; however, this appears to be becamt 3.611-04. 2.49 Thus, for equivalent waste vol-excessive for this analysis. Therefore, the average volume umes, the amount of radioactivity released by excav. ting of waste excavated, for each borehole peaetrating a hori, host rock will be about 0.01 percent of the amount re-zontally emplaced waste package, is taken to be: leased by excavating waste directly, given the assumptions which tend to overestimate the amount of contamination )2 in the rock. V= (IISb) e 2 Ilowever, in addition, the rock could be contaminated in much of the space below the emplacement horizon. H,3,2 Concent-ation of Waste Released Given a single borehole, the length of a cylinder of con-taminated rock could be as much as the distance from the The need w hether to consider the excavated rack, in addi' emplacement horizon to the water table (assuming that tion to the excavated waste itself, depends upon the rela' the much larger quantities of water in the saturated rone th e concentrations of radionuelides in the two locations. will substantially reduce the concentration).This is a con-Now, the concentration of waste in the waste package servative estimate of the amount of contaminated host could be considered to be: rock, since there may not be enough waste emplaced to raise all the vadose water below the repository horizon to ^5 C.#= N (ll.10a) the saturation limit. According to the SCp Overview Y (DOli,1988), the static water table is 1300 to 2000 feet T p below the ground surface, and the proposed repository where A, is the speci3e activity of the emplaced waste will be about 1000 feet below ground surface.Therefore (curies per hilllM),0 w is the total quantity of emplaced the length of a contaminated iock celumn would be from waste (Afl11F !). N r is the total number of waste pack-300 to 1000 fect (approumately 100 to 330 meters) tong. ages,and V is the volume of a single waste package. For vertically emplaced waste packages, the leng'h of the p NURiiG-1327 11-4

i Appendix 11 Borehole #1, Borchole #2, Not Along the f fAlong the Radius of the Canister canister Radius l, y\\ aste Canister W a

c ix 4

l End View f I h of Waste j Canister r /_ y_ p N Borehole Axis Iligure 11-2 Geonictry for volume excavated hy a txtrehole through a horizontally D emplaced waste canister. 6

Appendix H intercepted waste column was 43 m (SCP, p.S.3.5.13-83) ts gnen by the pioduct, N6, and l ln is the ms entory i (DOE,1988), and the averape imercepted waste column present in the repository as a function of tune, for hori/ontal emplacement was 0.53 m. Thus, the ctm-taminated rock solume could be from 77 times the inter. The available imentory in the icpository at any time ecpted volume of waste, for vertical emplacement, up to changes, because of: (1) radioactne decay and, m some approximately 630 times the intercepted volume for hori-cases, production, and (2) radionuclide dissolution and rontal emplacement. Therefore, the ratio of conse-migration caused by grounJwater. Changes due to taJio-quences between rock excavation and waste excavation ective decay are treated using conventional equations. would be approximately 0.011 and 0.091, for venical and Changes due to dissolution and migration t. irnpot tant horizontal emplacement respectively. Since the probabil-to treat in this type of analysis because of the coupling ity of excavatmp of waste is only about 0.0015, for vertical between onlling into a repository and groundwater trans-emplacement, anJ 0.0114. for hori/ontal emplacement, por t. That is, it might be ex pected that a " light" r c;witory the excavation of contaminated rock alone could contrib-would be mot e vulnerable to relcases by dnlling, because ute significantly to the consequcnces of this scenario. more of the waste would remain in the canisters. Alterna-That is, the meremental risk (i.e., the consequences of ths tively, a " loose" repository that t cleases a lot of radioac-event multiplied by the probability of the event occurnng) tis e material to the geospher e, beginning at an early time, from the excavation of rock could be 73 to 8.0 (vertical would be less vulnerabic to drilling because less waste and horizontal emplacement, respectively' times that of remams in placcJlhe coupling is not crecisely this dear, the excavation of waste alone. because, as discussed preuously, substantial conse-quences could result from excavating contaminated rock. If the sorption of radionuclides by the host rock were accounted for, the sig..ificance of the excavation of rock 'I hese two factors could be considered explicitly, I y wrib could be prcater than the previous estimate; however, if ing: radionuchdes were soUed, tnen their concentratmn in the groundwater would decrease. l'or very long times and Ign = D'(OL(t) (11.13) for radionuclides vath long half lives, it is possible that the entire rock column down to the water table could be at the concentration limit for a particular radionuclide and where D,(t) is the function of time desenbing radio-that the tock, itself, could be contaminated by sorption to nuclide decay and pioduction (whtch is radmnuclide-several times that concentration limit. Such considera' dependent),and I(t)is the function of time desenbing the tions, which were omitted from the previous estimates, removal of inventory from the repository by dissolution could more th:m compensate for other assumpt ons that and miermion (this function can be radionuchde-i may have ovesestimated the consequences of excavatinE depende'nt also, but it was assumed to be the same for all host rock. For example,if a saturation condition of 0.2, on ra&onuclides in the Phase 1 demonotration) average, was assumed instcad of a value of 1.0 (corrplete saturation), the previously estimated consequences 1:or decay only, would be reduced by a factor of 0.2. Regardless, the result of this evaluation appears to be clear: consequences fcom D (t) = 1 (0) exp(-n t) (H.14) i i i the exca,ation of host rock should be included in the model. where l (0) is the emplaced inventory of radionuclide i i 11.3.3 Consequence Estimation for the entire repository and a i s the decay constant for i the ith radionuclide. 11.3.3.1 Ihcavation of waste To consider decay and production, the llateman equa- .Io begm. a more precise consideration of the conse' tions must be used mstcaJ of l'quation i1.14: how ever, for quences from this scenano class, one considers the consc~ the Phase 1 analysis, only decay was considered. quences of excavating waste first. The consequence of excavating waste by a single borehole at some time, t, is 1 or the source tenu treatment used in the analysis of just the release of radionuclides at that time: raWonuclide release to the groundwater, as discussed in Section 5 of this report, it was assumed that releases C (t) = b ;(t) i V ~ beFan from the engineered turrier system (Ells) at some 1 (11.12) T time after closure, T,and that release from the uramum s mainx occured at a constant rate, untd the entire matrix where C,(t) is the incternental release of nuclide i at time was gone after pawape of the time period, Tg. This 1, Ve is given by Equations ll.9a and 9h, VT s the total means that the function, I.(tt in Equation 11.13 is given i volume of waste emplaced and as used m Equation It lua by: NUREG-1327 11 4

