Seepage Calibration Model and Seepage Testing Data Rev 03, ICN 00 MDL-NBS-HS-000004 September 2004 1. PURPOSE The purpose of this Model Report is to document the Seepage Calibration Model (SCM). The SCM was developed (1) to establish the conceptual basis for the Seepage Model for Performance Assessment (SMPA), and (2) to derive seepage-relevant, model-related parameters and their distributions for use in the SMPA and seepage abstraction in support of the Total System Performance Assessment for License Application (TSPA-LA). This Model Report has been revised in response to a comprehensive, regulatory-focused evaluation performed by the Regulatory Integration Team [Technical Work Plan for: Regulatory Integration Evaluation of Analysis and Model Reports Supporting the TSPA-LA (BSC 2004 [DIRS 169653])]. The SCM is intended to be used only within this Model Report for the estimation of seepage- relevant parameters through calibration of the model against seepage-rate data from liquid-release tests performed in several niches along the Exploratory Studies Facility (ESF) Main Drift and in the Cross-Drift. The SCM does not predict seepage into waste emplacement drifts under thermal or ambient conditions. Seepage predictions for waste emplacement drifts under ambient conditions will be performed with the SMPA [Seepage Model for PA Including Drift Collapse (BSC 2004 [DIRS 167652])], which inherits the conceptual basis and model- related parameters from the SCM. Seepage during the thermal period is examined separately in the Thermal Hydrologic (TH) Seepage Model [see Drift-Scale Coupled Processes (DST and TH Seepage) Models (BSC 2004 [DIRS 170338])]. The scope of this work is (1) to evaluate seepage rates measured during liquid-release experiments performed in several niches in the Exploratory Studies Facility (ESF) and in the Cross-Drift, which was excavated for enhanced characterization of the repository block (ECRB); (2) to evaluate air-permeability data measured in boreholes above the niches and the Cross-Drift to obtain the permeability structure for the seepage model; (3) to use inverse modeling to calibrate the SCM and to estimate seepage-relevant, model-related parameters on the drift scale; (4) to estimate the epistemic uncertainty of the derived parameters, based on the goodness-of-fit to the observed data and the sensitivity of calculated seepage with respect to the parameters of interest; (5) to characterize the aleatory uncertainty of the parameters as a result of spatial variability; (6) to evaluate prediction uncertainty based on linear uncertainty-propagation analyses and Monte Carlo simulations; (7) to validate the SCM during model development, and validate the SCM using the post-development activities outlined in the Technical Work Plan (TWP, see below); (8) to provide the technical basis for the resolution of unconfirmed issues previously labeled “to be verified” (TBV); and (9) to provide the technical basis for screening of certain seepage-related features, events, and processes (FEPs). The primary caveats and limitations in the scope of this Model Report and the results from the SCM are as follows: 1. The seepage models are intended to provide estimates of the seepage flux averaged over a 5 m drift segment (the approximate length of a waste package). The seepage models are not expected to quantitatively predict individual seepage events or the precise spatial seepage distribution along the drift. 2. By definition, the derived parameters are related to the specific model structure used, i.e., these parameters are only applicable to a conceptual and numerical model similar to the SCM. (Note that the SCM and the SMPA are compatible in this sense.) The parameters are also process specific and scale dependent, i.e., while they can be considered optimal for seepage calculations on the drift scale, they are not necessarily applicable to other processes on different scales. 3. The effective parameters derived in this Model Report capture many processes and features leading to dripping of formation water into a large underground opening. However, this does not include water dripping as a result of condensate accumulation on the drift surface or other in-drift moisture redistribution processes. More detailed discussions of the appropriateness of the modeling approach, the sufficiency of the data, and the inherent limitations and caveats can be found throughout this Model Report. The technical scope, content, and management of this Model Report are described in the planning document Technical Work Plan for: Unsaturated Zone Flow Model Report Integration (BSC 2004 [DIRS 169654], Section 2). This document does not deviate from the TWP; no additional criteria were identified in the TWP. Direct inputs to this Model Report are listed in Section 4.1. These source data include the air-permeability and liquid-release test data described in the report In Situ Field Testing of Processes (BSC 2004 [DIRS 170004], Sections 6.2 and 6.11), calculated percolation flow fields described in the report UZ Flow Models and Submodels (BSC 2001 [DIRS 158726]) and the related numerical grid described in the report Development of Numerical Grids for UZ Flow and Transport Modeling (BSC 2001 [DIRS 159356]), fracture property data described in the reports Analysis of Hydrologic Properties Data (BSC 2004 [DIRS 170038]) and Calibrated Properties Model (CRWMS M&O 2000 [DIRS 144426]). This Model Report mainly supports the reports that document the SMPA (BSC 2004 [DIRS 167652]) and seepage abstraction [Abstraction of Drift Seepage (BSC 2004 [DIRS 169131])]. In addition, the discussions and results are used in the reports Features, Events, and Processes in UZ Flow and Transport (BSC 2004 [DIRS 170012], Section 6.1.24), and Drift-Scale Coupled Processes (DST and TH Seepage) Models (BSC 2004 [DIRS 170338]). This report also addresses the following issues: The development of a collection system in Niche 5 (also referred to as Niche 1620) for mass balance considerations (see Sections 6.5.3 and 6.8); monitoring and estimation of evaporation effects (see Sections 6.3.3.4, 6.5.4, 6.6.1.3, 6.6.1.4, 6.6.2.3, and 6.6.3.3); inclusion of film flow effects (see Sections 6.1.2, 6.3.3, 6.3.3.2, 6.3.4, and 6.6.3.1); inclusion of effects from small-scale irregularities at the drift surface (see Sections 6.3.3, 6.3.3.2, 6.3.3.3, 6.3.3.5, 6.3.4, 6.6.2.2, 6.6.3, 6.6.3.3, 8.2, and Appendices C–E); justification of the continuum approach (see Sections 6.3.2, 6.3.3.2, 6.3.4, and 6.4.1); discussion of differences between continuum models and discrete fracture network models (see Sections 6.3.2 and 6.4.1); and the use of Niche 5 data to improve parameter estimates (see Sections 1, 4.1, 6.5, 6.6.2, 6.6.3, 7.3, 7.4, 8.1, and 8.2). INTENTIONALLY LEFT BLANK 2. QUALITY ASSURANCE Development of this model report and the supporting modeling activities have been determined to be subject to the Yucca Mountain Project’s quality assurance program as indicated in Technical Work Plan for: Unsaturated Zone Flow Model Report Integration (BSC 2004 [DIRS 169654], Section 8.1). Approved quality assurance procedures identified in the TWP (BSC 2004 [DIRS 169654], Section 4) have been used to conduct and document the activities described in this model report. The TWP also identifies the methods used to control the electronic management of data (BSC 2004 [DIRS 169654], Section 8.4) during the modeling and documentation activities. This model report examines the properties of natural barriers identified that are classified in the Q-List (BSC 2004 [DIRS 168361]) as “Safety Category” because they are important to waste isolation, as defined in AP-2.22Q, Classification Analyses and Maintenance of the Q-List. The report contributes to the analysis and modeling data used to support performance assessment (PA). The conclusions of this model report do not affect the proposed repository design or engineered features important to safety, as defined in AP-2.22Q. INTENTIONALLY LEFT BLANK 3. USE OF SOFTWARE The software programs used in this study are listed in Table 3-1. These programs were selected because they are appropriate for the intended application. They were used only within the range of validation; there are no limitations on outputs due to the selected software. The software programs were obtained from Software Configuration Management; their qualification and baseline status is given in the Document Input Reference System (DIRS). Table 3-1. Qualified Software Used in this Report Software Name Version Software Tracking Number Reference iTOUGH2 4.0 10003-4.0-00 LBNL 1999 [DIRS 139918] iTOUGH2 5.0 10003-5.0-00 LBNL 2002 [DIRS 160106] GSLIB Module SISIM 1.203 10001-1.0MSISIMV1.203-00 LBNL 1999 [DIRS 134136] GSLIB Module SISIM 1.204 10397-1.0SISIMV1.204-00 LBNL 2000 [DIRS 153100] GSLIB Module GAMV2 1.201 10087-1.0MGAMV2V1.201-00 LBNL 1999 [DIRS 134139] GSLIB Module GAMV3 1.201 10398-1.0GAMV3V1.201-00 LBNL 2000 [DIRS 153099] EarthVision 4.0 10174-4.0-00 Dynamic Graphics 2003 [DIRS 162369] AddCoord 1.0 10355-1.0-00 LBNL 2000 [DIRS 152814] MoveMesh 1.0 10358-1.0-00 LBNL 2000 [DIRS 152824] AddBound 1.0 10357-1.0-00 LBNL 2000 [DIRS 152823] Perm2Mesh 1.0 10359-1.0-00 LBNL 2000 [DIRS 152826] CutNiche 1.2 10356-1.2-00 LBNL 2000 [DIRS 152815] CutNiche 1.3 10402-1.3-00 LBNL 2000 [DIRS 152828] CutDrift 1.0 10375-1.0-00 LBNL 2000 [DIRS 152816] AddBorehole 1.0 10373-1.0-00 LBNL 2000 [DIRS 152822] ECRB-XYZ .03 30093-V.03 CRWMS M&O 1999 [DIRS 147402] EXT 1.0 10047-1.0-00 LBNL 1999 [DIRS 134141] The use of the software programs identified in Table 3-1 is documented in Section 6 and in the supporting scientific notebooks (SNs). A summary description of the programs and their use is given below. The software program iTOUGH2 V4.0 (LBNL 1999 [DIRS 139918]) provides forward and inverse modeling capabilities for unsaturated and multiphase flow in fractured porous media. The iTOUGH2 V5.0 (LBNL 2002 [DIRS 160106]) program has—among other features—the extended capability of efficiently simulating evaporation effects [Requirements Document (RD) for iTOUGH2 V5.0-00 (BSC 2002 [DIRS 161067], Section 1.2)]. Both programs are used in this Model Report for simulating liquid-release experiments and predicting seepage rates. Moreover, they solve the inverse problem by automatically calibrating the model against measured data, and calculate prediction uncertainties for model validation. The GSLIB modules GAMV2 V1.201 and GAMV3 V1.201 (LBNL 1999 [DIRS 134139]; LBNL 2000 [DIRS 153099]) analyze spatial correlation of, respectively, two-dimensional (2-D) and three-dimensional (3-D), irregularly spaced datasets. These programs are used for the geostatistical analysis of air-permeability data. The GSLIB module SISIM V1.203 (LBNL 1999 [DIRS 134136]) generates 3-D spatially correlated random fields by means of sequential indicator simulations. It is used in this Model Report to generate spatially correlated fields of log-permeability modifiers. Module SISIM V1.204 (LBNL 2000 [DIRS 153100]) is an extended version of SISIM V1.203 (LBNL 1999 [DIRS 134136]), in which coordinates are directly output along with the log-permeability modifiers, making the use of software program AddCoord V1.0 (see below; LBNL 2000 [DIRS 152814]) unnecessary. The following utility programs support the generation of computational meshes. The software program MoveMesh V1.0 (LBNL 2000 [DIRS 152824]) adds a constant to the coordinates of a mesh file, translating the coordinate system. The software program AddBound V1.0 (LBNL 2000 [DIRS 152823]) adds boundary elements to a mesh file. The software program AddCoord V1.0 (LBNL 2000 [DIRS 152814]) adds coordinates to the output file of SISIM V1.203 (LBNL 1999 [DIRS 134136]) in preparation for its use by the software program Perm2Mesh V1.0 (LBNL 2000 [DIRS 152826]), which maps a field of log-permeability modifiers onto a mesh file. The visualization software EarthVision V4.0 (Dynamic Graphics 2003 [DIRS 162369]) is used to extract coordinates of the rough ceilings of Niches 3 (also referred to as Niche 3107) and Niche 4 (also referred to as Niche 4788) in preparation for the use of the software program CutNiche V1.2 (LBNL 2000 [DIRS 152815]), which cuts a niche with a rough ceiling from a mesh file. The software program CutNiche V1.3 (LBNL 2000 [DIRS 152828]) cuts a smooth niche from a mesh file. The software program CutDrift V1.0 (LBNL 2000 [DIRS 152816]) cuts a cylindrical drift from a mesh file. The software program AddBorehole V1.0 (LBNL 2000 [DIRS 152822]) inserts a borehole into a mesh file. The software program ECRB-XYZ V.03 (CRWMS M&O 1999 [DIRS 147402]) calculates the coordinates of a given ECRB station number, so the location of ECRB test beds can be related to the coordinates of the computational mesh. The software program EXT V1.0 (LBNL 1999 [DIRS 134141]) takes the forward output file from iTOUGH2 (V4.0 or V5.0) and converts it into a Tecplot (all versions, see Table 3-2) input file; this software is used for visualization purposes only. Table 3-2 summarizes the commercial off-the-shelf software used in support of this Model Report. This software is exempt from software qualification. Computations performed using the standard functions of the software products listed in Table 3-2 are described in the model documentation (Section 6) and the cited appendices. For visualization purposes, certain units have been converted using the equation utility of Tecplot. A factor of 1/86,400 was used to convert time from seconds to days; a factor of 1/60,000 was used to convert water flow rates from milliliter per minute (ml/min) to kilograms per second (kg/s), which implies a water density of 1 gram per milliliter (g/ml). Information needed to reproduce the work, including the input, formulae or algorithm, and output, is included in this Model Report and the cited references. Table 3-2. Software Products Exempt from Software Qualification Software Name Version Platform Information Used for Microsoft EXCEL 97 (SR-2) PC, Windows 98 Data reduction, computation, graphical representation of output 2000 (9.0.3821 SR-1) PC, Windows 98 2000 (9.0.3821 SR-1) PC, Windows 2000 Professional Microsoft WORD 2000 (9.0.3821 SR-1) PC, Windows 98 Word processing 2000 (9.0.3821 SR-1) PC, Windows 2000 Professional vim 6.0.12 PC, Linux Text editing Adobe Illustrator V8.0.1 Mac, MacOS 9.0.4 Schematic figures Microsoft PowerPoint 2000 (9.0.3821 SR-1) PC, Windows 98 Tecplot 8.0-1-0 Sun, SunOS 5.5.1 Technical figures 8.0-0-6 PC, Windows 98 7.5 PC, Windows 98 9.0-3-0 PC, Windows 2000 Professional Exceed V6.1/V5.3 PC, Windows 98 Communication and file transfer between PC and Unix workstation F-Secure V5.1 (Build 21) PC, Windows 2000 Professional INTENTIONALLY LEFT BLANK 4. INPUTS 4.1 DIRECT INPUT Input data and parameters needed for the development of the Seepage Calibration Model (SCM) are obtained from the Technical Data Management System (TDMS). As stated in Section 1, the SCM is used to estimate seepage-relevant parameters through model calibration. In general, calibration is a process of fixing certain parameters considered known, relatively certain, or insensitive, and adjusting others that are unknown, uncertain, or highly sensitive to minimize the misfit between measured data and model output. Input data were measured in or refer to the middle nonlithophysal and the lower lithophysal zones of the Topopah Spring welded unit (the repository units). Appropriate data for the middle nonlithophysal zone have been measured in Niches 2, 3, and 4, and appropriate data for the lower lithophysal zone have been measured in Niche 5 and in boreholes SYBT-ECRB-LA#1–#3 drilled into the ceiling of the ECRB Cross- Drift. Specific input data sets and the associated Data Tracking Numbers (DTNs) are listed in Table 4-1; specific input parameters are listed in Table 4-2; Technical Product Output (TPO) used as input to calculate local percolation fluxes is summarized in Table 4-3. These data and parameters are considered appropriate as input for the development of the SCM for the following reasons: 1. Profile alignments and borehole (BH) survey information (Table 4-1). These survey data are accurate and thus considered appropriate as a basis for defining niche geometry and identifying injection elements in the numerical mesh. 2. Air-permeability data (Table 4-1). These data are used as a basis for the geostatistical analysis and generation of spatially correlated permeability fields near the niches and the ECRB Cross-Drift. The data are location-specific and on the appropriate scale, and thus suitable for representing the local rock properties and the structure of sub- drift-scale heterogeneities. 3. Liquid-release test data (Table 4-1). These data are used for calibration and validation of the SCM. Liquid-release test data are appropriate for the calibration of the SCM and the estimation of seepage-relevant parameters, because they reflect the salient processes and features affecting seepage. Moreover, they are taken on a representative scale comparable to that of a waste emplacement drift. 4. Calibrated drift-scale fracture properties for the middle nonlithophysal and lower lithophysal zone of the Topopah Spring welded unit. Because they are directly measured or derived from data collected at Yucca Mountain, these fracture parameters are considered appropriate to be used as reference input parameters. Only the parameters that are fixed during an inversion, and for which no location-specific data are available, are needed as input; this subset is summarized in Table 4-2. Because of their small sensitivity on predicted seepage rates (see Section 6.6.3.1), a minor change in any of these input parameters has a negligible impact on the estimated model parameters or the conclusions of this Model Report. 