Appendix 11 Since C (t) and exp(-R61)do not depend on the number 1 t < T, i of boreholes, N, these terms can be taken from under the ~ T s t < (Ts + T ) (11.15) L(t) = { l -. 3 t (lM t)N = C (t) exp(- Rot) I t 2 (T 4 T ) i(t) O 3 L i N = 1 (N - 1)1 w here T, is the time at w hich teleases from the 1:DS start or and TL s the time for the waste to move completely from i the !!!!S to thr peosphere, = C (t) exp(- Rot)(Rot) I (ygg)N-1 = C (t) i N' = 1 (N - 1)I 'the previous equation assumes, implicitly, that the amount of waste excavated by drilling is small compared or to the total amount of waste in the repository, since the inventory is reduced in time only by the amount of radi- = C (t) exp(- Rot) (R6t) 1}. (ISt)N onuclides removed by groundwater and not by the C;(t) i nmount removed by drilhng. or it is clear from liquations 11.12 through 11.15 that the consequences of excavating waste by a single borehole i(t) = C (t) exp(- Rot) (R6t) exp(+ 101) depend significantly on the time at which such excavation i occurs. One approach to treating this issue is to simulate a number of realizations in which the number and drilling or times of the boreholes are random variables. Another approach would be to divide the time interval of interest d (t) = C (t) (Rot) (11.17) i i into subintervals and to use a representative consequence for each interial.The npproach chosen for this Phase 1 Ily taking the limit of liquation 11.17 as 01 approaches analysts was to calculate the expected value of conse-7ero, the new equation is: quences over the entire intesval of interest. Sm.cc liqua-tion 11.12 represents the consequence (incremental in- = C (t) (Rdt) (11.18) crease in radionuclide i released to the environment) dC (t) i i from a single borehole excavated at time, t, the conse-quence of N boreholes excavated at time, t, is just: 'Iten, by integrating this result over the time penod of interest, one fmds: C (t) = NC;(t) (11.16) Tp d (t) = R Cj(t) di Taking this va'ue of the consequence of N boreholes as (I1,19) i Te representative cf the consequences that occur over some time increment, St (where t is some time in the interval 61), then the expected value of consequences, averaged Combining Equations 11.12 and 11.13 with liquation 11.19, one obtains: over all possible values of number of boreholes,is: T i(t) = P(N) C (t) U lt) = Rh D (t) L(t) dt (H.20) i i N=1 \\ TsTo where D (t) can be obtained from liquation 11.14, and I P(N) N C (t) i = i N=1 1 f t) can be obtained frem liquation IL15. 4 w here P(N)is the probability of N boreholes over the time To perform the integration indicated in I!quation i L20, it interval, 6t.110 wever, sinc'c P(N) is given by liquation is useful to note the following: (1) if Td s T, then the p 11.2, the equation becomes: ~ integral in liquation IL20 is zero (no dnlling during the time penod of interest); and (2)if T > T, + T, then the d L integral in liquation 11.20 is rero (drilling commences N *'P ~ M II after all the waste has migrated from the lillS). Assuming = f d (t) = E -I II N C (t) i i' N=1 N! that Td > Tp and that Ta < T, + TL, then one can write: 11-7 NURl!G-1327

Appcadix 11 I As an approximation to the total volume of contamincted mk (me can use: d (t) = R (la+Ig (11.21) i Vy = A dr (11.26) t where A has been defined previously as the area of the t reoository and dr is the depth of rock underlying the u To > T. repository subject to c<mtaminatic g' IG*i Ig(0) (exp(-a3 d)

• 8"P(-81 s)] if T < Ts T

T d As discussed previously, a reasonable assumption is that aI the concentration of waste in the rock is reduced substan-tially below the water table, because more flow is avail-and able to dilute and disperse the contamination entering from the unsaturated ione. Ilowever,if a borehole is used 2 ((14 ai{T,- T. - TO ) exp(-ai t,). (11.23) to extract water from the saturated yone, radionuclides t t 1pa dispersed in the aquifer could be brought into the accessi. T.a t I (1 + ai{T. - T - TO ) esp (- ai.)] b!c environment. l'or the Yucca Mountain repositoty, the i T water table is at dtfferent depths below the repository horizon dtpc:nimg on the lateral hwation in the reposi-tory, This variation in depth is shown in Table 93.1 of d ere' Section 93. liased r n the information in this table and taking the weirWd average of these depths, one obtains d I < Ts Ta " (Is d (ll.24a) an averan ocpth of 249.1 m. H T > Ts A consistent approximation to the volume of rock exca-Td d vated is then given by; and Ta, (I + I II I > T + TL (11.24a) S l-P 3 Then T if T < T + TL p p s