5. Coordinates of the Unsaturated Zone Flow and Transport Model (UZ Model) grid and calculated flow rates for extraction of background percolation flux (Table 4-3). In the absence of direct observations of percolation flux, the percolation fluxes calculated by the UZ Flow Model, which is based on site-specific data, are considered appropriate for their intended use in the SCM. Table 4-1. Input Data DTNa Data Description Niche Geometry MO0003GSC00096.000 [DIRS 152167] ESF Niche 2 (Niche 3650) profile alignment MO0002GSC00076.000 [DIRS 152623] ESF Niche 2 (Niche 3650) borehole as-built information MO0003GSC00103.000 [DIRS 152176] ESF Niche 3 (Niche 3107) profile alignment MO0002GSC00064.000 [DIRS 152625] ESF Niche 3 (Niche 3107) borehole as-built information MO0008GSC00273.000 [DIRS 152626] ESF Niche 4 (Niche 4788) profile alignment MO0107GSC01069.000b [DIRS 156941] ESF Niche 4 (Niche 4788) borehole as-built information MO0009GSC00332.000 [DIRS 155370] ECRB Niche 5 (Niche 1620) profile survey data MO0107GSC01061.000 [DIRS 155369] ECRB Niche 5 (Niche 1620) slot survey data MO0312GSC03176.000 [DIRS 169532] ECRB Niche 5 (Niche 1620) survey data for collars, bottoms, and intervals LB0301N5CEILNG.001 [DIRS 161733] ECRB Niche 5 (Niche 1620) detailed niche ceiling roughness data Air-Permeability Data LB0011AIRKTEST.001 [DIRS 153155] Air permeability data from ESF Niche 2 (Niche 3650) LB990601233124.001 [DIRS 105888] Air permeability data from ESF Niche 3 (Niche 3107) and Niche 4 (Niche 4788) LB0110AKN5POST.001 [DIRS 156904] Air permeability data from ECRB Niche 5 (Niche 1620) LB00090012213U.001 [DIRS 153141] Air permeability data from ECRB borehole SYBT-ECRB-LA#2 Liquid-Release Test Data LB0010NICH3LIQ.001 [DIRS 153144] Liquid-release test data from ESF Niche 3 (Niche 3107), March 1999 LB0010NICH4LIQ.001 [DIRS 153145] Liquid-release test data from ESF Niche 4 (Niche 4788), Nov. 1999 LB0207NICH5LIQ.001 [DIRS 160408] Liquid-release test data from ECRB Niche 5 (Niche 1620), June 2000 LB0209NICH5LIQ.001 [DIRS 160796] Liquid-release test data from ECRB Niche 5 (Niche 1620), June 2002 LB0211NICH5LIQ.001 [DIRS 160792] Liquid-release test data from ECRB Niche 5 (Niche 1620), August 2002 LB0110ECRBLIQR.002 [DIRS 156879] Liquid-release test data from ECRB borehole SYBT-ECRB-LA#1, Feb. 2001 LB00090012213U.002 [DIRS 153154] Liquid-release test data from ECRB borehole SYBT-ECRB-LA#2, May 2000 LB0110SYST0015.001 [DIRS 160409] Liquid-release test data from ECRB borehole SYBT-ECRB-LA#2, Oct. 2000 LB0203ECRBLIQR.001 [DIRS 158462] Liquid-release test data from ECRB borehole SYBT-ECRB-LA#3, May 2001 a Traceability to the specific information extracted from these DTNs is given in the appendices and cited Scientific Notebooks. b This DTN superseded MO0008GSC00310.000 [DIRS 152627], which was the source for borehole coordinates available at the time of model development for Niche 4. Borehole coordinates in both DTNs are identical, i.e., there is no impact on the models, analyses, and conclusions presented in this Model Report. DTN=Data Tracking Number Table 4-2. Hydrogeologic Input Parameters DTN Parameter Value Units Middle Nonlithophysal Zone of Topopah Spring Welded Unit (Fracture Parameter for tsw34) LB997141233129.001a [DIRS 104055] van Genuchten parameter, m 0.608 [dimensionless] LB997141233129.001a [DIRS 104055] Residual liquid saturation, Slr 0.01 [dimensionless] LB997141233129.001a [DIRS 104055] Satiated saturation, Sls 1.00 [dimensionless] Lower Lithophysal Zone of Topopah Spring Welded Unit (Fracture Parameters for tsw35) LB0205REVUZPRP.001 [DIRS 159525] Porosity 0.96 [%] LB997141233129.001a [DIRS 104055] van Genuchten parameter, m 0.611 [dimensionless] LB997141233129.001a [DIRS 104055] Residual liquid saturation, Slr 0.01 [dimensionless] LB997141233129.001a [DIRS 104055] Satiated saturation, Sls 1.00 [dimensionless] a The superceded fracture parameters of DTN: LB997141233129.001 [DIRS 104055] [which is a qualified product output from a previous revision of the Calibrated Properties Model (CRWMS M&O 2000 [DIRS 144426])] are suitable for their intended use within this Model Report. The superceded and superceding values are identical with the exception of the van Genuchten parameter m; the superceding value for both units is 0.633. The difference between the superceded and superceding values are inconsequential for the estimation of drift seepage, because (1) the sensitivity of seepage to the m parameter (or the related n parameter) is very limited (as discussed in Section 6.6.3.1), and (2) consistent values are used in the calibration and prediction models. Moreover, the superceded values, which originated from a reliable source, have been used in previous analyses of flow, transport, and seepage for the same units; the superceded value is considered pertinent to the property of interest. The values were superceded because (1) the numerical grid was modified and (2) a new inversion methodology was employed. DTN=Data Tracking Number Table 4-3. Mesh Coordinates and Flow Field Used to Calculate Local Percolation Flux DTN TPO Description LB990701233129.001a [DIRS 106785] 3-D UZ model grid, including coordinates LB990801233129.003 a [DIRS 122757] Calculated percolation flux, flow field #3 a The calculated percolation fluxes from the superceded DTN: LB990801233129.003 [DIRS 122757] (which are based on the related numerical grid contained in the superceded DTN: LB990701233129.001 [DIRS 106785]) are considered suitable for the intended use in this Model Report, because (1) the estimated parameters, output, and conclusions presented in this Model Report are insensitive to the specified background percolation flux, (2) the superceded data originated from a reliable source, and (3) the superceded data have been used in previous analyses of flow, transport, and seepage in the unsaturated zone, i.e., they are pertinent to the properties of interest. The UZ Flow model providing percolation fluxes was revised (1) to accommodate a new repository design (requiring a new numerical grid), (2) to incorporate revised property sets, and (3) to employ a finer vertical discretization of the PTn hydrogeologic unit. The flow fields calculated with the revised UZ Flow model (see DTN: LB03023DSSCP9I.001 [DIRS 163044]) yield local percolation fluxes at Niches 3650, 3107, 4788, and 1620 of 4.0, 4.7, 2.4, and 6.2 mm/year, respectively. Given the low sensitivity of the estimated parameters to the background percolation flux (see discussion in Section 6.6.2.3), the differences between these and the superceded values are inconsequential. 3-D=three-dimensional: DTN=Data Tracking Number; UZ=Unsaturated Zone; PTn=Paintbrush nonwelded tuff; TPO=Technical Product Output Equations are discussed in the context of model development in Section 6 with appropriate citations to their sources. The collection of the input data used for the development and calibration of the SCM is described in detail in the report In Situ Field Testing of Processes (BSC 2004 [DIRS 170004], Sections 6.2 and 6.11) and is summarized in Section 6.5. The analysis of the seepage-rate data is described in Section 6.6.3. Uncertainties in the input data and parameters are addressed throughout Section 6 and are summarized in Section 8.2. 4.2 CRITERIA The licensing criteria for postclosure performance assessment are stated in 10 CFR 63 [DIRS 156605]. The requirements to be satisfied by TSPA are identified in the Yucca Mountain Project Requirements Document (Canori and Leitner 2003 [DIRS 166275]). The acceptance criteria that will be used by the U.S. Nuclear Regulatory Commission (NRC) to determine whether the technical requirements have been met are identified in Yucca Mountain Review Plan, Final Report (YMRP; NRC 2003 [DIRS 163274]). The pertinent requirements and criteria for this Model Report are summarized in Table 4-4. Section 8.5 provides cross-references to demonstrate how the acceptance criteria are addressed. Table 4-4. Project Requirements and Yucca Mountain Review Plan Acceptance Criteria Applicable to this Model Report Requirement Number Requirement Title 10 CFR 63 Link YMRP Acceptance Criteria PRD-002/T-015 (Canori and Leitner 2003 [DIRS 166275]) Requirements for Performance Assessment 10 CFR 63.114 (a-c) [DIRS 156605] Criteria 1 to 4 for Quantity and Chemistry of Water Contacting Waste Packages and Waste Forms (NRC 2003 [DIRS 163274], Section 2.2.1.3.3.3). Criteria 1 to 4 for Flow Path in the Unsaturated Zone (NRC 2003 [DIRS 163274], Section 2.2.1.3.6.3). YMRP=Yucca Mountain Review Plan Where a subcriterion includes several components, only some of those components may be addressed. How these components are addressed is summarized in Section 8.5. The acceptance criteria identified in Section 2.2.1.3.3.3 of the YMRP (NRC 2003 [DIRS 163274]) are given below. Section 2.2.1.3.3.3, Quantity and Chemistry of Water Contacting Waste Packages and Waste Forms Acceptance Criterion 1, System Description and Model Integration are Adequate: Subcriterion (2): The abstraction of the quantity and chemistry of water contacting engineered barriers and waste forms uses assumptions, technical bases, data, and models, that are appropriate and consistent with other related U.S. Department of Energy abstractions. For example, the assumptions used for the quantity and chemistry of water contacting engineered barriers and waste forms are consistent with the abstractions of “Degradation of Engineered Barriers” (Section 2.2.1.3.1); “Mechanical Disruption of Engineered Barriers (Section 2.2.1.3.2); “Radionuclide Release Rates and Solubility Limits” (Section 2.2.1.3.4); “Climate and Infiltration” (Section 2.2.1.3.5); and “Flow Paths in the Unsaturated Zone” (Section 2.2.1.3.6). The descriptions and technical bases provide transparent and traceable support for the abstraction of quantity and chemistry of water contacting engineered barriers and waste forms. Subcriterion (8): Adequate technical bases are provided, including activities such as independent modeling, laboratory or field data, or sensitivity studies, for inclusion of any thermal-hydrologic-mechanical-chemical couplings and features, events, and processes Acceptance Criterion 2, Data are Sufficient for Model Justification: Subcriterion (1): Geological, hydrological, and geochemical values used in the license application are adequately justified. Adequate description of how the data were used, interpreted, and appropriately synthesized into the parameters is provided. Subcriterion (2): Sufficient data were collected on the characteristics of the natural system and engineered materials to establish initial and boundary conditions for conceptual models of thermal-hydrologic-mechanical-chemical coupled processes, that affect seepage and flow and the engineered barrier chemical environment. Acceptance Criterion 3, Data Uncertainty is Characterized and Propagated Through the Model Abstraction: Subcriterion (1): Models use parameter values, assumed ranges, probability distributions, and bounding assumptions that are technically defensible, reasonably account for uncertainties and variabilities, and do not result in an under-representation of the risk estimate. Subcriterion (2): Parameter values, assumed ranges, probability distributions, and bounding assumptions used in the total system performance assessment calculations of quantity and chemistry of water contacting engineered barriers and waste forms are technically defensible and reasonable, based on data from the Yucca Mountain region (e.g., results from large block and drift-scale heater and niche tests), and a combination of techniques that may include laboratory experiments, field measurements, natural analog research, and process-level modeling studies. Subcriterion (4): Adequate representation of uncertainties in the characteristics of the natural system and engineered materials is provided in parameter development for conceptual models, process-level models, and alternative conceptual models. The U.S. Department of Energy may constrain these uncertainties using sensitivity analyses or conservative limits. For example, the U.S. Department of Energy demonstrates how parameters used to describe flow through the engineered barrier system bound the effects of backfill and excavation-induced changes. Acceptance Criterion 4, Model Uncertainty is Characterized and Propagated Through the Model Abstraction: Subcriterion (1): Alternative modeling approaches of features, events, and processes are considered and are consistent with available data and current scientific understanding, and the results and limitations are appropriately considered in the abstraction. Subcriterion (2): Alternative modeling approaches are considered and the selected modeling approach is consistent with available data and current scientific understanding. A description that includes a discussion of alternative modeling approaches not considered in the final analysis and the limitations and uncertainties of the chosen model is provided. Subcriterion (3): Consideration of conceptual model uncertainty is consistent with available site characterization data, laboratory experiments, field measurements, natural analog information and process-level modeling studies; and the treatment of conceptual model uncertainty does not result in an under-representation of the risk estimate. The acceptance criteria identified in Section 2.2.1.3.6.3 of the YMRP (NRC 2003 [DIRS 163274]) are given below. Section 2.2.1.3.6.3, Flow Paths in the Unsaturated Zone Acceptance Criterion 1, System Description and Model Integration are Adequate: Subcriterion (1): The total system performance assessment adequately incorporates important design features, physical phenomena, and couplings, and uses consistent and appropriate assumptions throughout the flow paths in the unsaturated zone abstraction process. Couplings include thermal-hydrologic-mechanical- chemical effects, as appropriate. Subcriterion (2): The aspects of geology, hydrology, geochemistry, physical phenomena, and couplings that may affect flow paths in the unsaturated zone are adequately considered. Conditions and assumptions in the abstraction of flow paths in the unsaturated zone are readily identified and consistent with the body of data presented in the description. Subcriterion (6): Adequate spatial and temporal variability of model parameters and boundary conditions are employed in process-level models to estimate flow paths in the unsaturated zone, percolation flux, and seepage flux. Subcriterion (7): Average parameter estimates used in process-level models are representative of the temporal and spatial discretizations considered in the model. Acceptance Criterion 2, Data are Sufficient for Model Justification: Subcriterion (1): Hydrological and thermal-hydrological-mechanical-chemical values used in the license application are adequately justified. Adequate descriptions of how the data were used, interpreted, and appropriately synthesized into the parameters are provided. Subcriterion (5): Sensitivity or uncertainty analyses are performed to assess data sufficiency, and verify the possible need for additional data. Subcriterion (6): Accepted and well-documented procedures are used to construct and calibrate the numerical models. Subcriterion (7): Reasonably complete process-level conceptual and mathematical models are used in the analyses. In particular: (i) mathematical models are provided that are consistent with conceptual models and site characteristics; and (ii) the robustness of results from different mathematical models is compared. Acceptance Criterion 3, Data Uncertainty is Characterized and Propagated Through the Model Abstraction: Subcriterion (1): Models use parameter values, assumed ranges, probability distributions, and bounding assumptions that are technically defensible, reasonably account for uncertainties and variables, and do not result in an under-representation of the risk estimate. Subcriterion (4): The initial conditions, boundary conditions, and computational domain used in sensitivity analyses and/or similar analyses are consistent with available data. Parameter values are consistent with the initial and boundary conditions and the assumptions of the conceptual models for the Yucca Mountain site. Subcriterion (6): Uncertainties in the characteristics of the natural system and engineered materials are considered. Acceptance Criterion 4, Model Uncertainty is Characterized and Propagated Through the Model Abstraction: Subcriterion (1): Alternative modeling approaches of features, events, and processes, consistent with available data and current scientific understanding, are investigated. The results and limitations are appropriately considered in the abstraction. 4.3 CODES, STANDARDS, AND REGULATIONS No specific, formally established standards have been identified as applying to this modeling activity. INTENTIONALLY LEFT BLANK 5. ASSUMPTIONS This section contains a list of assumptions used for the development of the Seepage Calibration Model (SCM). Each statement of an assumption is immediately followed by the rationale for why the assumption is considered valid or reasonable. Assumptions in immediately preceding upstream documentations have no significant impact on the results of the present model or they are discussed in the following subsections. 5.1 CONTINUUM APPROACH Assumption: The continuum approach is assumed to be a valid concept to calculate percolation flux and drift seepage at Yucca Mountain. Rationale: The continuum approach can be considered appropriate (1) if it appropriately represents the key features and processes determining seepage into large underground openings, and (2) if it is capable of reproducing and predicting seepage rates into a drift in a fractured formation. As discussed in detail in Section 6.3.2, diversion of water around an underground opening on account of the capillary barrier effect predominantly occurs within fracture planes that are oriented approximately perpendicular to the drift axis. Flow within a fracture plane (or a collection of fracture planes) can be described by a continuum model with a heterogeneous permeability field. As demonstrated in Sections 6.6.3 and 7.2.2.1, a continuum model is capable of reproducing and predicting seepage rates into a drift section, i.e., on the scale of interest. The continuum approach is therefore considered appropriate for seepage studies if applied within the framework described in this Model Report. Inverse modeling should be used for the estimation of process- specific, model-related, and scale-dependent parameters, and the same or similar conceptual model should be used for the subsequent seepage predictions, specifically the SMPA. No further confirmation is required for this assumption, which is used throughout Sections 6 and 7. 5.2 UNSATURATED FLOW Assumption: Water flow under unsaturated conditions is assumed to be governed by Richards’ equation [“Capillary Conduction of Liquids Through Porous Mediums” (Richards 1931 [DIRS 104252], pp. 318–333)]. Rationale: This assumption is justified because (1) gravitational force is ubiquitous, and (2) rough-walled or partially filled fractures exert varying degrees of capillary pressure at different saturation levels. The constant of proportionality—relative permeability—is saturation- dependent because (1) porous-medium continuum laws also apply to water flow through fractures filled with porous material, and (2) in the absence of fracture fillings, the thickness of the water film and connectivity of liquid islands on the fracture surface are saturation dependent [“Water Film Flow Along Fracture Surfaces of Porous Rock.” (Tokunaga and Wan 1997 [DIRS 139195], pp. 1287–1295). Richards’ equation follows from (1) the continuity equation and (2) the Buckingham-Darcy equation [Dynamics of Fluids in Porous Media (Bear, 1972 [DIRS 156269], pp. 487–502)]. Richards’ equation states that isothermal flow of water in a porous medium or rough-walled fracture occurs under the combined effect of gravitational and capillary forces, that flow resistance is a function of saturation, and that, for the purposes of this representation, movement of the nonwetting air phase can be neglected. This general concept, which is further discussed in Section 6.6.1.1 and used throughout Sections 6 and 7, is reasonable for unsaturated water flow through both porous matrix as well as partially filled or rough-walled fractures and does not require further confirmation. 5.3 CHARACTERISTIC CURVES Assumption: Relative permeability and capillary pressure are assumed to be described as continuous functions of effective liquid saturation, following the expressions given by the van Genuchten-Mualem model [“A Closed-Form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils.” (van Genuchten 1980 [DIRS 100610], pp. 892–893)]. Rationale: The van Genuchten-Mualem model is the standard model used in the suite of UZ models; it was chosen in this work to ensure consistency. Furthermore, the applicability of relative permeability and capillary pressure functions is consistent with the continuum assumption (see Sections 5.1 and 6.6.1.1) and is appropriate to represent fractures that are rough- walled and/or partially filled with porous material. The calibration process and the consistent conceptualization in the downstream models (specifically the SMPA) make this assumption a valid approach. No further confirmation is required for this assumption, which is used throughout Sections 6 and 7. 5.4 EFFECTIVE CAPILLARY-STRENGTH PARAMETER Assumption: The effective, seepage-relevant capillary-strength parameter to be estimated for each test location is assumed to be spatially uniform on the drift scale and thus not correlated to the small-scale heterogeneous permeability field. Rationale: The capillary-strength parameter to be estimated by calibration of the model against seepage-rate data is considered an effective parameter that includes a number of seepage-relevant features and processes, such as (1) the continuum capillarity of a network of rough-walled fractures, (2) capillary rise within finite fracture segments intersected by the underground opening, (3) small-scale drift-wall roughness (including effects of lithophysal cavities; see Section 5.7), and (4) capillary adsorption of water along drift wall leading to film flow. The capillary strength of the fracture system is correlated to the fracture aperture distribution. Similarly, permeability may be correlated to aperture, suggesting that capillarity and permeability are (negatively) correlated. However, given that these parameters describe continuum properties of a fracture network (rather than those of a single fracture), it should be noted that an increase in permeability might be associated with an increase in fracture density (rather than an increase in aperture). An increase in fracture density does not affect capillarity. Consequently, capillarity and permeability are not necessarily correlated. Items (2) through (4) are features and processes related to capillarity, and are thus well represented by a capillary-strength parameter; however, they are not related to permeability. Finally, since capillary strength is an effective parameter estimated by inverse modeling for a given conceptual model, its value is appropriate for use in a prediction model that has the same model structure, i.e., that uses the same assumption regarding the uniformity of this parameters. Given that (1) capillarity and permeability are not necessarily correlated, (2) seepage-relevant features and processes not related to permeability are represented by the capillary-strength parameter, and (3) the effective parameter is estimated and used within a suite of conceptually consistent models, it is appropriate to consider the capillary-strength parameter uniform on the drift scale and not correlated to the small-scale heterogeneous permeability field. This assumption, which is used throughout Sections 6 and 7, does not require further confirmation. 5.5 EVAPORATION IN CLOSED-OFF NICHES Assumption: The effect of evaporation on the seepage rates observed in closed-off niches in the middle nonlithophysal zone is assumed to be insignificant, i.e., water removal from the formation, at the drift surface, and from the capture system by evaporation and vapor diffusion is assumed to be small. Rationale: Under isothermal conditions, potential evaporation at the wall or in the capture system of a closed-off and humidified niche is small compared to the amount of water being released. Seepage experiments in the middle nonlithophysal zone of the Topopah Spring welded unit were conducted in niches that were closed off by a bulkhead, which leads to comparatively high relative humidity and low air circulation. Moreover, a humidifier was used in some of the experiments to ensure high relative humidity. For these conditions, Ho [“Evaporation of Pendant Water Droplets in Fractures.” (1997 [DIRS 141521])] and Or and Ghezzehei [“Dripping into Subterranean Cavities from Unsaturated Fractures under Evaporative Conditions.” (2000 [DIRS 144773])] provide a detailed description of evaporation mechanisms on the scale of individual water droplets within fractures or emerging from fractured formations. The evapo-infiltration threshold calculated by Ho (1997 [DIRS 141521], p. 2670) is significantly lower than the applied injection rates, suggesting a very minor influence of evaporation on measured seepage rates in experiments conducted in the niches. Evaporation effects were included in the modeling of liquid-release tests performed in Niche 5, which exhibited relative humidity conditions slightly lower than those expected to have prevailed in Niches 2, 3, and 4. The impact of evaporation on seepage rates and thus on the estimation of seepage-relevant parameters is minor (as demonstrated in Section 6.7), confirming that neglecting evaporation effects in closed-off niches is appropriate. No further confirmation is required for this assumption, which is used in Sections 6.6.3 and 7.2.2.1. Note that evaporation effects in the open, ventilated ECRB Cross-Drift are considered significant and are taken into account in the model based on relative humidity and evaporation rate measurements (see Sections 5.6, 6.5.4, and 6.6.1.4). 5.6 EVAPORATION IN OPEN DRIFT Assumption: Evaporation from the drift surface is assumed to be governed by one-dimensional vapor diffusion across an evaporative boundary layer, the thickness of which can be estimated from measurements of relative humidity and evaporation rate from a free water surface. Rationale: As water injected during a liquid-release test reaches the opening, it spreads along the surface on account of capillarity within the rough surface. As a result, water potentially seeping into the opening may not only form droplets or lines along fracture traces with a small surface area, but may spread across the drift surface over a relatively large area. This phenomenon is qualitatively confirmed by the geometry of the wet spot observed at the niche ceiling during seepage experiments [(BSC 2004 [DIRS 170004], Section 6.2.1.3.4, Figure 6.2.1-7); [“Seepage into an Underground Opening Constructed in Unsaturated Fractured Rock under Evaporative Conditions.” (Trautz and Wang 2002 [DIRS 160335], Figures 7 and 9)]. The geometry of the wet spot does not have a clear correlation with the visible fractures traces. Even though water first appears along fracture traces (Trautz and Wang (2002 [DIRS 160335], Figure 10), the wet spot grows in an areal fashion. The short arrival time and the average speed at which the leading edge of the plume moves across the ceiling makes it obvious that the water is not transmitted through the matrix, but spreads along the ceiling as a surface film, possibly supported by flow through microfractures. Evaporation from such wet areas is similar to evaporation from a free water surface, where the evaporation rate is governed by one- dimensional vapor diffusion across a relatively thin boundary layer of linearly decreasing vapor concentration. A detailed description of the corresponding conceptual and mathematical model and the estimation of the evaporation boundary-layer thickness is given in Sections 6.6.1.3 and 6.6.1.4. No further confirmation is required for this assumption, which is used in Sections 6.6.3 and 7.2.2.1. 5.7 LITHOPHYSAL CAVITIES Assumption: The impact of lithophysal cavities on seepage is assumed to be appropriately captured in the estimation of an effective capillary-strength parameter. Rationale: The impact of lithophysal cavities on flow and seepage is twofold: (1) lithophysal cavities are essentially obstacles to water flow because they act as capillary barriers, focusing the water that flows around them; (2) lithophysal cavities intersected by the drift lead to a rough drift wall, potentially creating seepage points at local topographic lows. Both effects tend to promote seepage. The assumption states that the effect of lithophysal cavities on seepage can be captured through the estimation of an effective capillary-strength parameter, making the explicit inclusion of lithophysal cavities into the process model unnecessary. This approach is considered appropriate for the following reasons: (1) omitting lithophysal cavities in the process model used for inverse modeling yields lower estimates of the capillary-strength parameter and is thus conservative; (2) consistency between the calibration model (the SCM) and the prediction model (the SMPA) removes the impact of a potential estimation bias; (3) the approach followed allows for the development of a single SMPA conceptual model for both the middle nonlithophysal and lower lithophysal zones, yielding a single look-up table for TSPA to sample from; and (4) explicit modeling of lithophysal cavities is not warranted because of insufficient information regarding their location, shape, and frequency. No further confirmation is required for this assumption, which is used in Sections 6.6.3 and 7.2.2.1. INTENTIONALLY LEFT BLANK 6. MODEL DISCUSSION 6.1 MODELING OBJECTIVES AND DEFINITIONS 6.1.1 Objectives The following sections describe the development, calibration, and validation of the Seepage Calibration Model (SCM). The purpose of the SCM is to provide a methodological and conceptual basis for the subsequent development of the Seepage Model for Performance Assessment (SMPA). Furthermore, seepage-relevant parameters are derived as input to the abstraction for drift seepage. The seepage models are not expected to accurately predict individual seepage events or the precise spatial distribution along the drift. Instead, the seepage models are intended to provide estimates of the seepage flux averaged over a 5 m drift segment (the approximate length of a waste package) as a function of the percolation flux on the drift scale. The seepage experiments and modeling approach are designed to address seepage on this specific scale. A list of the data corroborating and supporting the SCM (including the corresponding source DTNs) is provided in Table 6-5 below. 6.1.2 Definitions Seepage is defined as flow of liquid water into an underground opening such as a niche, the ECRB Cross-Drift, or a waste emplacement drift; the water originates from the rock mass and forms drops that subsequently detach from the opening surface. According to this definition, seepage does not include advective or diffusive vapor flow into the opening or condensation of water vapor on surfaces, which may lead to drop formation and drop detachment. Some of the water entering an underground opening may also evaporate or flow along the wall, thus not contributing to seepage in the narrow sense defined here. Note, however, that evaporation, condensation, and film flow along the surface of the opening affect the moisture conditions in the waste emplacement drift and may thus impact repository performance. Seepage rate is the amount of water seeping into the opening per unit of time. Seepage flux is defined as the seepage rate per unit area of the projected drift outline. Seepage percentage is defined as the ratio of seepage flux divided by percolation flux. As outlined in Section 6.1.1, a five-meter long drift section (the approximate length of a waste package) is used as the reference scale for calculating percolation and seepage fluxes. In the context of liquid-release tests, seepage percentage is the ratio of the rate or amount of water that seeped into the niche divided by the rate or amount of water released. Seepage threshold is defined here as the critical percolation flux below which no seepage occurs, i.e., all percolating water is diverted around the opening, evaporates, or flows along the drift surface as a thin water film. Note that Philip et al. [“Unsaturated Seepage and Subterranean Holes: Conspectus, and Exclusion Problem for Circular Cylindrical Cavities.” (1989 [DIRS 105743])] did not consider evaporation and film-flow effects when defining the critical seepage conditions. Seepage fraction is defined as the fraction of waste packages affected by seepage. This is equivalent to the fraction of 5 m drift sections that exhibit a nonzero seepage percentage. Capillary barrier is a technical term used to describe the fact that water is diverted around an underground opening, preventing seepage or reducing the seepage flux below the incident percolation flux. This technical definition of a barrier is different from regulatory definition of 10 CFR 63.2 [DIRS 156605]. However, the usage of the term in the context of the capillary barriers discussed in this Model Report is unambiguous. 6.1.3 Scientific Notebooks The scientific notebooks (SN) listed in Table 6-1 provide details potentially needed to reproduce the modeling work discussed in this Model Report. Table 6-1. Scientific Notebooks LBNL Scientific Notebook ID M&O Scientific Notebook ID Relevant Pages Citation YMP-LBNL-SAF-1 SN-LBNL-SCI-087-V1 1–4, 100–102, 139 Finsterle 1999 [DIRS 153448] YMP-LBNL-SAF-2 SN-LBNL-SCI-171-V1 1–2, 34–42, 47–95 Finsterle 2002 [DIRS 161043] YMP-LBNL-SAF-3 SN-LBNL-SCI-228-V1 1–26, 31–37 Wang 2003 [DIRS 161456] YMP-LBNL-SAF-TG-1 SN-LBNL-SCI-223-V1 9–44 Wang 2003 [DIRS 161456] YMP-LBNL-RCT-DSM-1 SN-LBNL-SCI-157-V1 1–37 Trautz 2001 [DIRS 161044] YMP-LBNL-RCT-2 SN-LBNL-SCI-156-V1 35–45 Trautz 2001 [DIRS 156903] YMP-LBNL-JSW-6C SN-LBNL-SCI-122-V1 108–123 Wang 1999 [DIRS 153449] YMP-LBNL-DSM-CFA-1 SN-LBNL-SCI-180-V1 4–6, 8–10, 13, 15–58 Ahlers 2002 [DIRS 161045] YMP-LBNL-YSW-JH-2 SN-LBNL-SCI-143-V1 124 Hinds 2001 [DIRS 155955] YMP-LBNL-RCT-RH-1 SN-LBNL-SCI-175-V1 27–29 Hedegaard 2002 [DIRS 161046] LBNL=Lawrence Berkeley National Laboratory 6.2 FEATURES, EVENTS, AND PROCESSES Table 6-2 contains a list of FEPs taken from the LA FEP List (DTN: MO0407SEPFEPLA.000 [DIRS 170760]). The selected FEPs are those taken from the LA FEP List that are associated with the subject matter of this report (BSC 2004 [DIRS 169654], Table 2.1.5-1). The results of this model are part of the basis for the treatment of FEPs as discussed in the Total System Performance Assessment-License Application Methods and Approach (BSC 2003 [DIRS 166296], Section 3.2.2). The cross-reference for each FEP to the relevant sections of this report is also given in Table 6-2. Table 6-2. FEPs Addressed in this Model Report FEP No. FEP Name Relevant Sections of this AMR 1.2.02.01.0A Fractures Air-permeability and seepage testing as well as the heterogeneous fracture continuum model characterize and account for flow through and seepage from fractures (see Sections 6.3.2, 6.3.3.2, and 6.5) 2.1.08.02.0A Enhanced influx at the repository The impact of an underground opening on the unsaturated flow field (including dry-out from evaporation, capillary-barrier effect, and flow diversion around the drift) is captured in the seepage process model by solving the equations governing unsaturated flow in fractured porous media and by specifying appropriate boundary conditions at the drift wall. It leads to reduced (not enhanced) influx (see Sections 6.3, 6.3.2, 6.3.3, 6.6, and 6.8). 2.2.01.01.0A Mechanical effects of excavation/ construc- tion in the near field Excavation effects are taken into account through the use of post- excavation air-permeability data and the estimation of a capillary-strength parameter determined from seepage data that reflect seepage from an excavation-disturbed zone around a large opening (see Sections 6.3.3.2, 6.3.4, 6.5.2, 6.6.3.1, and 6.6.3.3). 2.2.03.02.0A Rock properties of host rock and other units Location-specific rock properties are (1) taken from supporting reports (see Table 4-2), (2) determined from local air-permeability data (including measures of heterogeneity and spatial correlation), and (3) determined through inverse modeling. Variability is accounted for on various scales (see Sections 4.1, 6.5.2, and 8.2). 2.2.07.02.0A Unsaturated ground- water flow in the geosphere Unsaturated flow processes are accounted for in the conceptual and mathematical model (see Sections 6.3.2 and 6.6.1.1). 2.2.07.04.0A Focusing of unsatu- rated flow (fingers, weeps) Explicitly modeled heterogeneity induces flow focusing. Impact of small- scale flow focusing effects on seepage is included in effective parameter (see Sections 6.3, 6.3.3.1, 6.6.2.1, and 6.6.3.3). 2.2.07.08.0A Fracture flow in the UZ Liquid flow through unsaturated fractures is simulated using site-specific fracture properties; explicit inclusion of heterogeneity leads to flow channeling (see Sections 6.3.2, 6.3.3.2, and 6.6.2.1) 2.2.07.09.0A Matrix imbibition in the UZ Matrix imbibition is considered small under near-steady seepage conditions and is therefore neglected (see Section 6.3.3.2). 2.2.07.18.0A Film flow into the repository If water originating from film flow seeps into the opening during a liquid- release test, it is reflected in the corresponding seepage data point used for model calibration, i.e., film flow is automatically accounted for in the estimated seepage-related parameter and thus in the prediction of seepage into waste emplacement drifts (see Sections 6.1.2, 6.3.2, 6.3.3.2, 6.3.4, and 6.6.3.1). 2.2.07.20.0A Flow diversion around repository drifts The impact of an underground opening on the unsaturated flow field (including dry-out from evaporation, capillary-barrier effect, and flow diversion around the drift) is captured in the seepage process model by solving the equations governing unsaturated flow in fractured porous media and by specifying appropriate boundary conditions at the drift wall. Drift shadow is simulated as a result of seepage exclusion (see Sections 6.3, 6.3.2, 6.3.3, 6.6, and 6.8). FEP=features, events, and processes; UZ=Unsaturated Zone 6.3 BASE-CASE CONCEPTUAL MODEL 6.3.1 Seepage Phenomena and Processes To understand the seepage process and to identify the factors affecting seepage, a description is given of the fate of water percolating through the unsaturated zone of Yucca Mountain, eventually encountering the immediate vicinity of a waste emplacement drift. This description is based on and consistent with the related discussion found in the scientific literature (see, for example, Philip et al. (1989 [DIRS 105743]) and “Using the Continuum Approach to Model Unsaturated Flow in Fractured Rock.” (Finsterle 2000 [DIRS 151875]) and references therein). Water that penetrates the ground surface and reaches a depth that is unaffected by evapotranspiration starts to percolate downwards, driven by gravity and capillary forces. The detailed flow path is determined by the degree of fracturing, fracture geometry, orientation, and connectivity, as well as the hydrogeologic properties of the fractures and the matrix. Depending on these factors, the water phase in the unsaturated fracture network will either disperse or focus along the flow path. Tilted contacts between hydrogeologic units (especially between welded and nonwelded tuffs) may affect the overall flow pattern or lead to a change in the frequency and spacing of flow channels. However, the channeling process (i.e., the redistribution of water leading to local fluxes in portions of the fracture network that are higher and lower than the average flux) is likely to diminish with depth. As flow concentration continues to occur, the distance between the individual channels carrying focused flow increases, so the likelihood of two channels meeting and merging decreases with depth. Flow focusing and dispersion of flow paths also happens within a rough-walled fracture, where asperity contacts and locally larger fracture openings lead to small-scale redistribution of water within the fracture. A general discussion of channeling effects under unsaturated flow conditions can be found in “Solute Channeling in Unsaturated Heterogeneous Porous Media” (Birkholzer and Tsang 1997 [DIRS 119397]). Flow focusing is important for seepage, because seepage depends on the local rather than average percolation flux. As water approaches the potential waste emplacement drift (one to several meters above the drift ceiling), conditions change in several ways, all affecting the amount of water that will eventually seep into the opening. The water may first encounter a dry-out zone caused by drift ventilation. The dry-out zone may also develop as a result of increased temperatures, in which case it is referred to as a boiling zone. Under these thermal conditions, the dry-out zone may be surrounded by a two-phase zone in which heat-pipe effects determine water, vapor, and heat fluxes, and a condensation zone with increased saturation. (Note that ventilation and elevated temperatures are limited in time and thus do not affect long-term seepage.) In addition, formation properties around the openings are likely to be altered as a result of stress redistribution during drift excavation, which leads to local opening or partial closing of fractures and potentially the creation of new fractures. Thermal expansion of the rock matrix may also induce changes in apertures. Finally, the local chemical environment, which is altered by evaporation and thermal effects, may lead to dissolution and precipitation of minerals, again affecting porosity, permeability, and capillarity of the fracture system as well as fracture-matrix interaction. Such thermally and geochemically induced alterations were of no significance during the ambient liquid-release tests analyzed by the SCM. In general, however, all the conditions discussed above lead to a flow pattern in the vicinity of a waste emplacement drift different from that in the undisturbed formation under ambient conditions. Provided that liquid water penetrates the boiling or dry-out zone (for details, see Drift-Scale Coupled Processes (DST and TH Seepage) Models (BSC 2004 [DIRS 170338])), it reaches the immediate vicinity of the drift wall, where (at least under ambient conditions) a boundary layer of increased saturation is expected to develop as a result of the capillary barrier effect (Philip et al. 1989 [DIRS 105743]). The water is prevented from seeping into the drift because of capillary suction, which retains the wetting fluid in the pore space. If permeability and capillarity of the fracture network within this boundary layer are sufficiently high, all or a portion of the water is diverted around the drift under partially saturated conditions. Locally, however, the water potential in the formation may be higher than that in the drift, and water appears at the drift surface. At the drift surface, the water either evaporates, follows the inclined, rough wall in a water film, or forms a drop that grows and eventually detaches (Or and Ghezzehei 2000 [DIRS 144773]). Only this last mechanism is considered drift seepage according to the definition of Section 6.1.2. To summarize, the rate of water dripping into an opening in an unsaturated geologic formation is expected to be less than the downward percolation rate because (1) the cavity acts as a capillary barrier, (2) water may flow along the drift surface without dripping into the opening, and (3) water may evaporate. Even if the seepage threshold was exceeded and seepage occurred, the seepage flux would be lower than the percolation flux. Section 6.3.2 describes the rationale and justification for using a heterogeneous continuum model for the simulation of drift seepage. Section 6.3.3 discusses specific factors and properties affecting seepage during liquid-release tests and how they are incorporated into the conceptual model. 