• /A (II'28)

Vf/V'T = fr r r d 113 3.2 Ihcavation of contaminated rock The forrnulas in the previous section provided an esti. As in the analysis for the excavation of waste, une can mate of the consequences of excavating waste. Since the express the time dependence of the repository inventory prior analysis also showed that the excavation of contami, as the product of a term related to radioactive decay, nated rock could be significant and that the probability of D'(t), and a term related to migration of radionuclides excavating rock was much higher than excavating waste, from the repository into the geosphere, L'(t). In an ana-the following analysis was developed to estimate the con-logue to F.quation 11.13, one has: sequences of excavating contaminated rock. I'j (t) = D' (t)L'(t) (11.29) One can proceed in a manner very similar to that used for estimating the consequer.ces of excavating waste. One where Di(t) is the function of time describing radio-can rewrite !!quation IL12 for the excavation of rock: nuclide decay and production (which is radionuelide-dependent), and 1 '(t) is the function of time describing y *a movement of inventory from the repository to the host Cl(t)= l'i(t) (11.25) rock by dissolution and migration (this function cm be Vj radionuclide-dependent also, but it was assumed to be the same for all radionuclides in the Phase 1 demonstration), where Ci(t) is the ine emental release from excavated To be consistent with the analysis for the excavation of rock of nuclide i at time t Vf is the volume of rock waste, one assumes that the decay function for waste in excavated by a single borehole, V$ is t he total volume of the geosphere is the same as for waste in the repository, rock that is potentially contaminated by waste from the and that the removal function for the geosphere, L'(t),is repository, and Ii (t)is the inventory present in the con-the complement of the removal fLactirn for the reposi-taminated rock as a function of time. tory. That is: NUlWG-1327 11-8

Appendix 11 (" ) D' (t) = D (t) 1'(0) i In= y [(1 + ng(Ta - Ts) ) exp(- u,T l - a Tu (1 + aj{T - T }) exp(-a T )] (ll.34a) L i p 3 p and L'(t) = 1 - L (t) (ll.30b) where D (t)isgiven19 !!quation i1.14 and 1.(t)is given by where i 1iguation i1.15. T if Td<T 3 3 'these assumptions comprise a compartmental analysis T= { a for the waste:(1) whatever waste does not decay rnust be i d> 5 d cither in the repository or in the geosphere; and (2) fur. ther undecayed waste moving out of the tcpository must be in the geosphere.This estimate of waste escavated is if T + T < Tp, then: L conservative, because the waste transt orted by other means, such as groundwater or gas movement to tiie accessible environment, nre not assumed to be removed 1 (0) [exp(- ng p) - exp(- n;Te)) (ll.35a) from the corapartments subject to excavation. I T e g lly combining Equations 11.15 and ll.30b, one obtains: when 0 t<T 3 if T < T + T T+Tg~ d 3 L' s L'(t) = { ~ T < t < (T + T ) (11.31) T={ 3 s t' T if Td>T+TL (ll.35b) TL d 3 I t > (T + T ) 3 L and In a manner similar to the derivation of liquations ILl6 through 11.20, one obtains the analogous result for exca-vation of contaminated rock: 0 if Td > Ts + TL ,4 ,,D 1 (0) 1." 2 l(I + U (T -T }) exp(- ni a) - T d ' (t) = R $ D (t) L'(t) dt (11.32) i a 3 i (1 + ui L) exp(-ni(Ts + T I)l (ll.35c) 1 i T to T L As in the previous analysis, to perform the integration if Td < Ts + TL indicated in Equation 11.32, it is useful to note the fr.llow-ing: (1) if Td 2 T, then the integral in liquation.1.20 is I p zero (no drilling during the time period of interest); and (2)if T < T, then the integralin liquation 11.20 is zero p (leaching commences after the time period of interest, so Ts if Td<Ts all the waste is in the lillS). Assuming that To < Tp and T={ (ll.35d) 3 that T > Ts, then one may write: T if T > T p d d 3 dj(t) =R (1 y + 1 ';) (IL33) j If the events of excavating rock and waste were mutually exclusive, then one could just multiply each consequence by its probahtlity of occurence and sum to find the average if T, + TL > T, then To < T + TL and: corsequence. Ilowever, every tii 2 waste is excavated, p there is nothing to prevent the column of host rock from Ig = 0 also being excavated (unless it is assumed that drilling would stop if waste were brought to the surface). There-and fore, it was assumed for this l'hase 1 analysis, that the

Appendir il consequences gis en by INuations 11.33 to 11.35 occurred average comequence of excavating host uvL. 'lhat is with a conditional probability of 1.0 (i.e., if drilling oc. curred with protubility given by 11guation 11.6). To find C (t) + C i(t) (11.36) C (t) = P g e the average consequences, the consequences of excavat-i ing waste, as given by I!quations 11.21 to 11.24, must be multiplied by the conditional probability of such excava-where C,(t) is the oserall aserage consequence and the tion Pc, as given by I!quation 11.7, and summed with the other tern 4s have been oefined presiously. NURIIG-1327 1l_10

.- - ~-:.~..-.-.. - ,.- - - ~ ~. Appendix H DRILLING SCENAlliO E '5g,

• h

-i / --ew.- ./; 4.. 0. ~**"~..a....; y..$'

c., a;.

9....y@;. 1 : .. 6 t.c:..... z. ; ; y.... i...; ..l.l, *. [.,'.*a. '/ U 8.; '..- "O C. *.