6.3.2 Continuum Approach The Seepage Calibration Model is conceptualized as a heterogeneous continuum model. The continuum approach can be considered appropriate for seepage studies if it is capable of predicting seepage rates for a drift in a fractured formation. Water flow through the Topopah Spring welded unit (TSw) and seepage into openings at Yucca Mountain occurs predominantly through the fracture network, suggesting that a discrete fracture network model is more appropriate than a fracture continuum model for the reproduction and prediction of drift seepage. However, it is important to recognize that flow diversion around the opening occurs primarily within the fracture plane (in-plane diversion). The need to engage multiple fractures arises only if the fracture is too short and the flow path within the fracture plane is interrupted. In this case, water is diverted into the next connected fracture. This fracture is again unlikely to be parallel to the drift axis, allowing the in-plane flow-diversion process to continue. The situation is schematically illustrated in Figure 6-1, which shows two fractures intersected by a drift. In Figure 6-1a, the two fractures are aligned with the drift axis (which is an implicit assumption made in two-dimensional fracture network models used to predict drift seepage). As an artifact of this specific and unrealistic fracture orientation, in-plane flow diversion is prevented, and the resulting impact of discreteness on seepage is exaggerated. Two-dimensional fracture network models (including those shown by Finsterle (2000 [DIRS 151875], Plate 1) and in “A Note on Unsaturated Flow in Two-Dimensional Fracture Networks” (Liu et al. 2002 [DIRS 160230], Figures 1–6)) represent extreme cases that are not representative of and appropriate for site-specific seepage modeling. (The advantages and disadvantages of the discrete fracture network model are further discussed in Section 6.4.1). In Figure 6-1b, the fractures are approximately perpendicular to the drift axis. Flow diversion occurs within the fracture plane, a process that is appropriately captured by a heterogeneous fracture continuum model even for a single fracture. In-plane flow occurring in multiple fractures can be readily combined and described by an effective fracture continuum. (a) (b) NOTE: (a) Drift intersected by network of fractures that are parallel to drift axis; note that a 2-D fracture network model (see, for example, Figure 6-3a) assumes that all fractures are parallel to the drift axis, preventing flow diversion within the fracture plane: (b) Drift intersected by randomly oriented fractures; note that a 2- D (and 3-D) fracture continuum model considers flow diversion occurring within multiple, randomly oriented fracture planes. Figure 6-1. Schematic Showing Two Fractures Intersecting a Drift In Geology of the ECRB Cross-Drift – Exploratory Studies Facility, Yucca Mountain Project, Yucca Mountain, Nevada, Mongano et al. (1999 [DIRS 149850], pp. 65–72, 76–79) documented in their detailed line survey report that fracture frequencies observed in the Topopah Spring upper lithophysal (Tptpul), middle nonlithophysal (Tptpmn), and lower lithophysal (Tptpll) zones range from 3.2 m-1 to 4.3 m-1, and that fractures are predominantly developed in two or three orientations resulting in well connected networks. Connectivity of fractures is further enhanced by the presence of numerous microcracks as observed at the site. Fracture network connectivity has been determined at a drift scale through air-injection tests, which indicate that fractures networks are well connected within the moderately to densely welded rocks selected to host the repository (see Section 6.5.2). Given the significance of in-plane flow diversion around the drift in combination with relatively high fracture density of variable orientation, a three-dimensional, heterogeneous fracture continuum model is an appropriate conceptualization. The continuum concept captures the relevant processes more realistically than, for example, a two-dimensional discrete fracture network model. In addition, the appropriateness of the continuum approach to simulate flow through fractured rock was studied by Jackson et al. [“Self-Consistency of a Heterogeneous Continuum Porous Medium Representation of a Fractured Medium” (2000 [DIRS 141523])] using synthetic and actual field data. They concluded that heterogeneous continuum representations of fractured media are self-consistent; i.e., appropriately, estimated effective continuum parameters are able to represent the underlying fracture-network characteristics. Finsterle (2000 [DIRS 151875]) demonstrated that seepage into underground openings excavated from a fractured formation could be simulated using a model based on the continuum concept, provided that the model is calibrated against seepage-relevant data (such as data from liquid-release tests). Synthetically generated data from a model that exhibits discrete flow and seepage behavior were used to calibrate a simplified fracture continuum model. The calibrated continuum model was used to predict average seepage rates into a sufficiently large section of an underground opening for low percolation fluxes, i.e., conditions significantly different from those encountered during calibration. The extrapolation from high-rate liquid-release tests to low-rate percolation fluxes is equivalent to the extrapolation from the calibration runs performed with the SCM to the predictive simulations that will be performed by the SMPA. As discussed by Finsterle (2000 [DIRS 151875]), the extrapolated seepage predictions performed with the continuum model were consistent with the synthetically generated data from the discrete-feature model under low percolation conditions. This demonstrates that (1) the calibrated continuum model and discrete-feature model yield consistent estimates of average seepage rates, and (2) that the continuum approach is appropriate for performing seepage predictions even if extrapolated to percolation fluxes that are significantly lower than those induced by liquid-release tests. The tests were performed at relatively high injection rates to generate seepage data useable for model calibration. Note that the discrete-feature model used in the study makes the extreme assumption that all fractures are oriented parallel to the drift axis, as discussed above and illustrated in Figure 6-1a. Even under these unfavorable conditions, the continuum approach proved to be appropriate. Note that the fracture density and hydraulic parameters used by Liu et al. (2002 [DIRS 160230]; see also Figure 6-3a below) result in very little flow diversion around the opening. This is a direct result of the unrealistic assumption that all fractures are parallel to the drift axis, which prevents in-plane flow diversion. In such a two-dimensional discrete fracture network model, flow diversion occurs only if the fracture density and/or the capillary-strength parameter are high. This was recognized by Liu et al. (2002 [DIRS 160230], p. 15-8), who concluded that fracture network models need to be three-dimensional for them to be able to realistically evaluate the capillary barrier effects in fractured formations. As discussed above, in-plane flow diversion in a three-dimensional fracture network can be appropriately represented by a heterogeneous continuum model. A calibrated continuum model is appropriate even in the extreme case where all fractures are perfectly parallel to the drift axis, as demonstrated by Finsterle (2000 [DIRS 151875]) and discussed in the previous paragraph. Note that the synthetic fracture network and hydraulic parameters used in the discrete model by Finsterle (2000 [DIRS 151875]; see also Figure 6-2a below) induced some flow diversion. This difference in flow diversion capability between the models by Finsterle (2000 [DIRS 151875]) and Liu et al. (2002 [DIRS 160230]) is caused by their respective parameter choices. This difference, however, does not affect the finding that the continuum approach captures the seepage-relevant processes more appropriately than two-dimensional discrete fracture network models. The advantages and disadvantages of the discrete fracture network model are further discussed in Section 6.4.1. The continuum approach is considered appropriate for seepage studies if applied within the framework described in this Model Report. Inverse modeling should be used for the estimation of process-specific, model-related, and scale-dependent parameters, and the same or similar conceptual model should be used for the subsequent seepage predictions, specifically the SMPA. Adopting the continuum approach, water flow under unsaturated conditions is governed by Richards’ equation (Richards 1931 [DIRS 104252]), which states that (1) isothermal flow of water in a porous medium or rough-walled fracture occurs under the combined effect of gravitational and capillary forces, (2) flow resistance is a function of saturation, and (3)—for the purposes of this representation—movement of the nonwetting air phase can be neglected. This general concept is reasonable, because gravitational force is ubiquitous, and rough-walled or partially filled fractures exert varying degrees of capillary pressure at different saturation levels. Relative permeability and capillary pressure are described as continuous functions of effective liquid saturation, following the expressions given by the van Genuchten-Mualem model (van Genuchten 1980 [DIRS 100610], pp. 892–893) as implemented in the iTOUGH2 code [User’s Manual (UM) for iTOUGH2 V5.0 (BSC 2002 [DIRS 161066], Section 4.3.2)]. The applicability of relative permeability and capillary pressure functions is appropriate also for fractures that are rough-walled and/or partially filled with porous material. The constant of proportionality—relative permeability—is saturation-dependent because porous-medium continuum laws also apply to water flow through fractures filled with porous material, and in the absence of fracture fillings, the thickness of the water film and connectivity of liquid islands on the fracture surface are saturation dependent (Tokunaga and Wan 1997 [DIRS 139195]). Capillary strength (represented by the parameter) and permeability are considered uncorrelated. The functional relationship describing the potential correlation between permeability and capillary strength is unknown. An increase in the effective (continuum) permeability of a fracture block may be attributed to either larger fracture apertures (which would reduce capillary strength) or to an increase in fracture density (which would not affect capillary strength). The capillary-strength parameter is taken to be constant for a given test bed, and will be estimated by inverse modeling. The van Genuchten-Mualem model is the standard model used in the suite of UZ models; it is appropriate for modeling of unsaturated flow and seepage (as discussed in Section 5.3) and was chosen in this work to ensure consistency. The mountain-scale models may use a modified version of the van Genuchten-Mualem functions to account for the fact that unsaturated flow is restricted to a limited number of (active) fractures and that flow within a fracture is likely to be channelized. Both effects lead to different effective saturations determining capillary pressure and relative permeability, and they reduce fracture-matrix interaction. This revised model was developed by Liu et al. [“An Active Fracture Model for Unsaturated Flow and Transport in Fractured Rocks” (1998 [DIRS 105729])] and is referred to as the Active Fracture Model (AFM). For drift-scale seepage models under ambient conditions, the standard van Genuchten-Mualem model is employed rather than the AFM, because (1) flow segregation into active and inactive portions of the fracture network is a large-scale effect not engaged during the short-distance liquid-release tests; (2) flow channeling within fractures is partially accounted for through explicit modeling of small-scale heterogeneity; (3) the correction of the fracture-matrix interface area (the main effect captured by the AFM) is insignificant for seepage because of insignificant matrix imbibition during the calibration period (see Section 6.3.3.2); and (4) the potential impact of AFM effects on seepage are automatically reflected in the observed seepage- rate data, which are used to estimate an effective capillary-strength parameter suitable for simulations with a conceptually consistent seepage-prediction model. This general model conceptualization is consistent with that of the UZ Model. The calibration process and the consistent conceptualization in the downstream models (specifically the SMPA) make this a valid and reasonable approach. 6.3.3 Factors and Properties Affecting Seepage During Liquid-Release Tests Seepage is a process that occurs at the interface between the natural and engineered systems. Consequently, seepage is not only affected by hydrogeologic factors (such as formation properties and flow conditions in the natural environment), but also by the engineered system itself. This second set of factors affecting the amount and distribution of seepage includes the design of the repository and waste emplacement drifts (location and geometry), the method of construction (excavation effects, drift surface roughness, ground support, backfill), and the conditions within the drifts (heat load and ventilation, which determine the relative humidity, evaporation potential, and the extent of the dry-out zone). The engineered barriers in the waste emplacement drift (specifically the drip shield and waste packages) will be exposed to seeping water only if (1) a flow channel exists that carries water through the (potentially dry) zone around the drift, (2) the local percolation flux in this flow channel is high enough to overcome the local seepage threshold, and (3) the water droplets forming at the drift wall do not evaporate or dissipate in a water film flowing along the surface. The following subsections describe in more detail the key factors affecting drift seepage and how they are included in the base-case conceptual model. Based on theoretical insights discussed in the scientific literature [see, e.g., Philip et al. (1989 [DIRS 105743]); “Modeling Studies and Analysis of Seepage into Drifts at Yucca Mountain” (Birkholzer et al. 1999 [DIRS 105170]); “Seepage into Drifts with Mechanical Degradation” (Li and Tsang 2003 [DIRS 163714])] as well as the sensitivity analyses presented in Section 6.6.3.1, the most important factors affecting seepage are the magnitude of the local percolation flux in relation to the formation’s permeability, the strength of the capillary forces in the fractures, the connectivity of the fracture network in the boundary layer, the local topography of the rough drift wall, and the thermodynamic conditions in the drift. 6.3.3.1 Percolation Flux General Description The magnitude of the percolation flux is a key factor determining seepage. Seepage is initiated if the local percolation fluxes in individual flow channels and their accumulation near the drift ceiling exceeds the diversion capacity of the capillary barrier (which is caused by the presence of the drift), the evaporation potential of the atmosphere in the drift, and the capacity of water films to carry water along the drift surface. Because the local—rather than average—percolation flux controls the onset of seepage, the distribution of flow channels on all scales becomes a critical aspect for drift seepage. Flow focusing could concentrate water onto a particular drift segment and lead to a flux that exceeds the seepage threshold. On the other hand, if flow is concentrated in one location, flow will be reduced in other areas (potentially below the prevalent seepage threshold), leading to overall less seepage. Therefore, the distribution of flow channels, their frequency, width, and hydrologic properties determine the seepage probability and seepage amounts. The spatial distribution of flow channels may change with the average percolation flux and potentially with time. The flux in a flow channel may be near steady state or episodic with a wide spectrum, ranging from high-frequency fluctuations triggered by flow instabilities, to intermediate variabilities in percolation fluxes in response to changing weather conditions, to long-term variations from climate changes. In summary, the local (rather than average) percolation flux reaching the drift is the most important factor determining whether seepage occurs, the seepage rate, and the spatial and temporal distribution of seepage events. Model Conceptualization The actual percolation flux and its distribution cannot be measured directly. Estimates of the average, steady-state percolation fluxes at the locations of the liquid-release tests are taken from the UZ Model (see Section 6.6.2.3) and applied at the top of the corresponding drift seepage models. Note that large-scale redistribution of infiltration and percolation fluxes is captured in the mountain-scale UZ Model; intermediate-scale flow concentration is accounted for in the TSPA calculations through the use of a probabilistic flow focusing factor (BSC 2004 [DIRS 169131], Section 6.6.4.2). Small-scale flow concentration is included in the SCM by explicitly modeling small-scale heterogeneities (see Section 6.6.2.1). The transient SCM simulations capture the time-dependent boundary conditions, saturation, and seepage-rate changes induced by the intermittent water release during seepage testing. Potential occurrence of small-scale, high-frequency episodic flow events is reflected in the seepage-rate data used for calibration. The cumulative effect of these episodic events on seepage is therefore appropriately captured in the estimation of an effective capillary-strength parameter. Low-frequency fluctuations in the background percolation flux on account of weather-condition or climate changes are insignificant because of the comparatively short duration of the liquid-release tests. In summary, the high-frequency episodic flow events are captured in the effective, seepage-relevant capillary-strength parameter, whereas the low-frequency transient events are accounted for in the UZ Model, which provides a time-dependent percolation flux as input to the seepage TSPA calculations. Additional issues related to the amount, variability, and uncertainty of percolation flux, lateral flow diversion, as well as large- and intermediate-scale flow concentration are also addressed by the UZ Model, seepage abstraction, and TSPA calculations. 6.3.3.2 Formation Properties General Description The key formation properties determining the effectiveness of the capillary barrier are (1) the capillary strength and (2) the tangential conductivity in the boundary layer near the drift wall. Geologic formations with strong capillarity and high tangential conductivity exhibit a high seepage threshold (i.e., low seepage), whereas a weak capillary barrier effect (i.e., high seepage) is expected if water retention is small or if the tangential permeability is insufficient to promote flow diversion. Porous formations with strong capillarity tend to have low permeability and vice versa, which is a correlation that reduces the probability of encountering parameter combinations conducive to extreme (low or high) seepage behavior, making seepage relatively uniform across different geologic units. However, this negative correlation between conductivity and capillary strength may not apply to a fractured system, specifically if considering the seepage process. A certain hydraulic conductivity may result from a network consisting of a few, large fractures or, alternatively, many small, well-connected fractures. The first network would exhibit weak capillarity, whereas the second network has strong capillarity, i.e., capillarity is not necessarily correlated to permeability. Moreover, if the predominant fracture orientation happens to be aligned with the drift axis (see Figure 6-1a), little or no tangential conductivity is available, flow diversion is reduced or prevented, and seepage is increased. Even if fractures are normal to the drift axis, they may be too small or poorly connected, i.e., they would not be able to facilitate a continuous flow path from the apex of the drift to its spring line. For flow diversion to occur, the fracture system must have sufficient connectivity and permeability to provide the necessary effective conductivity in tangential direction around the drift. In the repository units, matrix permeability is low, and the potential for imbibition of substantial amounts of water into the matrix is limited because of relatively low porosity and relatively high initial liquid saturation. In a fracture-matrix system, the transient effects from matrix imbibition are restricted to intermediate times, i.e., they are insignificant (1) for a short-term liquid-release test with insufficient time for matrix imbibition, and (2) for a long-term seepage experiment, when near-steady late-time data are no longer affected by matrix imbibition. Most liquid-release tests analyzed in this Model Report are sufficiently long to yield near-steady seepage rates that are insignificantly affected by potential matrix imbibition. Finally, potential seepage from the matrix during a liquid-release test is captured by the seepage-rate data used for calibration, and is thus reflected in the effective, seepage-relevant parameter. Heterogeneities in formation properties impact seepage as they promote flow concentration and increase the probability of locally breaching the capillary barrier. Model Conceptualization Seepage-related fracture properties on all relevant scales are not available and cannot be reliably derived from fracture-trace maps, considering that the mapped geometric characteristics and hydraulic properties are not related in a simple or unique way. However, as discussed in Sections 6.3.2 and 6.4.1 and demonstrated by Finsterle (2000 [DIRS 151875]), adequate average seepage prediction on the scale of a waste package does not require a discrete fracture network model. In this work, the capillarity and the conductivity are conceptualized as effective properties that are specifically determined for their intended use in a drift seepage model. The corresponding model parameters must represent the average hydraulic characteristics of individual fractures as well as the connectivity, density, geometry, and orientation of the fracture network as it relates to the geometry and orientation of the underground opening. Moreover, they must account for seepage processes that cannot be explicitly implemented in the conceptual model (such as film flow and small-scale roughness in the drift ceiling), and compensate for certain artifacts related to the finite discretization of the numerical model. Model calibration using data that reflect all relevant seepage processes is the approach relied upon to determine these effective parameters. The SCM is conceptualized as a heterogeneous fracture continuum model (see also Section 6.3.2). Given the specifics of the seepage process, the overall modeling approach, the purpose of the SCM, and the consistency with the downstream model (the SMPA), a single-continuum representation of the fractured formation is appropriate. (Note that the impact of the rock matrix may not be ignored when considering other processes; a dual-continuum model is selected in these instances.) The seepage-relevant capillary-strength parameter is determined by calibrating the model against seepage-rate data from liquid-release tests (see Section 6.5.3). These data reflect the seepage process and contain information about seepage-relevant capillary properties of the fractured formation in the vicinity of an open drift. Thus, the inversely determined effective capillary-strength parameter is considered pertinent and appropriate for the intended use of the model. The simulated seepage can be increased by decreasing capillary strength or permeability. Consequently, the two parameters are negatively correlated if inversely determined from seepage-rate data. Because only seepage data are available for calibration, the parameters are expected to be strongly correlated. That is, it is unlikely that they can be determined independently from one another and with a reasonably low estimation uncertainty. To reduce correlations and to improve the conditioning of the inverse problem, only the capillary-strength parameter is estimated through inverse modeling, whereas the permeability is fixed during the inversion. The choice of this calibration parameter is further discussed in Section 6.6.3.1. The permeability field is considered the result of a stochastic process. The geostatistical properties of the field are determined from air-injection tests (see Section 6.5.2). Multiple realizations of the permeability field are generated and used in the inversions of data from the lower lithophysal zone. The permeability fields generated for simulations with the SCM are representative of the conditions currently encountered at the test locations of Yucca Mountain. Therefore, thermally and geochemically induced property changes do not need to be considered in this Model Report. They are addressed by the TH Seepage Model (BSC 2004 [DIRS 170338]) and the Thermal-Hydrologic-Chemical (THC) Seepage Model [Drift-Scale Coupled Processes (DST and THC Seepage) Models (BSC 2004 [DIRS 168848])]. 6.3.3.3 Drift Geometry General Description The overall drift size and geometry impact the seepage threshold and the seepage amount. Generally, a larger drift exhibits a significantly lower seepage threshold because more water accumulates in the boundary layer as it migrates over a longer diversion distance around the wide opening. Because of the nonlinear impact of cavity size on seepage (Philip et al. 1989 [DIRS 105743]), seepage into large openings cannot be easily inferred from cumulative seepage into small cavities. The effectiveness of a capillary barrier is highest if the shape of the cavity follows an equipotential surface. In a homogeneous medium, parabolic cavities are more efficient in preventing seepage than circular or flat-roofed openings. Breakouts in the drift ceiling, as a result of rock fall and general drift degradation, may change the overall drift geometry and lead to local topographic lows, which may trap water, reduce or prevent flow diversion, and thus initiate seepage. In addition, small-scale surface roughness tends to increase seepage if the amplitude of the irregularity is on the order of boundary-layer thickness. The latter is determined by the capillary strength of the formation. In a heterogeneous, fractured formation, the importance of drift shape and drift geometry may be diminished relative to that of flow channeling and local ponding conditions (see Birkholzer et al. (1999 [DIRS 105170], pp. 372–379) and Section 6.4.2). Model Conceptualization The impact of the overall geometry of the underground opening (ECRB Cross-Drift or niche) on seepage is accounted for through explicit discretization of the cavity. The ECRB Cross-Drift is approximated as being cylindrical, with a diameter of 5.0 m. The overall geometry of the niches is taken from survey data, thus including some medium-scale roughness from rock fall and large lithophysal cavities. Small-scale roughness is indirectly included through a discretization effect. The length of the last vertical connection from the gridblocks representing the formation and the interface denoting the drift surface is 0.05 m (see Appendix C, Appendix D, and Appendix E; see also related discussion in Section 6.6.1.2). The choice of this nodal distance affects seepage because (in the model) no horizontal flow diversion can occur closer than 0.05 m from the drift wall. Since water is laterally diverted only if capillary suction is on the order of 0.05 m or higher, the discretization has an effect similar to that of (1) drift-wall roughness of amplitude of 0.05 m, with troughs at the gridblock centers and ridges along the gridblock interfaces, or (2) short fractures cutting into the opening, with a distance to the next fracture intersection of 0.05 m. Consequently, the effective capillary-strength parameter estimated by inverse modeling depends on the chosen discretization; it contains a geometric component related to the length of the nodal distance between the formation and the drift. The estimate is thus model-related, and the discretization between the calibration model and the prediction model must be consistent. In summary, the geometric factors affecting seepage are accounted for through (1) explicit discretization of the opening (which includes the overall shape as well as medium-scale roughness from break-outs lithophysal cavities), (2) by preventing flow diversion in a 0.05 m thick layer around the drift (mimicking small-scale surface roughness with a 0.05 m amplitude of the irregularities), and (3) the estimation of an effective capillary-strength parameter. The inclusion of small-scale surface roughness (exceeding an amplitude of 0.05 meters) and discrete effects from small fractures into an effective capillary-strength parameter is appropriate because their impact on seepage rates is directly related to capillarity. Note that the nominal diameter of a repository drift is 5.5 m, which is slightly larger than that of the ECRB Cross-Drift (5.0 m). This difference is of no significance, because the seepage-related parameters are determined using a model with the correct diameter (5.0 m) to be used for the analysis of liquid-release tests in the ECRB Cross-Drift. These parameters are then applied in the prediction model, which simulates seepage into an opening with a 5.5 m diameter. The impact of drift-shape changes as a result of drift degradation is discussed in Seepage Model for PA Including Drift Collapse (BSC 2004 [DIRS 167652]). 6.3.3.4 Evaporation Conditions General Description Reduced relative humidity in the underground opening leads to evaporation of water at the drift surface and the development of a dry-out zone in the vicinity of the cavity. Part or all of the water reaching the ceiling of the opening during a liquid-release test may evaporate, depending on the evaporation potential in the drift and the wet area exposed to evaporation. The evaporation potential depends on the relative humidity in the opening and the thickness of a diffusive boundary layer at the drift surface, which in turn is governed by the air velocity in the ventilated drift. The size of the wet spot developing at the drift ceiling as a result of liquid release above the drift depends on the formation properties, the spreading mechanism along the drift surface, and evaporation itself. As water injected during a liquid-release test reaches the opening, it spreads along the surface on account of capillary tension within the rough drift wall. As a result, water potentially seeping into the opening may not only form droplets or lines of water along fracture traces with a small surface area, but may spread across the drift surface over a relatively large area. This phenomenon is qualitatively confirmed by the geometry of the wet spot observed at the niche ceiling during seepage experiments (BSC 2004 [DIRS 170004], Section 6.2.1.3.4, Figure 6.2.1-7; Trautz and Wang 2002 [DIRS 160335], Figures 7 and 9). The geometry of the wet spot does not have a clear correlation with the visible fracture traces. Even though water first appears along fracture traces (Trautz and Wang 2002 [DIRS 160335], Figure 10), the wet spot grows in an areal fashion. It is obvious from the short arrival time and the average speed at which the leading edge of the plume moves across the ceiling that the water is not transmitted through the matrix, but spreads along the ceiling as a surface film, possibly supported by flow through microfractures. Evaporation from such wet areas is similar to evaporation from a free water surface, where the evaporation rate is governed by one-dimensional vapor diffusion across a relatively thin boundary layer of linearly decreasing vapor concentration. Temporal shrinkage of the wet spot can be correlated to increased evaporation as a result of changed ventilation regime, highlighting the coupled nature of the process. In a closed-off and humidified niche, potential evaporation at the wall or in the capture system is expected to be small compared to the amount of water being released. Seepage experiments in the middle nonlithophysal zone of the Topopah Spring welded unit were conducted in niches that were closed off by a bulkhead, which leads to comparatively high relative humidity and low air circulation. Moreover, a humidifier was used in some of the experiments to ensure high relative humidity. For these conditions, Ho (1997 [DIRS 141521]) and Or and Ghezzehei (2000 [DIRS 144773]) provide a detailed description of evaporation mechanisms on the scale of individual water droplets within fractures or emerging from fractured formations. The evapo-infiltration threshold calculated by Ho (1997 [DIRS 141521], p. 2670) is significantly lower than the applied injection rates, suggesting a very minor influence of evaporation on measured seepage rates in experiments conducted in the niches. Model Conceptualization Evaporation effects are included in the modeling of liquid-release tests performed in the ventilated ECRB Cross-Drift as well as in Niche 5. Evaporation effects are neglected in the modeling of liquid-release tests conducted in the closed-off niches in the middle nonlithophysal zone, i.e., Niches 2, 3, and 4. As demonstrated in Section 6.7, the impact of slight evaporation in a closed-off and moisturized niche on seepage rates—and thus on the estimation of seepage-relevant parameters—is minor. Evaporation effects are accounted for in the model by prescribing the measured relative humidity in the opening as a temporally varying water-potential boundary condition. Evaporation is calculated as a function of the water-potential gradient at the drift surface, the vapor diffusion coefficient, and the thickness of the diffusive boundary layer, which is estimated from evaporation pan measurements. A detailed description of the corresponding conceptual and mathematical model and the estimation of the evaporation boundary-layer thickness is given in Sections 6.6.1.3 and 6.6.1.4. Predictions of long-term seepage using the SMPA are based on the presumption of 100 percent relative humidity in the waste emplacement drifts, yielding higher seepage estimates than those expected in a ventilated environment. 6.3.3.5 Lithophysal Cavities General Description The impact of lithophysal cavities on flow and seepage is twofold: (1) lithophysal cavities are essentially obstacles to water flow because they act as capillary barriers, focusing the water that flows around them; (2) lithophysal cavities intersected by the drift lead to a rough drift wall, potentially creating seepage points at local topographic lows. Both effects tend to promote seepage. Model Conceptualization The effect of lithophysal cavities on seepage can be captured through the estimation of an effective capillary-strength parameter, making the explicit inclusion of lithophysal cavities into the process model unnecessary. This approach is considered appropriate for the following reasons: (1) omitting lithophysal cavities in the process model used for inverse modeling yields lower estimates of the capillary-strength parameter; (2) consistency between the calibration model (the SCM) and the prediction model (the SMPA) removes the impact of a potential estimation bias; (3) the approach allows for the development of a single SMPA conceptual model for both the middle nonlithophysal and lower lithophysal zones, yielding a single look-up table for TSPA to sample from; and (4) explicit modeling of lithophysal cavities is not warranted because of insufficient information regarding their location, shape, and frequency. Note that the impact of lithophysal cavities on surface roughness in Niche 5 is accounted for through explicit discretization of the niche’s geometry, based on survey data (see Appendix E). 6.3.4 General Modeling and Data-Analysis Approach The key element of the approach chosen to simulate seepage and determine seepage-relevant parameters is the reliance on inverse modeling. Given the complexity of the seepage process in a fractured porous medium, it is considered unfeasible to develop a detailed process model with a deterministic calculation of unsaturated water flow, through a fracture network that exhibits multiscale variabilities in hydraulic properties. Such a model would also require an accurate representation of the seepage process, which includes effects from small-scale roughness and small-scale heterogeneities, film flow within fractures and along the drift surface, drop formation and detachment, and other processes. The necessary characterization data needed to carry out such a detailed simulation are not available. As discussed in the following paragraph, such a detailed simulation is not necessary for an adequate treatment of the issue. The difficulties mentioned above can be effectively overcome by recognizing that (1) detailed simulation of individual seeps is not necessary to estimate average seepage rates into waste emplacement drifts, (2) certain factors affecting seepage can be lumped into an effective parameter, (3) calibrating a model against data from seepage experiments ensures that the model captures the relevant processes, (4) estimating effective parameters partly compensates for processes and features that are not explicitly considered in the model, and (5) the estimated parameters are optimal and can be directly used in the prediction model. The main advantage of this approach is that it relies directly on seepage-rate data, which inherently contain information about the relevant processes. Moreover, the calibration data (seepage rates on the scale of a drift section) are very similar to the measure of interest for the subsequent predictions. The consistency between the calibration model used to derive seepage-relevant parameters and the prediction model used to forecast seepage minimizes potential conceptual differences and large systematic errors. The advantages of the selected method over alternative approaches are further evaluated in Section 6.4. The SCM is conceptualized as a three-dimensional, heterogeneous continuum model. The continuum mainly represents the dense fracture network that dominates the seepage process. The SCM is conceptually consistent with the site-scale model of the unsaturated zone at Yucca Mountain and submodels thereof, specifically the SMPA and TH Seepage Model (BSC 2004 [DIRS 170338]). This makes it straightforward to embed the SCM into the current modeling framework. As will be discussed in Section 6.6.3.2, the SCM is calibrated against late-time seepage-rate data from liquid-release tests. Early-time seepage data are discarded because they are affected by storage effects and the properties of a few fractures connecting the injection interval with the opening. These fractures are not necessarily representative of the fracture network that is engaged in flow diversion around the entire opening under steady-state conditions. Late-time data are more representative of near-steady conditions and are less influenced by storage effects. Moreover, the relatively large amount of released water at late time has likely encountered a significant portion of the capillary barrier. As a result, the late-time seepage data better reflect average conditions on the scale of interest. The duration of the liquid-release tests is on the order of days and weeks, whereas the calibrated parameters are intended to be used in a steady-state prediction model (the SMPA). Nevertheless, the late-time seepage-rate data are considered suitable for calibrating a model that subsequently will be used for the prediction of long-term seepage behavior. The approach is appropriate for the following reasons. First, any data that are sensitive to the parameters of interest are generally adequate for model calibration. There is no inherent requirement that the data used for model calibration have to reflect steady-state conditions if the ultimate purpose of the model is to predict steady-state behavior. If the model is capable of capturing the transient effects occurring during the liquid-release tests, no unwanted bias is introduced; such a bias would only be introduced if a steady-state model were calibrated against non-steady-state data. The SCM is a transient model that simulates time-dependent liquid release, flow, storage, and seepage processes. If the SCM can be successfully calibrated and validated, the