• [.

?;..'*;;. ! ] ;.*. *;T*.;) ...:;.. i... 4.'.. ?.: ...... r f.y ' '. $..'4,;..d ll ~. i...,..... ll:i' : ....'. 4 ?... :... ..;. b. : l ? ; *.:. r. .?.'... '. :..,..* '*.' 4....n,,;*.;i:.;.*?;,.?;$:; s ? q. t ' '..:...;: d; ' ? ' '.,. %,'.;-}. '.*,:.c: :.'.l.M* ;;A.. ;, *:.;. - .. c. 9: i '. ..:. ;.; t.'";. ,.;,. : e.. .. *. *.* z *...e-

,,.? ;

la..:?. t;:.F ?? ..,o. c.

• .l.....
• *ll:..l. h *;.h. t y '*':l;'. '.".y$s. ;. W.*

!;j;D.,.;.;,?;ci. *lc :?y.*..; 1p' ; c.x:.;. :g. ;... ';.x. .-.;f.r .:y.;:.r.;*... ;- .'....%, W.l;Ql' efte'ho1Q*:f; ..j:R.. \\... i;.Not intersectin0W.....' 1 ,. : [h ........e... c...M. 5.. ....,. t /,. 1..y;. W ;*.:*l... f*. 9/,,Cir. l: ".',;.*?; : " 8>'..,. ;. '., 7.;: ; r.::'s.W::. ' :.}l:;; ;; z..; :ll;.;;;t:;.:..:.. y.*;,':. Weste. :C.an..Iste.t$..::. i.:.. ;;;;. b ln.' .....*..;... :r; ..,.c ,;.c::.,s', :. 1?. *:........ w m:.;..:. i.. e...:6::: :..: :. : e .....w.. ~.?..? :?.a r. '..u....;;;;;:c.a.c...::.. <.....;. n* l; .:.?: . z..v: w... a.,,.. g:.e'. ll.c.;O...**.Y..*b...:;,. :;. :.. i..:; i..' ~..?.:.T. ..y...

l;'. ;.;*l 5,;*),g. ra;.. ;i:.,.tl. :. * :v.r..'. & Boe eho1e

. *.';.;. l. s.'.;f *4

. ;r :;.

.. ;y.. '.c. y, :.......l;: . intersecting

6...:.:?. :: q.;?s't,s:...s;;y U..

N.*;;;,[ag:. ~,Md.3 ..b ?. Waste, Car! s'tet T 7'y[d,f.*ddj;';.'le'.Q 'J'.Of t :l-$Ni:!,'.'

U.

. :':. c..'5'a>;Q; p..l,'IIf.J' .Q" .'.q;:'.R 1.:::.'.'. v !;', 4..*: ?.';.. b '.:.;7 T l q. ?;O l.}.\\ :l$%. l f' l {l},'.h?)h\\ ,i5 f h q.I.bh:.ph!ib.t.? l; h $,$ glIj(pri, '$k 5'Wi7 v.dk e Jf;i .hc.' Yw. w g h'-

l ft v

i f Ernplaced - U f3% b.tg. ti j M@jd%Pakges).~} ?,5,5hhhfIh' 4I h' h,h

• % 'h

$p$Ny g@ M b,.afI!!f4.d @ N$s s[w&; py e t ) r, e -3'- re @M bd!dh

1. aw[@$

t QY6l&NWR$g{$i2b.Q?8Rtfi@%$ Q hi$$se[d$Mfh5NNN k R $y W S@p$$n e e y fWb@ 2 W 4@: enM.5 4 pg 22s2 .hY.Y'O.Nl 5G0N.?, ' Ye$" $ 5 + 5 $ b Yo W t'Ao ^ = _-

m w.

._.=::-. -- =.=,- =,

== ~ = =. = - - Figure H Conceptual view of drilling through waste canister and contarninated rock below the repository and drilling into contaminated rock. H-ll NUREG-1327

l\\ ' Apiwndix H TIME CONSIOGRATIONS Time Line i Waste Package End of Failure Leaching I 1 l i I l I n I I i 1 i Waste in l I Package l I l I l s' l l I l I l l I. I L l I I I Waste in l l Host Rock I l l L l I Figure H-4 Depiction of quantity of waste in packages and host rock from the time of waste emplacement until the end of teaching (all waste leached from waste package). NURI!G-OT' H-12

4 APPENDIX I-SYSTEM CODE STEPS .l.1 DISCUSSION C. Determine (from SYS.INP) whether the groundwater release pathway is accessed for The following is a more detailed step by-step outline of radionuclide release from the repository, the system code operation than that provided in Section D. If the groundwater release pathway is accessed, 4,4,4, la) Run the groundwater flow and transport 1. Set parameters and determine the dimensions of the model(NEFTRAN)(if the run is inter-neccessary arrays. nal). Ib) Request the name of the ground-water model output file to access (for ex-2. Open input and output files, ternal system code runs); the file name is supplied by SYS.INP. A. EPAllM.D AT:an input file of Environmental -2) Open the groundwter flow and transport Protection Agency (EPA) release limits (40 CFR Part 191, Appendit A, Table 1) for 28 model output file' selected radionuclides per 10,000 Metric Tons

3) Read in the radionuclides and cumulative Ileasy Metal (MTHM) releases for each parametric input vector until all the data are input ta the program.

B. SYS.INP : a file of analyst-supplied system code input for a particular run, consisting of

4) Call the ORDER subroutine to input / output flags, a flag denoting the manner a.

Compare the released radionuclide of execution (i.e., internal v3. external), and names against radionuclide names mformation about the specific scenarios to be for which EPA limits are given. modeled in this particular run b. Calculate the normalized radionu-C. SYS.DAT: a detailed output file, with the type clide releases using the EPA release and amount of data placed in this file depend. limit weighting factors (see step 4 ent on the system code input flag values above). D. - CCDF.DAT:an output file which willcontain four-dimensional array (CUMREL), data needed to graph a complementary cumu-according to scenario class, released lative distribut,on function (CCDF) i radionuclide, input vector, and re-lease pathway. 3. - Read i_nto the program the input / output flags, simu-l lation time, and scenario information from the

5) Return to the main program from the l

SYS.INP file. ORDER subroutine. E*""

1. ^** E"

• "I * " "*

4. Read in the EPA release limits from EPALIM.DAT c ssed for radionuclide release, continue with and calculate a wei hting factor for each radio-8.. nuclide (based upcm an imtial repository inventory step F below. of 70,000 MTi!M) using the formula: F. Repeat steps D and E above for the gaseous release pathway and then for the direct release weighting facter pahay, if either pathway is accessed for radio-for nuclide release. (This information is in radionuclide n 7 x (Release lirnit for radionuclide n) gygyp 5. Sequence through the analyst-suppl;cd scenario G. Return to stcp A and determine, by checking classes. SYS.INP, whether more scenario classes are to be modeled. If so, continue at step B; other-A. Select the first scenario from SYS.INP. wise, continue with step 6 below. B. If the system code is executing internally, run Summing Calculations the latin Hypercube Sampling (LHS) routine 6. Sum the normalized radionuclide releases in the to generate consequence module parametric CUMREL array over release pathway, into the input vectors. three-dimensional PEPASUM array. 1-1 NUREG-1327

Appendix 1 7. Sum the normalized releases in the CUMRELarray match is found, the probabilitics are combined, and by radionuclide released into the three-dimensional the duplic.e release values are deleted. REPASUM array.

14. Calculate the total CCDF.

8. Sum the REPASU M array over release pathway into the two-dimensional SEPASUM array. A. Copy the releases and their associated likeli-hoods by scenario from the EPASUM array 9. Sum the PEPASUM array by radionuclide released into TSDF, a three-dimensional array, multi-into two-dimensional FEPASUM array, plying the likelihoods by the probability of the scenario itself, as the transfer occurs.

10. Check for any errors by comparing the results in the SEPASUM array against those in the FEPASUM 11.

Sort l LDF so that the releases are in ascending

array, order, kom the top of the array to the bottom.

C. Compress the TSDF array and recalculate the Calculations for CCDF iad:enuclide release probabilities, as was done

11. For each scenario treated, sort the summed normal.

for the EPASUM array in step 13. ized releases in SEPASUM in ascending order..from D. Create a running cumulative probability,in the the top of the array to the bottom, using the ASORT subroutine, third dimensior of TSDF, by summing prob-abilities from

• he bottom of the array to the top.

12. Place the ordered releases into the EPASUM arrav

15. Generate the system code output files.

by scenario and vector, along with the probability or each consequence, given that each vector in a sce-A. Write the date and time of the run to nario is equally probable, i.e., SYS.DAT. Probabuity or a particular release. I*[,(' []' 3 11. Fill the SYS.DAT ft!c according to the output flags set in SYS.INP.

13. Compress EPASUM by comparing each release C.

Copy the TSDF array into the CCDF.DAT with all other releases within the same scenario. If a output file. e 3 o NUREG-1327 1-2

APPENDIX J-DOCUMENTATION OF FILES AND PROGRAMS ON INEL CRAY XMP/24 FOR REPOSITORY PERFORMANCE CALCULATIONS J.1. Introduction - Package, GRAPHER, on a personal computee. He CCDFIJM program also calculated the average contribu-His appendix documents briefly some significant com. tion to the EPA ratios by radionuclide and sorted them in puter programs, data files, and output files used to gener-descending order, in addition, the program could screen ate and manipulate the source term and radionuclide output vectors on t he basis of limits imposed on the input transport results presented in this report. parameters or combinations of parameters presided to N.EFFRAN. This capability was used in the analyses that examined the sersitivity of the radionuclide releases to .J.2 -I,'ortran Programs the NRC subsystem performance criteria (see Section 9.5). -J.2.1 NEFTRAN6 his is the modified version of NEFFRAN (12mgsine, et al.,1987) used for this demonstration. It was modified This program combined the TAPE 20 output files from . from the standard sersion in the following ways: NEFrRAN for the four columns into a single corabined TAPE 20 output file, The program was developed to a. All calculations having to do whh the determi-avoid: (1) sending four lengthy files from the CRAY sys-nation of the flow through saturated flow tubes - tem to the National Institutes of Health system via BIT-using Darcy's law were removed. Flux was an NET; and (2) a problem on HrrNET, which caused some input variable based on infiltration and fracture of the long output lines to be clipped at 79 columns.The flow, as determined by saturated hydraulic con-revised output file was identical to the old TAPE 20 out-ductivity. put, except that the long lines were no longer written in a list-directed format, but rati.er in formatted form with a b. Most input variab!cs needed for he unsatu-line length of 68 characters.11.Se output for a particular rated flow and transport calculations were con-radionuclide chain was all zeros,,Se list-directed output tained in subroutine GETRV. This subroutine was still used to take advantage of tUs format's compact read the random input vectors on input file structure. ( TAPE 10, as generated by the program IllSVAX, and generated an output vector file J.2A LIISM of radronuclide releases accumulated over time (either 10,000 or 100,000 years), that was writ-This program generated the LilS sample for input to f. ten to an output file, TAPE 20. NEFFRAN6. It was modified from the original version in c. Minor changes were made to the output format of TAPE 20 to include the scenario number on A rimdom number generator, R AN 1, from Nu-a. each record. merical Recipes, was added. J.2.21 CCDFLB1 b. It read in the names of the input and output - files. This program was used for the sensitivity and uncertainty analyses described in Section 0.5. He CCDFLIM pro-J.2.5 STEP . gram took the TAPE 20 output files generated by NEFlRAN6 for the four columns and generated a com. This program performed the stepwise linear regression . plementary cumulative distribution functior. CCDF) for and rank regression on the outputs of NEFIEAN6 for each scenario, it multiplied the output cumulative re-each *enario, to determine sensitivities and uncertainties leases for each radionuclide by their respective Environ-(see action 9.5 of this report).The main m(xlification to mental Protection Agency (EPA) release litnit factors this program was to take the TAPE 20 output from the -(from'40 CFR Part 191) to get an EPA ratio for each four columns generated by NEFFRAN6 and combine vector. The vectors were then combined for the four them in.o EPA ratios for each vector using the EPA - columns, sorted and W2itten to a file for transmission to release limit factors. The combined F.t A ratios are writ-2 the U.S. Nuclear Regulatory Commission (NRC). nese ten to a temporary file and ren.a int; the STEP program to results were plotted with the' commercial graphics generate the reeressiot J-l NUREG-1327