parameters determined by inverse modeling are not affected by the transient nature of the underlying data, and thus they are also suitable for the prediction of steady-state seepage. Second, the late-time seepage-rate data used for model calibration show near-steady behavior, i.e., they do not change significantly with time. They closely reflect the processes governing steady-state seepage and are thus suitable as calibration data for a prediction model of long-term seepage into waste emplacement drifts. The capillary-strength parameter will be determined by calibrating the model against multiple tests using different liquid-release rates. Some of these release rates induced a local percolation flux above the seepage threshold, i.e., water dripped into the opening and yielded seepage-rate data valuable for calibration. However, the joint inversion of multiple data sets also included data from tests performed below the seepage threshold. Moreover, the model was validated against tests conducted above and below the seepage threshold. That is, the system is probed and the model will be calibrated and validated for the critical range of percolation rates about the seepage threshold. Seepage predictions for natural percolation fluxes, that are even lower than the low fluxes (below the seepage threshold) induced during the low-rate tests, will yield the correct result, namely zero seepage. As a result of a high-infiltration climate or strong flow focusing, the natural percolation flux may be high and exceed the seepage threshold. This would be the critical scenario for performance. Obviously, the parameters estimated from the liquid-release tests would be most suitable for those critical circumstances, because they were determined under similar high-rate conditions. In summary, the parameters determined from relatively high-rate liquid-release tests are appropriate and provide a solid basis for seepage predictions under low and higher natural percolation fluxes. Liquid-release tests directly supporting the SCM were conducted in two different hydrogeologic units, in multiple test beds, boreholes, and intervals. Each test event probes a different portion of the rock and a different section of the underground opening. The question arises how the available data should be combined to yield suitable averages and reasonable measures of variability and uncertainty, which are needed for model validation and the subsequent PA calculations. The goal is to obtain a probability density function of the seepage-relevant parameters that reflects both estimation uncertainty and spatial variability. These two aspects are discussed separately in the following paragraphs. Parameter estimates determined by inverse modeling are uncertain because they are derived from limited data, which exhibit random and potentially systematic measurement errors, and because the model is a simplification of the real system, which introduces systematic and random modeling errors. As discussed above, estimating model-related parameters mitigates the impact of some of the residual systematic errors. Estimation uncertainty as a result of random noise in the seepage data is relatively minor (see Sections 6.6.3.3 and 8.2). However, there remains irreducible uncertainty because of small-scale heterogeneity that affects individual seepage tests. The details of these small-scale heterogeneities are unknown (i.e., they cannot be described deterministically) and vary from location to location (i.e., they are spatially variable). Consequently, they are considered the result of a stochastic process that must be described by geostatistical parameters and modeled by performing multiple geostatistical simulations. Each seepage data set is obtained from a certain test bed (niche or section of the ECRB Cross-Drift); it can be considered one realization from a number of statistically similar geologic systems. The lack of knowledge regarding the details of this specific realization makes the inversely determined parameter estimate uncertain. This uncertainty is examined by performing multiple inversions of the same data set using different realizations of the underlying heterogeneous permeability field, yielding a distribution of estimated capillary-strength parameters rather than a single value. In addition to capturing the random nature of the permeability field and its impact on seepage, each realization will induce some ergodic fluctuations, which reflect the fact that the model statistics are inferred from sparse air-permeability sampling (i.e., they are not deemed exactly representative of the population statistics). The average of all inversions performed with different permeability fields for a given interval yields one estimate representative of that location. The average parameters obtained from multiple simultaneous inversions of one or more seepage events conducted in a certain test interval are considered independent, each reflecting the seepage-relevant properties at a given location on the drift scale. These estimates are then combined to yield a parameter distribution for the entire hydrogeologic unit. This distribution reflects spatial variability. By sampling from the distribution of the resulting parameter estimates, probabilistic predictions of seepage across the repository horizon can capture the spatial variability of average seepage on the scale of the 5 m long drift segment. Uncertainty in this average seepage rate as a result of small-scale heterogeneity is calculated based on multiple seepage prediction runs by the SMPA, using multiple realizations of the underlying permeability field. A comparison of seepage predictions with observable data (such as seepage data from transient liquid-release tests involving a finite amount of water) is a necessary step in model development and confidence building. However, models are often developed—and most usefully—to infer behavior that cannot be directly observed (such as long-term near-steady seepage under naturally low percolation fluxes). The appropriateness of such an extrapolation of the model beyond its tested grounds needs to be assessed. While rigorous model testing is fundamentally not possible [“Verification, Validation, and Confirmation of Numerical Models in the Earth Sciences” (Oreskes et al. 1994 [DIRS 152512])], validation of the model for a limited purpose can be accomplished. The SCM is partially tested against observable data from seepage experiments that were not used for model calibration. The purpose of this validation exercise is to determine whether the model is appropriate and adequate for its intended use. Remaining uncertainty will be quantified during the seepage abstraction process and propagated through the PA models. Additional remarks about model validation can be found in Section 7.1. The development of the SCM involves the following steps (note that this is a general description; details about the implementation and execution of these steps can be found in Sections 6.6.2, 6.6.3, and the appendices): 1. Geostatistical parameters of the permeability field are determined from the results of air-injection test data. 2. Multiple realizations of the permeability field are generated, each being consistent with the geostatistical properties of the measured air permeabilities. 3. A numerical mesh is generated. This step involves (a) making a primary 3-D grid, (b) translating coordinates to center the mesh, (c) mapping the permeability field onto the mesh, (d) cutting out the opening (niche or drift) from the mesh, (e) adding top and bottom boundary elements as well as an evaporation boundary in the opening, and (f) modifying elements representing injection intervals. 4. An input file defining the forward problem is prepared. This step involves (a) assembling parameters representing hydrogeologic properties, (b) assigning appropriate properties to elements representing the excavation and borehole intervals, (c) extracting the background percolation flux from the UZ Model, (d) assigning appropriate initial and boundary conditions, and (e) selecting computational parameters and program options. 5. Simulations with the background percolation flux applied at the top of the model are run to steady state to obtain initial conditions for the subsequent simulation of transient seepage experiments. 6. Injection rates are specified as time-dependent source terms. 7. Test events are selected for calibration. Seepage rates are calculated from the cumulative seepage data. 8. An input file defining the inverse problem is prepared. This step involves (a) selecting the parameters to be estimated and their initial guesses, (b) selecting points in time at which calibration should occur, (c) specifying the data against which calibration should occur, and (d) selecting computational parameters and program options. 9. Seepage-relevant, model-related parameters are determined by automatic model calibration using iTOUGH2 V4.0 (LBNL 1999 [DIRS 139918]) and V5.0 (LBNL 2002 [DIRS 160106]). In each inversion, seepage-rate data from multiple test events are jointly inverted. 10. The model is tested by comparing predicted seepage rates to observed data from seepage experiments not used during model calibration. Prediction uncertainty is calculated by iTOUGH2 V4.0 (LBNL 1999 [DIRS 139918]) and V5.0 (LBNL 2002 [DIRS 160106]) using first-order-second-moment (FOSM) uncertainty propagation analysis and Monte Carlo simulations. 6.4 ALTERNATIVE CONCEPTUAL MODELS The following subsections contain short descriptions of potential alternative ways to evaluate seepage into waste emplacement drifts at Yucca Mountain. These alternative conceptual models are discussed in a qualitative manner, and references to more detailed analyses are given, if available. No quantitative evaluations of these alternative conceptual models are presented in this Model Report. Natural analogues for seepage also support the concepts of the base-case model; they are briefly reviewed in Section 7.2.1. In general, the choice of a conceptual model should be based on a careful consideration of the study objectives, the available database in comparison with the data needs, the uncertainty in the input parameters and the corresponding prediction uncertainties and computational aspects. 6.4.1 Discrete Fracture Network Model A discrete fracture network model (DFNM) is an alternative conceptual model to the heterogeneous continuum model used in this Model Report [“Alternative Concepts and Approaches for Modeling Flow and Transport in Thick Unsaturated Zones of Fractured Rocks” (Pruess et al. 1999 [DIRS 117112], pp. 307–309)]. A high-resolution DFNM is capable of generating channelized flow and discrete seepage events, as demonstrated by Finsterle (2000 [DIRS 151875], Plate 1) and Liu et al. (2002 [DIRS 160230], Figure 5). Note that two-dimensional DFNMs (such as those shown in Figure 6-2 and Figure 6-3) make the implicit assumption that the fractures are oriented parallel to the drift axis. This assumption exaggerates the discreteness of the flow and seepage behavior as flow diversion within the fracture plane is not possible (see also Figure 6-1 and related discussion in Section 6.3.2). In a DFNM, an individual fracture is discretely represented by appropriately small computational grid blocks. This is often considered the defining feature of such a model. (Note, however, that the flow equations solved in and between grid blocks are essentially identical to those solved by a continuum model. As outlined in Section 6.3.2, in-plane flow diversion is the key mechanism characterizing the capillary barrier effect and determining the seepage threshold. In comparison, the geometrical details and discreteness of fracture-to-fracture flow are secondary aspects for seepage.) (a) High-Resolution Permeability Field b) D (b) Flow Paths (a) Fracture Network Liu et al. (2002 [DIRS 160230], Figures 1 and 2). Figure 6-3. Two-Dimensional Discrete Fracture Network Model The development of a defensible DFNM requires collecting a very large amount of geometric and hydrologic data. While part of the required geometric information can be obtained from fracture mappings, the description of the network remains incomplete and potentially biased towards fractures of a certain orientation and a certain size. Moreover, unsaturated hydrological parameters on the scale of individual fractures are required, along with conceptual models and simplifying assumptions regarding unsaturated flow within fractures and across fracture intersections. The databases required to develop a defensible DFNM are currently not available and are generally difficult or even impossible to obtain for site-specific simulations. The cumulative effect of the input uncertainties is likely to outweigh the apparent advantage of a detailed representation of the fracture network, specifically since the DFNM must be calibrated against hydrogeologic data to reduce prediction uncertainty—that is, an approach very similar to that outlined in Section 6.3.4 must be followed. The appropriateness of using a continuum model for the prediction of average seepage quantities was demonstrated by Finsterle (2000 [DIRS 151875]). In that study, seepage predictions with a calibrated fracture continuum model were compared to those of a DFNM, yielding consistent results even when applied outside the range of calibration. Given these results, the parsimony of the continuum model is considered a key advantage over the complexity of the DFNM, which is difficult to support or justify despite its visual appeal. Moreover, a two-dimensional DFNM is not capable of capturing flow diversion within the fracture plane, a mechanism appropriately represented by a 2-D (or 3-D) continuum model. For the reasons outlined above, the full development of a DFNM as a potential alternative to the base-case continuum model was considered unwarranted, infeasible, and unnecessary. While seepage calculations with a calibrated DFNM are likely to corroborate the findings of this Model Report, this approach is not further considered. 6.4.2 Seepage Governed by Ponding Probability As an alternative conceptual model to a seepage process model, Birkholzer et al. (1999 [DIRS 105170], pp. 372–379) related seepage to the local ponding probability, which was derived from the variability of the permeability field. Their approach assumed that—in strongly heterogeneous formations—seepage is predominantly affected by pressure variations governed by local heterogeneity rather than the presence and geometry of the capillary barrier. This is different from the behavior in a homogeneous system, where the geometry of the capillary barrier has a strong impact on seepage (Philip et al. 1989 [DIRS 105743]). Strong medium- to small-scale heterogeneities tend to increase seepage because they increase channeling and local ponding. This effect is included in the current seepage process models through the estimation of effective, seepage-specific parameters for a heterogeneous medium with a heterogeneous permeability field. While the approach presented by Birkholzer et al. (1999 [DIRS 105170], pp. 372–379) may provide guidelines for how to extrapolate seepage predictions to other units or drift geometries, it nonetheless requires a calibration step similar to that described in this Model Report. The approach is therefore not further considered. Nevertheless, the concept that ponding probability affects seepage is consistent with and thus corroborates the base-case model, which produces random seepage locations as a result of local ponding in a stochastic permeability field. 6.4.3 Inferring Seepage from Geochemical Data Observations of calcite and opal in lithophysal cavities could be used to estimate long-term seepage rates into these small openings [Analysis for Geochemical Data for the Unsaturated Zone (BSC 2003 [DIRS 168343], Section 6.10.1)]. Calcite is assumed to precipitate from downward-percolating meteoric water because of (1) evaporation, (2) CO2 outgassing as a result of the geothermal gradient, and (3) interaction with a gas phase containing less CO2 than the gas with which the water was last equilibrated. Considering these calcite-precipitation mechanisms and assuming certain water-to-calcite ratios, seepage into lithophysal cavities can be estimated from calcite-deposition data. The analysis of calcite and opal precipitation data shows that (1) not all lithophysal cavities encountered seepage, and (2) seepage flux derived from mineral deposits is a very small fraction of percolation flux [“Estimation of Past Seepage Volumes from Calcite Distribution in the Topopah Spring Tuff, Yucca Mountain, Nevada” (Marshall et al., 2003 [DIRS 162891], Section 5)]. Both conclusions corroborate the general concept of a capillary barrier reducing seepage below the value of the percolation flux. The advantage of using geochemical information to infer seepage is the fact that calcite and opal were deposited over a long period of time under natural percolation conditions. The disadvantage of this approach is that (1) seepage is inferred in an indirect manner, requiring a number of geochemical models with their associated assumptions—in addition to hydrogeologic model assumptions; (2) the calcite depositions on lithophysal cavity floors may not originate from dripping water (i.e., seepage); in fact, there is a lack of evidence of dripping from cavity ceilings (absence of stalactites or stalagmites), even where fractures containing coatings intersect lithophysae ceilings [“Physical and Stable-Isotope Evidence for Formation of Secondary Calcite and Silica in the Unsaturated Zone” (Whelan et al. 2002 [DIRS 160442], p. 744)]; (3) the data reflect seepage into (small) cavities instead of seepage into a (large) waste emplacement drift; since the size of the underground opening impacts seepage in a nonlinear fashion, a hydrological, physically based process model is required to estimate seepage on the scale of interest; (4) seepage into lithophysal cavities does not include potential impacts from the excavation- disturbed zone around a mechanically constructed drift; and (5) the historic record and the approach does not allow making predictions into the future under changed conditions. As shown by Marshall et al. [“Seepage Flux Conceptualized from Secondary Calcite in Lithophysal Cavities in the Topopah Spring Tuff, Yucca Mountain, Nevada” (2000 [DIRS 151018], Figure 1), the seepage rates estimated from the calcite-deposition data are significantly lower than those predicted by TSPA using data derived from the SMPA, which is based on the methodology outlined in this Model Report. Inferring seepage from secondary mineral depositions in lithophysal cavities is not further considered as an approach to quantitatively estimate seepage into waste emplacement drifts. 6.4.4 Inferring Seepage Threshold Directly From Liquid-Release Tests Trautz and Wang (2002 [DIRS 160335], Section 5) estimated the seepage threshold directly from the liquid-release test data, based on a number of simplifying assumptions (with regard to the cross-sectional area of the flow path between the borehole and the ceiling, evaporation, and the steady-state flow field). Once the seepage threshold was determined, a capillary-strength parameter was derived assuming seepage into a cylindrical cavity excavated from a homogeneous porous medium (Trautz and Wang 2002 [DIRS 160335], Section 6). The base- case model outlined in this Model Report relies on fewer assumptions than the simplified alternative conceptual model and predicts a lower seepage threshold; the base-case model described in this Model Report is therefore the preferred conceptualization. 