~ Appendix J .l.2.6 C14B t52nl: This is the same as 151ni above, but for Column H. His program calculated the carbon-14 release from the waste packages as a function of time. The program tS3nl: This is the same as 151n t alvve, but for assumed that the canisters failed with a normal probabil. Column C. ity distribution. Following failure, oxygen attacked the fuel matrix and the C-14 inventory was released accord. tS4nl: This is the same as tSin 1 above, but for ing to a rate based on the spallation time, randomly Column D. picked from a uniform probability distribution bounded-by two lines that are functions of temperature. To this (51100: nis is the same as (Sini above, but for release rate was added the prompt release of C-14 at the 100,000 years. Results based on this - time of canister failure. The auxiliary analysis of the re. data was used in the sensitivity and un. lease of gaseous C-14 is discussed in Appendix D of this certainty analyses discussed in Section report. 9.5 of this report. tS2100: This is the same as L52nl above, but for J.3 Batch Script Files 100,000 ye rs. R suhs based on this data was used in the sensitivity and un- ' The following batch file executed programs in the batch certainty analyses discussed in Sectir,a mode on the Cray using the batch queue function QSUB: 9.5 of this report. 153100: This is the same as 153nI above, but for - The STATCON.SUB batch file' executed in sequence 100,000 years. Results based on this with the programs LHSVAX and NEFFRAN6, for all - data was usec' in the sensitivity and un-four columns, and then with the program CCDFLIM, to certainty analyses discussed in Section generate a CCDF,The main purpose of this script file was 9.5 of this report. - to simplify the multistep operation for generating a CCDF, particularly for the statistical convergence exer-154100: This is the same as tS4nt above, but for cisc that demonstrated the sensitivity of the CCDF to the 100,000 years. Results based on this number of latin Hypercube vectors samples chosen data was used in the sensitivity and un-(either 100 or 500)(see Appendix E for discussion). A new certainty analyses discussed in Section random seed for LHSVAX was chosen for each of the 9.5 of this report. runs with 100 vectors. TAPE 20: The random vectors produced by LHSVAX for the input file ympyuc.dat = J.4 Data Files were stored in this file. The following is a list of the input data files used: J.5 Output Files ympyuc2.dat: This file was used for all scenarios to The following is a list of the output data files used: generate the LHS samples for NEFTRAN6 based on the distribution tape 2051.10: These were the TAPE 20 output files and ranges specified for 47 variables. tape 1052.10: from NEFERAN6 for the base-case .When used in the statistical conver-tape 2053.10: scenario, 10,000g cumulative re-gence test (Appendix E), a new random-tape 2054.10: Icases for the four wiumns referred to input seed (specifiable in this file) was in the text as Columns A, B, C, and D, chosen for each CCDF run. respectively, epalim.dat: This file contains the EPA release limits tape 201.500: These wete the TAPE 20 output files by radionuclide in terms of permissible tape 202.500: from NEfTRAN6 for the base-case releases per 10,000 metric tons heavy tape 203.500: scenario, 100.000-ycar cumulative re-metal, as provided in 40 CFR Part J1, tape 204.500: leases for the four columns referred to in the text as Columns A, B, C, and D, Appendix A. Table 1. respectively. These data were used in the sensitivity and uncertainty analyses 151nl: This was the NEFFRAN6 card image

• ##E input file for the basic parameters in Column A, for the base-case scenario, tape 20cmb.10: This was the combined output for 10,000 years, 500 vectors.

tape 2051.10, tape 2052.10, tape 2053.10, NUREG-1327 J-2

Appendix J and tape 2054.10, es produced by pro-ccdfl00.out: This was the ottput file used in plotting / gram COMilINE. the 100,000-year base-case scenario CCDFs in Section 9.5. ccdfl0.out: This was the output file used in plotting the 10,000-year base-case scenario TAPE 6: This was the normal printed output file CCDFs in Section 9.5. for each NEFTRAN6 run. J-3 NUREG-1327