6.5 DESCRIPTION OF SEEPAGE EXPERIMENTS 6.5.1 Test Location and Borehole Configuration The data used for the development, calibration, and validation of the SCM were collected as part of the ESF Drift Seepage Test and Niche Moisture Study, an ongoing field-testing program. Drift-scale seepage tests were initiated in 1997 to investigate potential seepage into an underground opening representing a waste emplacement drift. Short drifts ranging from 6.3 m to 15.0 m in length were constructed at various locations along the ESF and the ECRB Cross-Drift. Boreholes were installed prior to and after the drifts were excavated to facilitate characterization of the rock using air-injection tests and investigation of seepage processes using liquid-release tests. The short excavations are called “niches,” and the drift-scale seepage tests are collectively referred to as the Niche Study. In Niche 5, a horizontal slot on the side of the niche (also referred to as “batwing”) was excavated to obtain direct evidence of the flow-diversion capability of the capillary barrier (see Section 6.8). A second study referred to as the Systematic Borehole Testing Program was initiated in 2000 to complement the niche seepage experiments. The purpose of the program is to provide broad, systematic coverage and characterization of the lower lithophysal zone (Tptpll) of the Topopah Spring welded unit (TSw). Systematic characterization of the Tptpll is accomplished by performing air-injection and liquid-release tests in approximately 20 m long boreholes drilled into the ceiling approximately every 30 m along the ECRB Cross-Drift. The data used in this Model Report are a subset of seepage tests from the Niche Studies and the Systematic Borehole Testing Program. A few tests failed and their data are not used in this Model Report (see discussion of Table 6-5 below). Data include air permeabilities and seepage-rate values from tests conducted at three niche sites located along the Main Drift of the ESF, one niche in the ECRB Cross-Drift, and in three systematic testing boreholes drilled into the ceiling of the Cross-Drift (see Figure 6-4). The first three niche sites are located along the west side of the ESF in the Tptpmn and they were selected for study based on fracture and hydrologic data collected in the ESF. The first niche site at construction station (CS) 31+07 (Niche 3107, hereafter referred to as Niche 3) consists of a 6.3 m long drift located in an area of relatively low fracture density. Niche 3 is located in close proximity to CS 30+62, where the Cross-Drift crosses over the Main Drift of the ESF. The second niche site, at CS 36+50 (Niche 3650, hereafter referred to as Niche 2), consists of a 9 m long drift located in a competent rock mass exhibiting relatively moderate fracture density. The third niche site, at CS 47+88 (Niche 4788, hereafter referred to as Niche 4), consists of an 8.2 m long drift located in a 950 m long exposure of an intensely fractured zone. Fractures in this zone are not uniformly spaced, but instead they occur in clusters of closely spaced fractures. The 15.0 m-long Niche 5 is located on the south side of the ECRB Cross-Drift in the Tptpll. NOTE: The shape of the openings is approximate. Figure 6-4. Schematic Geologic Map Showing Approximate Location of Niches and Systematic Testing Boreholes SYBT-ECRB-LA#1–3 Prior to niche excavation, horizontal boreholes were drilled to gain access to the rock for testing and monitoring purposes. The boreholes above each niche are approximately one meter apart and within the same horizontal plane. Table 6-3 provides the correlation between the borehole designations shown in the schematic cross sections of Figure 6-4 (and used throughout this document) and their respective designations in the survey DTN. Note that throughout Project documents, the systematic testing boreholes are designated as either SYBT-ECRB-LA#x or ECRB-SYBT-LA#x; these designations are unambiguous and thus interchangeable without loss of traceability. The format SYBT-ECRB-LA#x is used in this Model Report, consistent with most DTN entries. Table 6-3. Borehole Designations in Niches Niche Borehole Designation in DTN DTN of Borehole Survey 3107 (Niche 3) UL UM UR ESF-MD-NICHE 3107 #5 ESF-MD-NICHE 3107 #6 ESF-MD-NICHE 3107 #7 MO0002GSC00064.000 [DIRS 152625] 3650 (Niche 2) UL UM UR ESF-MD-NICHE 3650 #1 ESF-MD-NICHE 3650 #2 ESF-MD-NICHE 3650 #3 MO0002GSC00076.000 [DIRS 152623] 4788 (Niche 4) UL UM UR ESF-MD-NICHE 4788 #5 ESF-MD-NICHE 4788 #6 ESF-MD-NICHE 4788 #7 MO0107GSC01069.000 [DIRS 156941] 1620 (Niche 5) #2 #3 #4 #5 #6 #7 ECRB-NICHE 1620 #2 ECRB-NICHE 1620 #3 ECRB-NICHE 1620 #4 ECRB-NICHE 1620 #5 ECRB-NICHE 1620 #6 ECRB-NICHE 1620 #7 MO0312GSC03176.000 [DIRS 169532] NOTE: No liquid-release tests were performed in Niche 3566 (Niche 1). DTN=Data Tracking Number; UL=upper left; UM=upper middle; UR=upper right The boreholes listed in Table 6-3 are approximately parallel to the niche axis. Air-injection tests were conducted in several, 1 ft (0.3 m) long, packed-off intervals, both prior to and after niche excavation, to determine the permeability distribution of the formation, as well as to study potential permeability changes as a result of stress relief during niche excavation. After niche construction, water was injected at a specified rate into intervals of the same boreholes to observe, document, and quantify any water migrating to and seeping into the niche. The systematic testing boreholes SYBT-ECRB-LA#1, 2, and 3 are drilled from the ECRB and located in the moderately to densely welded, devitrified, and vapor-phase altered lower lithophysal zone (Tptpll). Borehole SYBT-ECRB-LA#1 is collared from the drift crown at ECRB construction station CD 17+49. It is upward-inclined at nominal 15° from the drift axis. Packers are set to isolate an injection zone between 10 ft (3.0 m) and 16 ft (4.9 m) (zone 2) from the collar (DTN: LB0110ECRBLIQR.002 [DIRS 156879]). Borehole SYBT-ECRB-LA#2 is collared from the drift crown at ECRB construction station CD 17+26. It is upward-inclined at nominal 15° from the drift axis. Packers are set to isolate three 6 ft (1.8 m) long injection zones between 17 ft (5.2 m) and 23 ft (7.0 m) (zone 1), 33 ft (10.1 m) and 39 ft (11.9 m) (zone 2), and 49 ft (15.0 m) and 55 ft (16.8 m) (zone 3) from the collar (DTN: LB00090012213U.002 [DIRS 153154]). Borehole SYBT-ECRB-LA#3 is collared from the drift crown at ECRB construction station CD 16+95. It is upward-inclined at nominal 15° from the drift axis. Packers are set to isolate three 6 ft (1.8 m) long injection zones between 18 ft (5.5 m) and 24 ft (7.3 m) (zone 1), 34 ft (10.4 m) and 40 ft (12.2 m) (zone 2), and 50 ft (15.2 m) and 56 ft (17.1 m) (zone 3) from the collar (DTN: LB0203ECRBLIQR.001 [DIRS 158462]). 6.5.2 Air-Injection Tests The purpose of the air-injection tests was to estimate permeabilities as a basis for the stochastic generation of heterogeneous permeability fields. The tests were performed by isolating a short section of the boreholes (1 ft [0.3 m] in niches, 6 ft [1.8 m] in systematic testing borehole SYBT-ECRB-LA#2), using an inflatable packer system, and then injecting compressed air at a constant rate into the isolated injection interval. The pressure buildup in the injection interval and in nearby observation intervals was monitored with time until steady-state conditions were reached, which typically occurred within a few minutes. Air injection was terminated after reaching steady-state pressures, and the decline in air pressure was then monitored as it recovered to its initial pre-test condition. Air-permeability values were derived from the steady-state pressure data (BSC 2004 [DIRS 170004], Section 6.1.2) based on a commonly used analytical solution [Pneumatic Testing in 45-Degree-Incline Boreholes in Ash-Flow Tuff Near Superior, Arizona (LeCain 1995 [DIRS 101700], p. 10, Eq. (15))]. The air permeabilities around the niches and the ECRB Cross-Drift are affected by excavation (BSC 2004 [DIRS 170004], Section 6.1.2.2; “Permeability Changes Induced by Excavation in Fractured Tuff” (Wang and Elsworth 1999 [DIRS 104366]), pp. 752–756). Since seepage is determined by the formation properties in the immediate vicinity of the opening, it is reasonable to use post-excavation air-permeability data for seepage calculations. Note that the perturbation of the permeability in the drift vicinity depends on the excavation method. A tunnel-boring machine is used for the excavation of the ECRB, whereas a road header is used to mine out the niches. Since local post-excavation air-permeability values are directly used for the analysis of seepage-rate data, no bias is introduced. The permeabilities used during TSPA-LA are sampled from a distribution that describes variability and uncertainty, including uncertainty induced by excavation effects (BSC 2004 [DIRS 169131], Section 6.6). These permeabilities are considered representative of the absolute permeability of the excavation-disturbed zone around the opening, because the post-excavation air-injection tests were conducted in a network of essentially dry fractures, i.e., no empirical relative permeability function is needed to translate air conductivity into absolute permeability. Since air-injection tests are a standard method to obtain permeability values, the use of these values during both calibration and prediction of seepage ensures consistency. The distributions representing variability and uncertainty in permeability (BSC 2004 [DIRS 169131], Section 6.6) were developed also based on air-permeability data. This consistency reduces the impact of a potential bias. Data that are located outside the footprint of the niches were removed from the data set (Ahlers 2002 [DIRS 161045], p. 20; Trautz 2001 [DIRS 161044], p. 20) because they represent a separate population of air permeabilities performed in an area of relatively undisturbed, lower-permeability rock. Mean and standard deviations for each of the four locations are summarized in Table 6-4. Here, standard deviations reflect spatial variability within the test bed. The number of log-permeability values available is indicated in the last column. Mean permeabilities and their spatial variability as calculated for the three niches located in the middle nonlithophysal zone are consistent with one another. Permeability in the lower lithophysal zone is approximately one order of magnitude larger. The variability as measured in Niche 5 is significantly larger than that obtained in borehole SYBT- ECRB-LA#2. This is partly a result of the injection intervals of borehole SYBT-ECRB-LA#2 being six times longer than those in Niche 5 are. Note no air-permeability data are available from boreholes SYBT-ECRB-LA#1 and SYBT-ECRB-LA#3 because of equipment problems during air-injection testing. Table 6-4. Mean and Standard Deviation of Post-Excavation Log-Air-Permeability Values Location Input DTN Scientific Notebook Reference Mean Log (k [m2]) Std. Dev. Na Niche 2 LB0011AIRKTEST.001 [DIRS 153155] Trautz 2001 [DIRS 161044], pp. 19–25 -11.66 0.72 84 Niche 3 LB990601233124.001 [DIRS 105888] Ahlers 2002 [DIRS 161045], pp. 39–40 -12.14 0.80 78 Niche 4 LB990601233124.001 [DIRS 105888] Ahlers 2002 [DIRS 161045], pp. 15–21 -11.79 0.84 63 Niche 5 LB0110AKN5POST.001 [DIRS 156904] Wang 2003 [DIRS 161456], SN-LBNL-SCI-223-V1, pp. 19–20 -10.95 1.31 61 SYBT-ECRB-LA#2 LB00090012213U.001 [DIRS 153141] Finsterle 2002 [DIRS 161043], pp. 54–55 -10.73 0.21 6 a Number of log-permeability values DTN=Data Tracking Number; Std Dev=Standard Deviation 6.5.3 Liquid-Release Tests Multiple liquid-release tests were performed in the niches and the ECRB Cross-Drift to characterize seepage into a large underground opening (BSC 2004 [DIRS 170004], Sections 6.2 and 6.11). The tests were performed by sealing a short section of the borehole above the opening using an inflatable packer system and then releasing water at a specified rate into the isolated test interval. Any water that migrated from the borehole to the ceiling and dripped into the opening was captured and weighed. Only a small amount of water (approximately one liter per test event) was released during testing at Niche 2, and only the total amount of water that seeped into the capture system was recorded. Seepage experiments at Niches 3, 4, 5, and in the systematic testing boreholes SYBT-ECRB-LA#1–3 involved significantly more water, which was injected over longer periods, and cumulative seepage was recorded as a function of time. In many intervals, multiple liquid-release tests were conducted using different injection rates with different lengths of inactivity between individual test events. The reason for using different injection rates and different injection schedules was to collect data that are sensitive to percolation rate and water storage effects. While the inverse modeling approach pursued in this Model Report does not require data above and below the seepage threshold, increasing the sensitivity of the data to seepage-related effects improves the identifiability of seepage-relevant parameters. Table 6-5 summarizes the test events used for the calibration and validation of the SCM. The approximate release rate (defined as the injection rate minus the return flow) is indicated in Column 4. As shown in Column 5, 53 out of 90 test events led to seepage into the capture system. Potential seepage was not recorded in two cases (Events 5 and 46) because of an equipment failure. While no data are available to be used for calibration or validation from these three test events, the injections that occurred were nevertheless modeled because the released water has a potential impact on subsequent test events. Column 6 indicates whether a specific test event was used for calibration (C) or validation (V). The selection of each test event for calibration or validation purposes is discussed in detail in Section 6.6.3.2. A few additional seepage tests were conducted in Niche 2 that were not used, because only a very small amount of water was released and generally, no seepage was observed. Injection attempts at zone 3 of borehole SYBT-ECRB-LA#3 (Event 77) failed because the zone was too tight. A few test events in Niche 5 were not analyzed because of various difficulties (Events 79 and 80: packer problem; Event 82: seepage partially bypassed capture system; Event 83: pump problem). During Events 87 and 88 in borehole SYBT-ECRB-LA#5, water from Events 83 and 85 (conducted in the neighboring borehole SYBT-ECRB-LA#3) entered the seepage collection system, interfering with the test results (see detailed discussion of Figure 6-17 below). The events without any seepage cannot be used for calibration (unless jointly inverted with other tests that exhibit seepage), because the corresponding inverse problem would be ill posed. These tests (along with tests showing seepage) are therefore used for validation of the SCM. The small amount of water released during the short-term tests performed in Niche 1 makes it difficult to reliably estimate seepage parameters on the drift scale. If used for calibration, these tests yield small-scale parameter values that are biased towards the properties of the few fractures connecting the release point with the niche ceiling. These fractures may not be representative of the fracture network taking part in the diversion of water around the entire niche, which is the behavior to be modeled under steady-state flow conditions. Moreover, storage effects are significant in short-term tests but are also poorly identifiable. For these reasons, the Niche 2 liquid-release tests are used for validation purposes only. The calculation of seepage rates from cumulative seepage data is described in Appendix F. Table 6-5. Liquid-Release Test Events, Approximate Release Rate, Occurrence of Seepage, and Their Use for Calibration or Validation Purposes Event Starting Date of Test Borehole, Interval Approximate Release Rate [ml/min] Seepage? Calibration, Validation Niche 2, DTN: LB980001233124.004 [DIRS 136583] 13 12/11/97 UL, 5.18–5.49 m 4.7 No V 14 02/12/98 UL, 5.18–5.49 m 0.4 No V 15 12/11/97 UL, 5.79–6.10 m 12.1 No V 16 12/11/97 UL, 6.40–6.71 m 12.7 No V 17 12/10/97 UL, 7.01–7.32 m 116.9 Yes V 18 01/06/98 UL, 7.01–7.32 m 11.4 No V 19 11/13/97 UM, 4.27–4.57 m 121.1 Yes V 20 12/03/97 UM, 4.27–4.57 m 30.2 Yes V 21 12/03/97 UM, 4.27–4.57 m 30.4 Yes V 22 01/07/98 UM, 4.27–4.57 m 2.8 Yes V 23 02/10/98 UM, 4.27–4.57 m 1.0 No V 24 11/12/97 UM, 4.88–5.18 m 173.5 Yes V Table 6-5. Liquid-Release Test Events, Approximate Release Rate, Occurrence of Seepage, and their Use for Calibration or Validation Purposes (Continued) Event Starting Date of Test Borehole, Interval Approximate Release Rate [ml/min] Seepage? Calibration, Validation 25 12/04/97 UM, 4.88–5.18 m 30.4 Yes V 26 12/05/97 UM, 4.88–5.18 m 8.6 Yes V 27 01/08/98 UM, 4.88–5.18 m 2.8 No V 28 03/06/98 UM, 4.88–5.18 m 0.8 No V 29 11/13/97 UM, 5.49–5.79 m 124.1 Yes V 30 12/04/97 UM, 5.49–5.79 m 30.2 Yes V 31 01/09/98 UM, 5.49–5.79 m 3.5 Yes V 32 02/11/98 UM, 5.49–5.79 m 0.8 No V 33 11/13/1997 UM, 6.10–6.40 m 30.8 No V 34 12/04/1997 UM, 6.10–6.40 m 11.5 No V 35 01/12/1998 UM, 6.10–6.40 m 47.5 No V 36 01/14/1998 UR, 4.27–4.57 m 11.9 Yes V 37 02/05/1998 UR, 4.27–4.57 m 3.3 No V 38 01/15/1998 UR, 4.88–5.18 m 11.4 Yes V 39 02/06/1998 UR, 4.88–5.18 m 3.2 No V Niche 3, DTN: LB0010NICH3LIQ.001 [DIRS 153144] 1 03/10/99 UL, 5.49–5.80 m 1.5 No V 2 03/30/99 UL, 5.49–5.80 m 2.0 No V 3 09/17/99 UL, 5.49–5.80 m 1.5 No V 4 03/04/99 UM, 4.88–5.18 m 0.9 No C 5 04/07/99 UM, 4.88–5.18 m 5.8 – a – a 6 04/27/99 UM, 4.88–5.18 m 2.4 Yes C Niche 3, DTN: LB0010NICH3LIQ.001 [DIRS 153144] (Continued) 7 04/30/99 UM, 4.88–5.18 m 0.8 No V 8 05/06/99 UM, 4.88–5.18 m 5.4 Yes C 9 09/21/99 UM, 4.88–5.18 m 5.0 Yes V 10 09/23/99 UM, 4.88–5.18 m 5.3 Yes V 11 09/27/99 UM, 4.88–5.18 m 5.4 Yes V 12 10/11/99 UM, 4.88–5.18 m 5.4 Yes V Niche 4, DTN: LB0010NICH4LIQ.001 [DIRS 153145] 40 11/03/1999 UL, 7.62–7.93 m 5.5 Yes V 41 11/30/1999 UL, 7.62–7.93 m 3.1 Yes C 42 01/24/2000 UL, 7.62–7.93 m 0.5 No V 43 06/26/2000 UL, 7.62–7.93 m 1.2 Yes C 44 11/16/1999 UM, 6.10–6.40 m 5.5 Yes V 45 12/10/1999 UM, 6.10–6.40 m 2.3 Yes C 46 02/09/2000 UM, 6.10–6.40 m 0.5 – a – a 47 03/14/2000 UM, 6.10–6.40 m 0.5 No V 48 06/08/2000 UM, 6.10–6.40 m 1.2 Yes C 49 12/07/1999 UR, 5.18–5.48 m 5.5 Yes V 50 01/05/2000 UR, 5.18–5.48 m 2.4 Yes C 51 02/14/2000 UR, 5.18–5.48 m 0.5 Yes C SYBT-ECRB-LA#2, DTN: LB00090012213U.002 [DIRS 153154] 52 05/11/2000 LA#2, zone 1 >450 Yes V 53 05/17/2000 LA#2, zone 1 34.9 Yes V 54 05/23/2000 LA#2, zone 1 26.3 Yes V 55 05/23/2000 LA#2, zone 2 29.5 Yes V 56 06/01/2000 LA#2, zone 2 31.6 Yes V 57 05/17/2000 LA#2, zone 3 16.8 No V 58 05/23/2000 LA#2, zone 3 26.1 No V 59 06/01/2000 LA#2, zone 3 35.6 No V 60 06/14/2000 LA#2, zone 3 37.8 Yes V SYBT-ECRB-LA#2, DTN: LB0110SYST0015.001 [DIRS 160409] 61 10/23/2000 LA#2, zone 2 33.0 Yes C 62 11/27/2000 LA#2, zone 2 35.3 Yes C 63 10/23/2000 LA#2, zone 3 38.0 Yes C 64 11/27/2000 LA#2, zone 3 40.8 Yes C SYBT-ECRB-LA#1, DTN: LB0110ECRBLIQR.002 [DIRS 156879] 65 02/28/2001 LA#1, zone 2 17.0 No C 66 04/03/2001 LA#1, zone 2 41.2 Yes C 67 04/09/2001 LA#1, zone 2 43.9 Yes C 68 04/17/2001 LA#1, zone 2 44.5 Yes C 69 04/25/2001 LA#1, zone 2 43.1 Yes C SYBT-ECRB-LA#3, DTN: LB0203ECRBLIQR.001 [DIRS 158462] 70 05/17/2001 LA#3, zone 1 36.4 No C 71 05/23/2001 LA#3, zone 1 24.7 Yes C 72 05/17/2001 LA#3, zone 2 71.2 No V 73 06/20/2001 LA#3, zone 2 31.2 No V 74 07/05/2001 LA#3, zone 2 65.7 No V 75 07/13/2001 LA#3, zone 2 47.9 No V 76 07/16/2002 LA#3, zone 2 32.4 No V 77 05/17/2001 LA#3, zone 3 0.0b No – b Niche 5, DTN: LB0207NICH5LIQ.001 [DIRS 160408] 78 05/06/2002 #5, 28–29 ft 72.0 Yes V 79 05/06/2002 #2, 21–22 ft 120.0 No – c 80 05/17/2002 #2, 21–22 ft 120.0 No – c 81 05/16/2002 #5, 28–29 ft 60.0 Yes V 82 05/21/2002 #5, 28–29 ft 72.0 Yes – d Niche 5, DTN: LB0209NICH5LIQ.001 [DIRS 160796] 83 07/17/2002 #3, 21–22 ft 55.0 Yes – e, f 84 07/29/2002 #3, 21–22 ft 33.0 Yes – f 85 08/14/2002 #3, 21–22 ft 9.0 Yes – g 86 07/15/2002 #5, 28–29 ft 25.8 Yes C 87 07/31/2002 #5, 28–29 ft 25.8 Yes V 88 08/05/2002 #5, 28–29 ft 11.3 Yes V Niche 5, DTN: LB0211NICH5LIQ.001 [DIRS 160792] 89 09/17/2002 #4, 10–11 ft 9.9 Yes C 90 10/01/2002 #4, 10–11 ft 4.8 No V a Events 5 and 46: Potential seepage could not be determined because of an equipment failure b Event 77: Return flow from injection interval indicates that no water was released to the formation c Events 79 and 80: Packer problem d Event 82: Seepage partially bypassed capture system e Event 83: Pump problem f Events 83 and 84: Test interference (see discussion of Figure 6-17) g Event 85: Test too short for analysis UL=upper left; UM=upper middle UL=upper left; UM=upper middle 6.5.4 Relative Humidity and Evaporation Rate Measurements Reduced relative humidity in the ESF Main Drift, the ECRB Cross-Drift, and the niches lead to partial evaporation of the water that reaches the opening, effectively reducing seepage. Note that a conservative treatment of a process in a forward model is nonconservative in an inverse model (and vice versa). Specifically, neglecting evaporation effects in a seepage prediction model (forward model) leads to higher seepage rates and is thus conservative. However, an overestimation of seepage in a model used for parameter determination (inverse model) is compensated by an increase in the estimated capillary-strength parameter, which is nonconservative if this parameter is subsequently used for seepage predictions. Following the recommendations made to address the evaporation issue [Seepage Calibration Model and Seepage Testing Data (CRWMS M&O 2001 [DIRS 153045], Section 7.5)], humidity in the closed-off Niches 3 and 4 was artificially increased to reduce the evaporation potential, and relative humidity was monitored (BSC 2004 [DIRS 170004], Figures 6-27 and 6-28). In the systematic testing area, additional curtains were installed on the two ends of the V-shaped seepage capture PVC curtains to reduce air circulation in the ventilated ECRB Cross-Drift (after June 2000). In addition, relative humidity and evaporation rates from an open pan were measured (see, for example, In Situ Field Testing of Processes (BSC 2004 [DIRS 170004], Figure 6-139). Relative humidity and evaporation rate were also measured in Niche 5 (DTN: LB0207NICH5LIQ.001 [DIRS 160408] and DTN: LB0211NICH5LIQ.001 [DIRS 160792]). The evaporation-rate data will be used to estimate the thickness of the diffusive boundary layer (see Section 6.6.1.4). The relative-humidity data will be applied as a time-dependent boundary condition determining the water potential in the opening. 