NRC FORM 335 U.S. NUCLEAR REGULATORY COMM>SO ON 1 REPORT NUMBER (2g g) ( Assyned t>y NRC, Add da!. NRCM 1102 Sam i Rev. and Addendam Num-3201.32c2 BIBLIOGRAPHIC DATA SHEET

• * ' * - " M (see instructions on 1*ie reverse)

NUREG-1327

2. TITLE AND SUBTITLE 3 DAT E HtPORT PUbuSHED Initial Demonstration of the U.S. NRC's Capability to Conduct a Performance yoy7g

[ ynn Assessment for a High-Level Waste Repository i May 1992

4. FIN 04 G'3 ANT NUMOE R

6. mUir*QR(53 6 T YPE OF RE POR T Technical R. Codell, N. Eisenberg, D. Fehringer, W. Ford, 7 WR o covt RER ""*"*

• D*'")

T. Margulies*, T. McCartin*, J. Park, J. Randall*

8. PEHFORMING ORGANIZATION - NAME AND ADDRESS Uf NRC, provice Divmon, Office or He@on, U E. Nucica' Heguia1My Comm,sston, and mailing address; ;f contractor, provide name and ma4'ng address.)

Office of Nuclear Material Safety and Safeguards

• Office of Nuclear Regulatory Research U.S. Nuclear Regulatory Commission Washington, DC 20555

9. SPONSORING ORJANIZAI:ON '- NA*,dt AND ADORESS Ut NRC, type

• Same as above'; it contractor, ptovide NHC Cmseon. Off <ce or Rege U S. Nuclear Regulatory Commission, and ma+ ling address )

Same as above 10, SUFMEMLNT ARY NOTES

11. ABSTRACT (200 worot or less)

In order to better redew licensing submittals for a High Level Waste Repository, the U.S. Nuclear Regulatorv Commission staff has expanded and improved its capability to condect performance assessments. This reput .N documents an initial demonstration of this capability. The demonstration made use of the limited data from Yucca Mountain, Nevada to investigate a small set of scenario classes. Models of release and transport of radionuclides from a repository via the groundwater and direct release pathways provided preliminary estir ates ~ of releases to the accessible environment for a 10,000 year simulation time. latin hypercube sampling of input parameters was used to express results as distributions and to investigate model sensitivities. This methodology demonstration should not be interpreted as an estimate of performance of the proposed repository at Yucca ' Mountain, Nevada.

13. AVAfLABILf f Y ST ATEMENT 12 KEY WORDS/DESCR;PTORS (List words or phrases tnat will assist resea cners in locating the report )

Unlimited a secuRm etAsstFicAiiON EPA Standard Geologic repository Unclassified High level waste (Thu R'rart) Performance assessment Probabilistic risk assessment Unclassified 15 NUWBtH OF PAGtS 16 PHiCE NRC F(SM 3'l5 (2-891 t

er .A 4 A r + v. s a _e-g a A THIS DOCUMENT WAS PRINTED USING RECYCLED PAPER

BMEMBMENM m a m m-a n n i x, m, a Y _c,.; m - c;s: Yk b,smnk

• vgyu _q w y a -

n w m

y;

.r. +n ~.- . - M,.pn gth@b ,< m+ \\. awn 7 .(, k-2 m tw 79

1. pmi s.

- ' ye;P M 4 WAp ,, e n 3-5 y p phpa @Mc e[g - " m4 hnh d k k M k g;g; w$ w m m m $es w w %e m[

d. Jhthih,, ',

[)m,, h U g( A. y m wwo ' m# @;a$m$ m$ $ $ m[ x @ w'/ ko hE M6 MMMMMMd M d ;. Aw upn,-n w[N-F e m w pey&.

s w

e m ,1sw,,>,,.. w n,!.

a. A_.

n w4,3,m f. M E.M - q$y$p$$[Op , p# -y k,. s. . ~e,,, ge,pl, y Wg.f ?.. .1 e' .m % k7 Li C,.l , ^vc : q 1 a gh 't i ' ' W"m n 8 <ccc 4, WQ,,pM iQ"b-y: :@n 4 #~ N, a +wa. ' %_ W4 ' ".. v"r e",T Wl.g W: r 4 /an@vthn >yv e c. _h . -SW3y v nA4V. - M2 m c .a n w.- m %mW w. , ~g-V;i W 0 S M w%g~ cU p b 1, ; @M G}@% a3? p C~ R R,'M+ W:% e % mW'. We 5 Q: 'm% %'? s Q f%M '? M 2f%. W MO> MM Y%W& T 'i L ' W M y q h $g~f4Nj &W - ( O t w (C M M $ k & n n L% - zgh t: D 7 N ' b' L j$@dD:: ,i h . #g,M.qpwe.pg . t, a w n[ v.r g a p o $i) ' m.;$ 5m4mk c a w ,:1 s g.g p m %m af_q ~ s Mk@O % % MW5 W MY Y_.$fp@wmgaupq h@ $w w & Y % r } &p l$hrhyI$db NN GMYY ' 39 $ Q@%di J .NNMWQlddW ,p% W W Myk&p4 # h&%:p% - A; ma # W % W W, 2 # w- @4'OfN $m M

xg d.y V

h w vm '; mism gh n y 9 y;i;% w q pd d. m d u yn y ~w w my a,.mn .;uym 7 i '"7J:;n:r, m m'a? $:3 % dm % dM"l""'l6 p %y g @g $ w w MMW.jM m m um e E

• e ge

. d. w% M M@v a m[$hk MShfh; s m- .g[M[gwan[ @ )N e,c S kNddh i M M$ M.3 3 a pc y2 q+e. ;y@a mn , m; n >R% o# v w=n.Mnew.eg;n a w n, mq m , m, ey,m%p +; xyx w w:; m, ws ~ m mg e, g ., ms L+ ..wea., p u . w+MMM;v m My pvM@M@b/M m4 e m.. . - ~ib q; 2 p u c 'w' < ' '" 7iN ME J Oqd:Q*m. > p r:a >wy?Q U D' ' $s ' [S M:p.'4:mN M. &. > : m u"M ~;yp

hu

pric

[ 4% 6% dlMu"M: V "J ' l.

. h.

W -4. sMm MN .[34% e > 4M7 M M;Vh. m,% '. M: ~ W: 9 $%m[ S ' ijd: P, c :7: m 1 'W MM j@!WNQ. 44 k '~t y a@%e%yy

t uq @q yM R F am

h. :. -

($.

p i &

9 v - u 8 qo c y.,ym o,m mm .m..r, .smc-4~ - em-mn c q A ng; 8? W i w A; %aW& w- .MM. h ,r kq g::MM iM n &y %.:i> $:u.smM : n arp n qn w s ? ~w %y #q Q)% Myh rlXL pp R "c:h,S. .y c n Al '3 P d e m%..qqM7W,d(,s@?::t. \\ - pp p mpq?j> ; m* m' ^ "; m 4y+ inyu ; - :v 1 <W .&pyg u

%f ; a

u.n:. a:n% v ? 9y rhP,?f b i./.n ), 4D

.pn g a cap n m& r-%pDW 4 k ' g O+@r y m& ,e

@ d:

- ?? a 4 W-w@$e$$cu@$W ' WhPG@!QB fn ApQ %.Q lc - 3% f Av 6 3 E d(m b %up! 2 = Nf w $$w 2CMf%w; P, i d, w u.n g s go? ? %e$J4e. 1X Wt

i. k.

mAm.u m1p-w+w m... %e -, a.9 1; 3 um m gm rm w,e yb ~y -4 hIhy %yg.-gpm%:v vm+wn%gghwhQ$Wjn; Og YdhD!N. d%wM'b

iMMM - 25

.s.., S t!I N s .=YA WDhf $4Q h y, su y 4 p 4 sgwtw4 x w wp x h' w%n, -ey w nm nw%, 9%c MA w mMg - e'2x. + 4 w 'wmy r'o o em u .h.7 n 4 9 g [ fk b hhh h v h%MN g~hhh5 k M N k ~ m; 5 s. 4. m n$ M W 4 @ Q,f k n n vn wt Q M & a n% n M h M e 4 ? P %p -l W & e i j $y & Qg n heuw w Mmw m % is .M-& M QGM Ne.%s& b O R M w m% n ~ Mi&y a MWM Mk BM

i w%~ NW;p%w$mg %

,o~ e m.nn y n % m. m[g. w% m p e m p A m n n-ct n 3w 06 W/;*'e*,y m mmmg@w s'66 WDM%y . - m & a w@g g' %h $h h @ w m m/p h+?.- g w km 4P y MFW 3

• 1

~ Ap my a+% m%gwan + m % a".$s w~4 TW - m.M TM O r mm ~. &ndy n %w w b : WMgw$ w? E r -qqw H dggg - W %T WWW M k h;y. h bbhbhhk ensp@h /( [hk pwanM y yn w x g w smw n~,. www ews -{g ~c ( A4 m 6 , h wm . xpm% me gy s o g 7 ~ m s k Ydk h \\q k k Wn v $xQ$N)hPMMM-wwww n ^ ^W% %< v" P, MMMW m J.WM f ( n$ @@4 v s% $W @$MM$M!M E N'M @ M a m m m w ee w N %WM EW w WM $[NW8QWM 3w .Cm %,4 5 % M M W .'W J/':4W hih hYNhbh w%au&mmww ' ff Y N N' w' + n YkYh ? i ?: %y $&p%M%gwn@Ww +

w-8 gauM J

7 ga d wiy,My~ C, ",. ?' i @ m@y mem%a&n&n~g;Sp ' cm w J a mm, , m, m e n.,j r- - m n gmx % e n ~ c m mL b .q_ ppgp'" O pp y nm r e, f f(Q., 6dd ? 4lyf sgl M 'T' M k ?? W m Wl f A W kk e mWM ",s' 'Ly-Q. - + ton. MWf . %g y w wdghmg g & g Ng:m. l y+ y q% ,y q v,a p = + -a c r x s Qp.y W aMy y ~JN3 Nr Nb %dt@g sig 4,W~@3 ' W 'wJ i p w.. RN,Q d(G ) &$%j##my/ ki%UW ^ V*' ? + M D N/Niye ww% 4 FNhTM3(&u-diY+k? -

• g.fdbQ J$W N

U f m d -M i 4L i s ~ w gg pp yy g Nk J $ehgM gN?p $ 'OWN~' 4 i g $$ e W; s 3 Li

3gV 1

1 - v c 43 -. 1. !i 9 b 'i! t e3.s l ; 11 2 f ^ . T e g t 4 w. p. g j 1 4 l1 'i~ [ er g c s r-N j +e 4 a. p .y p y N 'k jg 8 n \\ s L. w [* f l Nf i h l ~~ [ f .j y Ng 4'_ k ( o k [ ~ n i s y h c. h: 3 'IN M g. ,r_ f k f i,p 4; W4 ~ j.g t J-9 [.. ? -.. t! e,, 7 N I undesasemen&m}}