6.6 MODEL FORMULATION 6.6.1 Mathematical Model The mathematical model for unsaturated flow is based on the conceptual model outlined in the previous sections. The basic theoretical foundation for unsaturated flow in a continuum is outlined first, with a short discussion of the capillary pressure curve and its relevance for seepage (Sections 6.6.1.1 and 6.6.1.2). The incorporation of evaporation from a wetted porous surface is described in Sections 6.6.1.3 and 6.6.1.4. Section 6.6.1.5 contains a summary description of the inverse modeling methodology. 6.6.1.1 Unsaturated Flow Flow in unsaturated porous or fractured media is described by the rate of change in liquid saturation and the flow rate at any given point. The continuum concept (see Section 6.3.2) stipulates the following equation of continuity, which describes the rate at which liquid saturation changes at a given point (Bear 1972 [DIRS 156269], pp. 496, Eq. 9.4.39): (Eq. 6-1) Here, [s] is time, [dimensionless] is porosity, [dimensionless] is liquid saturation, [kg m-3] is liquid density, and [kg m-2 s-1] is the flow rate along the principal axes (, , and ). Considering that liquid flow is driven by gravity and pressure gradients (see Section 6.3.2), the liquid-flow rate is described by the Buckingham-Darcy law as follows (after Bear 1972 [DIRS 156269], pp. 487–488, Eqs. 9.4.20 and 9.4.21): , , and (Eq. 6-2) Here, [m2], [m2], and [m2] are the absolute permeabilities along the three coordinate axes [m], [m], and [m] (where is positive upward), [dimensionless] is relative permeability, [Pa·s] is liquid viscosity, [m s-2] is gravitational acceleration, and [Pa] is the capillary pressure defined as the difference between the liquid and gas pressure. Substituting Eq. 6-2 into Eq. 6-1 leads to the governing equation of flow in unsaturated porous media (after Bear 1972 [DIRS 156269], p. 496, Eq. 9.4.41): (Eq. 6-3) In Richards’ equation, the relative permeability () and capillary pressure () are functions of liquid saturation as given, for example, by van Genuchten’s model (after van Genuchten 1980 [DIRS 100610], after Eqs. [8] and [3]): (Eq. 6-4) (Eq. 6-5) In van Genuchten’s equations, the effective saturation, , is defined as (Eq. 6-6) where is residual liquid saturation, and [Pa] and [dimensionless] are fitting parameters. The roles of the parameters in the capillary pressure and relative permeability functions are illustrated in Figure 6-5. The parameter describes the point of inflection in the capillary-pressure function (Eq. 6-5) shown in Figure 6-5a. The factor scales the capillary pressure curve and is therefore referred to as the capillary-strength parameter. The parameter determines the slopes of the capillary pressure and relative permeability functions. It is a measure of the spread of the effective pore size distribution; a large m value implies a narrow pore size distribution. The use of continuous relative-permeability and capillary-pressure functions, which apply to porous media, is considered appropriate also for small fracture segments that are rough-walled and/or partially filled with porous material. 0.4 iveSatuS Figure 6-5. (a) Capillary-Pressure Curves and (b) Relative-Permeability Curves for Different Illustrative van Genuchten Parameters. 6.6.1.2 Onset of Seepage For a circular opening in a homogeneous medium, the threshold for liquid entry into the cavity is full saturation at the apex (Philip et al. 1989 [DIRS 105743]). For the liquid that enters the opening to form a drop at the opening wall and detach (see definition of seepage in Section 6.1.2), a positive pressure that offsets the drop pressure is required (Or and Ghezzehei 2000 [DIRS 144773], pp. 390-392). For a numerical model in which the continuum is subdivided into discrete gridblocks, the condition for seepage is determined by the total water-potential gradient at the connection between the fractured medium and the opening as depicted in Figure 6-6. From Eq. 6-2 it follows that downward seepage in a discrete numerical mesh, , occurs only when the following condition is satisfied: 0>.+ (Eq. 6-7) where is the capillary pressure at the last node adjacent to the opening. Given that the capillary pressure in the opening is zero, the numerical threshold capillary pressure is defined as , where is the distance between the last node and the opening. The numerical threshold capillary pressure therefore depends on the nodal distance between the last node and the opening. The opening surface does not need to be fully saturated for seepage to commence as given by the analytical solutions of Philip et al. (1989 [DIRS 105743]). As indicated in Figure 6-6, given a numerical grid, the seepage-threshold liquid saturation is lower for larger and for lower capillary strength (). Consequently, parameter determines whether liquid that reaches the surface seeps or is diverted around the opening (effectiveness of the capillary barrier). Note that the relative permeability function (Eq. 6-4) does not depend on . Hence, the capillary-strength parameter is the main subject of the SCM presented in this report (see also discussion in Section 6.6.3.1). The fact that the seepage threshold depends on the length of the nodal distance to the opening makes the values of the estimated capillary-strength parameter () applicable only to numerical models of comparable discretization (see Point 0 of Section 8.4). Figure 6-6 shows that a reasonable variation in the parameter has only a limited effect on the seepage threshold saturation; a stronger effect is seen for a change in , which tends to vary more than . Therefore, fixing the parameter appears reasonable as confirmed by the formal sensitivity analysis (see Section 6.6.3.1 below). Moreover, any potential variability of is accounted for in the calibrated parameter. The relative sensitivity and potential identifiability of seepage-relevant parameters are further discussed in Section 6.6.3.1. a perm.par$j xSisim << eof perm.par$j # Unix shell script file to generate TOUGH2 mesh echo Start shell script sh.LAx_zoneY_mesh echo Date : `date` i=0 # generate new seed number echo "Run $j of $runs" # nodal distance to b 1- erated: LAx_zoneY.mes$j ========================= pleted: `date` B 3AT5 1 10.1000E-090.5000E-010.3086E-010.9659E+00 B 4AT6 1 10.1000E-090.5000E-010.1549E-010.9659E+00 B 5AS6 1 10.1000E-090.5000E-010.2228E-010.9659E+00 B 6AS7 1 10.1000E-090.5000E-010.2394E-010.9659E+00 B 7AR7 1 10.1000E-090.5000E-010.1383E-010.9659E+00 B 8AR8 1 10.1000E-090.5000E-010.3239E-010.9659E+00 B 9AQ8 1 10.1000E-090.5000E-010.5377E-020.9659E+00 B 10AQ9 1 10.1000E-090.5000E-010.3004E-010.9659E+00 B 1B 2 20.1578E+000.2888E-010.1140E-02-.2588E+00 B 3B 4 20.1289E+000.6470E-010.1140E-02-.2588E+00 4B 5 20.6470E-010.9307E-010.1140E-02-.2588E+00 5B 6 20.9307E-010.1000E+000.1140E-02-.2588E+00 6B 7 20.1000E+000.5777E-010.1140E-02-.2588E+00 B 9B 10 20.2246E-010.1255E+000.1140E-02-.2588E+00 BOR 0B 1 30.1000E-090.1905E-010.3777E-010.1000E+01 0B 2 30.1000E-090.1905E-010.6913E-020.1000E+01 0B 3 30.1000E-090.1905E-010.3086E-010.1000E+01 0B 4 30.1000E-090.1905E-010.1549E-010.1000E+01 BOR 0B 6 30.1000E-090.1905E-010.2394E-010.1000E+01 BOR 0B 7 30.1000E-090.1905E-010.1383E-010.1000E+01 BOR 0B 9 30.1000E-090.1905E-010.5377E-020.1000E+01 BOR 0B 10 30.1000E-090.1905E-010.3004E-010.1000E+01 42113 55 42113 56 42113 115 42113 116 42113 175 42113 176 42113 235 42113 236 42113 295 42113 296 42113 355 42113 356 42113 415 42113 416 42113 475 ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... 42109 42110 42111 42101 42111 42102 42111 42103 42111 42104 42111 42105 42111 42106 42111 42107 42111 42108 42111 42109 42111 42110 Figure C-6. Excerpt from Sample Mesh File LAx_zoneY.mesZ INTENTIONALLY LEFT BLANK MESH GENERATION FOR SIMULATION OF SEEPAGE TESTS Four 3-D meshes for the simulation of liquid-release tests in niches located in the middle nonlithophysal zone were generated: two 1.5 m long sections of Niche 2 centered (a) 4.42 m and (b) 5.64 m from the collar of borehole UM, (c) a 1.5 m long section of Niche 3 centered at 0.25 m, and (d) a 2.0 m long section of Niche 4 centered at N The meshes were created in several steps as follows (where information for each mesh is preceded by the letter referring to a specific panel of Figure 6-16): 1. Primary meshes were generate × 5.0 m, and (d) 6.0 m × 2.0 m × 5.0 m, respectively, iche 2. The X-, Y-, an Z-coo dinates of each primary m (d), and Z = 0 is (a MoveMesh V1.0 Y = 0 coincides w LBNL ith Nich 000 [DIRS 152824]) so thtion 00 + 0.0 m for Mesh the bottom of the esh an (c & d) coincides with the local 3. The heterogeneous perroutine Perm2Mesh V1.0 (LBNL 2000 [DIRS 152826]). m ields were mapped onto their r espective meshes using 4. (a & b) A niche with vertical walls at X = -2 m and X = with the crown at (a) Z = 3.13 m and (b) Z = 3.33 m was cut from the mesh using software CutNiche V1.3 (LBNL 2000 [DIRS 152828]). (c & d) Niches with vertical walls at (c) X = -2.15 m and X = 2.35 m and (d) X = -2.00 m and X = 1.90 m and ceilings defined by survey data are cut from their respective meshes using software CutNiche V1.2 (LBNL 2000 [DIRS 152815]). A very small nodal distance e drift surace and the gridblocks denoting the drift, wh locks rep formatio terface de Software AddBound V1.0 (LBNL 2000 [Delements at the top and bottom of the model dom assigned to a special material domain to allow specifying a free-drainage boundary condition. 6. Gridblocks along the alignment of the injection boreholes were modified to represent 1 ft (0.3048 m) long injection intervals with a diameter of 3 inches (0.076 m). The final meshes (see Figure 6-16) conta 340 gridblocks and 108,000 c in various SNs (Finsterle 1], pp. 27–29, 42–44, 54; H999 [DIRS 153449], pp. 108–123; and Trautz 2001 [DIRS 156903], pp. 35–45). FT BLANK AP IX E ES ERATI OR SEE TEST ULA NS I sur oughness reproduc interp n fro y da Table E-1 s for the ng of N 5 DIRS ption Coo te Sys 003 Niche 5 p rvey data 155370] 010 Niche 5 sl y data ada S lane 155369] ILN 161733] Niche 5 suted bottom ta for colntervals LNG 161733] e 5 detai e surve e from referenc ent DTN=Data g Numbe follo n preparing th e interpo Niche 5 ceiling profiles are described below. poi the mesh s selecte e the in tion o enter ble E-2. m of Niche e Nevad rdina stem orthi Elev uth D [m [m] ELD A rees] 62.51 110 DTN: 9GSC0033 370 survey d DTN: M 00 [D o a regul ZY- inate em us the fo tion of the computational meshes for simulating liquid-release tests performed in Nimajor steps: ration of niche ceiling coord E PREPARA ION OF NICHE CEILING COORDINATES ()()()().·-+.·-=sincosDDNNESESX (Eq. E-1) ()()()().·-+.·-=sincosDESESNNY (Eq. E-2 A-°=.360 Dat d Thei dina Nor Elev X [m 23 11 -4.8 23 11 -2.8 23 11 23 11 0.00 23 11 0.00 23 11 0.00 11 0.00 23 11 0.00 23 11 0.00 23 11 11 3.03 23 11 3.95 23 11 -2.6 23 11 23 11 0.00 23 11 0.00 23 11 2.59 11 -2.7 23 11 2.66 23 11 0.00 23 11 0.00 11 0.00 233262.77 1108.62 2.51 13.59 0.00 13.98 233262.42 1110.09 233262.39 1106.57 0.00 14.01 233262.21 1108.48 -2.26 2323 11 2.04 11 0.00 23 11 0.00 23 11 2.03 23 11 23 11 0.00 23 11 0.00 23 11 2.08 11 -1.8 23 11 -0.0 23 11 -0.0 rofile Survey Data and Slot Survey Data and TZ Coordinate System (Conti e 5 Pro Nor Elev 23 11 2.13 23 11 -1.8 23 11 1.97 23 11 23 11 0.00 23 11 0.00 23 11 -2.4 11 -2.0 23 11 -0.0 23 11 -0.0 23 11 23 11 -2.0 233247.45 1110.10 -0.01 233247.45 1106.88 -0.01 233247.21 1108.73 0.00 29.19 e 5 Slo Nor Elev X [m 1 -2.0 23 1 -2.4 23 1 -2.8 23 1 -3.1 23 1 -3.3 23 -3.5 23 1 -3.5 23 1 23 1 -3.0 23 -2.7 23 1 -2.3 1 -2.0 23 1 -1.9 23 1 -1.9 23 1 -2.0 23 1 -2.5 23 -2.7 23 1 -3.0 23 1 -3.4 23 1 -3.7 23 -3.8 23 1 -3.8 23 1 -3.4 23 1 -3.2 23 1 -3.0 23 1 -2.42 Nor X [m 23 1 -2.2 23 1 -1.9 23 1 -2.2 23 -2.6 23 1 -3.2 23 1 -3.4 23 1 -3.7 23 1 -3.6 23 1 -3.5 23 1 -3.3 23 -3.0 23 11 -2.4 233254.48 1107.45 1109.11 -2.05 233253.96 -2.04 232 1 -2.2 1 -2.5 23 1 -2.9 23 1 -3.1 23 1 -3.6 23 1 -3.9 23 -4.0 23 1 -3.6 23 1 23 1 -3.2 23 -2.9 23 1 -2.5 23 1 -2.3 23 1 -2.2 23 1 -2.1 23 1 -2.1 23 1 -2.1 23 1 -2.2 23 1 -2.1 23 1 -2.2 23 1 -2.0 23 1 23 1 -2.0 23 1 -2.0 23 1 -2.1 23 1 -2.1 23 1 -2.3 1 -2.2 223 -2.5 23 11 -3.0 Nor X [m 23 1 -3.5 23 -2.0 23 -2.1 23 1 -2.5 23 1 -3.0 23 1 -3.2 23 1 -3.1 23 1 -2.7 23 1 -2.1 23 1 -2.0 23 1 -2.0 23 -2.1 233257.32 1 1108.38 -2.55 233257.16 1108.42 -2.69 23 1 -3.1 23 1 2.0 23 1 2.3 23 1 2.5 23 1 2.8 23 1 2.8 23 1 2.4 1 23 2.1 23 1 2.2 23 1 2.3 23 1 2.6 1 2.7 23 1 2.8 23 1 2.7 23 1 2.4 23 1 2.4 23 1 2.3 32.00 5370], 0106 .98 a Table E-4. Niche 5 Ceiling Roughness Data X [m] Y [m] Z [m] -1.40 27.86 -1.09 27.86 -0.17 27.8 0.14 27.80.45 27.8 0.76 27 1.06 271.37 27 1.68 27 -1.40 26 -1.09 26 -0.78 26.56 4.36 1.37 26.56 4.20 Table E4. Niche 5 Ceiling Roughness Data (Continued) X [m] Y [m] Z [m] -0.78 22.66 4.32 -0.48 22.66 4.42 -0.17 0.45 22.66 4.71 1.3 .66 4.26 -1.40 -1 36 4.13 -0 36 -0 36 4.35 4.33 0 36 0.76 1 36 4.19 1 36 -1 06 3.90 4.11 -1 -0 06 0 06 0 06 4.31 11.37 06 -1 76 3.96 3.89 -1 76 -0.48 18.76 4.10 -0.17 18.76 4.09 0.14 18.76 4.07 0.45 18.76 4.08 0.76 18.76 4.08 1.06 18.76 4.13 1.37 18.76 4.20 1.68 18.76 3.79 -1.40 17.46 4.00 -1.09 17.46 3.91 -0.78 17.46 3.93 7 14.87 8 14.87 0 15.19 0 15.51 0 16.49 0 16.81 17.14 17.79 18.11 18.44 19.09 19.41 0 19.74 0 20.39 20.71 0 0 0 21.69 0 21.04 Table E4. Niche 5 Ceiling Roughness Data (Continued) -0.78 16.16 4.06 -0.416.16 4.14 -0.17 16.16 4.12 0.14 16.16 4.09 0.45 16.16 4.18 0.76 16.16 4.09 3.92 3.77 1.40 9 14.87 14.87 14.87 14.87 14.87 14.87 14.87 14.87 26.24 (Continued) Z [m] 1.40 1.4 0 22.34 22.99 23.31 0 23.64 0 24.29 24.61 24.94 26.89 27.21 27.54 8 15.19 8 15.51 8 15.84 8 8 16.81 8 17.14 14.87 1 16.16 5 18.76 7 20.06 8 21.36 5 7 23.96 5 25.26 0 25.91 16.49 9 22.66 1 26.56 3 14.87 4 16.16 9 3 18.76 4 20.06 8 23.96 25.26 2 26.56 9 27.86 0 0.14 1 27.86 17.46 1 27.86 0.000000E+ Excerpts of the resulting Niche 5 ceiling and slot profiles are shown in Figure E-1, and Figure E-2 shows the plan view of the ceiling roughness. Niche 1620 ceiling 00E+0010.000000E+000 00E+001 4.222630E+000 E+001 4.225305E+000 2223 21 Y[m] 181920Z[m] 4.54.2 16173.4 -4-3-2-1012314 put DTN: LB0302SCMREV02.002. ble E-3 and Table E-4 were interpolated 152.82.5 E2. LOCATION OF BOREHOLES AND PREPARATION OF GEOSTATISTICAL PARAMETERS OF AIR-PERMEABILITY [DIRS 169532]) were transformed from the Nevada coordinate system to the regular ZYX-- rdinate system using Eqs. E-1 to E-4. The original borehole surveys and their transformed Table E-5. Original and Transformed Coordinates of Borehole Collars and Projected Bottoms rehole Collar/Bottom Easting (m) Northing (m) Elevation (m) Depth (m)X (m) Y (m) Z (m) Projected Bottom 170662.83 233246.93 1111.61 -0.83 29.48 5.70 ECRB-NICHE1620 #3 Collar #3 170662.25 233262.72 1111.00 0.02 13.68 5.09 Projected Bottom 170661.98 233247.23 1111.29 15.50 0.02 29.17 5.38 ECRB-NICHE1620 #4 Collar #5 170663.27 233262.84 1111.42 -1.00 13.58 5.51 rojected Bottom 170662.87 233247.08 1113.25 15.88 -0.87 29.33 P ECRB-NICHE1620 #6 Collar #6 ECRB-NICHE1620 #7 Projected Bottom 170661.17 233248.01 1113.27 14.81 0.84 28.37 7.36 easured by air-injection tests conducted in boreholes #2, #3, and #5 (see Collar #4 170661.26 233262.76 1111.04 1.01 13.63 5.13 Projected Bottom 170661.16 23315.02 Projected Bottom collars. These distances were first converted t ing ZYX-- coordinates by the following elementa ()()(22zyyxx-'+''-'+''-' yydyY'-'' +'= ()()()222zzyyxx''-'+''-'+''-' ()zzdzZ'-'' +'= (Eq. ()()()zzyyxx''-'+''-'+''-' ()'''()''''''ectively (see of the borehole are denoted by zyx,, and zyx,,, resp Table E-6. DTN: LB0110AKN5POST.001 [DIRS 156904] Calculated Start End 2Midpoint Midpoint 0 3 4 2.70E-12 3.5 1.07 -0.99 14.26 5.63 0 4 5 5.62E-12 4.5 1.37 -0 0 5 6 5.48E-09 5.5 1.68 -0.98 14.86 5.7015.16 5.73 0 6 7 4.27E-09 6.5 1.98 -0.98 0 7 8 4.08E-12 7.5 2.29 -0.98 15.47 5.7 0 9 10 7.77E-12 9.5 2.90 -0.97 16.07 5.8 16.38 5.87 0 10 11 3.30E-12 10.5 3.20 -0.97 0 11 12 2.79E-11 11.5 3.51 -0.97 16.68 5.9 0 12 13 3.83E-11 12.5 3.81 -0.97 16.98 5.94 0 13 14 1.65E-10 13.5 4.11 -0.97 17.28 5.9 0 14 15 1.82E-10 14.5 4.42 -0.96 17.59 6.01 4.72 -0.96 0 15 16 2.35E-11 15.5 17.89 6.04 3 4 5 1.61E-11 4.5 1.37 -0.96 14.57 5.135.14 3 5 6 3.18E-12 5.5 1.68 -0.96 14.87 3 6 7 1.56E-11 6.5 1.98 -0.96 15.18 5.1615.48 5.17 3 7 8 1.47E-12 7.5 2.29 -0.96 3 8 9 4.08E-10 8.5 2.59 -0.95 15.79 5.1 16.09 5.19 3 9 10 6.23E-10 9.5 2.90 -0.95 3 10 11 6.24E-10 10.5 3.20 -0.95 16.40 5.2 3 11 12 5.52E-10 11.5 3.51 -0.94 16.70 5.21 -9.2581 3 113 1.19E-12 12.5 3.81 -0.94 17.01 5.22 -11.9245 [DIRS 156904] BH Start End k [m2] Mi [ft] [ft] 3 15 16 2.23E-12 3 16 17 4.03E-09 3 17 18 1.92E-09 4 3 4 5.85E-09 4 4 5 9.51E-09 4 5 6 9.32E-12 4 6 7 8.85E-12 4 8 9 4.16E-13 4 9 10 1.87E-12 4 10 11 1.16E-13 4 11 12 4.87E-14 4 12 13 5.25E-13 4 13 14 2.20E-11 4 14 15 3.66E-11 4 15 16 4.82E-14 4 16 17 5.91E-13 4 18 19 3.71E-11 4 19 20 8.39E-13 4 20 21 2.48E-12 4 21 22 1.82E-12 4 22 23 1.86E-13 4 23 24 2.33E-13 4 24 25 2.65E-12 4 25 26 2.14E-12 4 26 27 2.11E-13 4 27 28 2.95E-13 4 28 29 6.71E-11 4 29 30 6.87E-11 4 30 31 1.64E-11 4 31 32 7.19E-12 4 32 33 2.43E-12 4 34 35 1.06E-12 4 35 36 1.57E-12 BH=borehole; DTN=Data Tracking Numb Multiple numerical meshes of a 2 m long section of the Niche 5 w different stochastic realization of the underlying heterogeneous perm three test zones, labeled Niche5a, Niche5b, and Niche5c. The locati of these meshes are listed below in Table E-8. Table E-8. Primary Dimensions of Niche 5 Meshes Locaax 0.0 + 15.60 0.0 + 19.60 0.0 + 21.90 ollowing mesh generation steps were perform The f mesh of borehole #5 (28-29 ft). 1. A mesh was ge nerated with X-Y-Z dimensions as listed in Table E-8, discre × 0.1 m × 0.1 m. The Y-axis was aligned Niche centerline. Figure E-3 sh gen erate the mesh. 2. The mesh was shifted using software Movtranslate the origin of the mesh to the datum of 3. The GSLIB module SISIM V1.204 (LBNL 2random, spatially correlated field of log-perm seed number was inserted into the SISIM V le measured_log-k_12_N5.dat (see Figure E-5). An excerpt of the meability field is shown in Figure E-6. 8. Gridblock (0.3 m) long injection intervals and 3 ft (0.9 m) long packers on both sides of the injection interval. 9. Drift elements (DRI78, DRI79, DRI88, volume so Dirichlet boundary conditio large represents seep 10. Six new evaporation eleadded and connected to the sam distance from the formation elements to the ev boundary-layer thickness. Flux into these elements represents eva 11. A single time step was performed using a generic TOUGH2 input iTOUGH2 V5.0 (LBNL 2002 [DIRS 160106]); s obtain cross-referencing information. The execution file Steps 1-11 listed above were The script file sh.N5BH5_28 -29ft_run (see Figure E-10) assigns new seed numneration of permeability field and generates multiple m ge sh.N5B sto The final mesh is underlying random permeability field. TOUGH2 input file for generating 3D grid for Niche 5 MESHMAKER ----*----2----*----3----*----4----*----5----*----6----*----7----*----8 XYZ NX NY 20 0.1000000 NZ 1 0.600E-10 50 0.1000000 NZ NZ 1 0.600E-10 ENDFI ---1-- Figure E-3. Input File N5BH5_28-29ft Used to Generate Primary Mesh Parameters for SISIM ******************** Niche 5 Borehole #5 (28-29 ft) TAG July 22, 2002 START OF PARAMETERS: measured_log-k_12_N5.dat 1 2 3 4 \column: x,y,z, -1.0e21 1.0e21 \data trimming l -2.0 5.0 \minimum and maximum data value 1 2.5 \lower t 1 1.0 \middle 4 dum 3 0 \column for variable, weight direct.ik \direct input of indi N5BH5_28-29ft_airK.dat \output file for 2 N5BH5_ 0 \0=standard order relation corr 59069 1 80 -4.45 0.10 20 50 2.05 0.10 1 0 2.0 0.0 0.0 0 0 20 \ min, max data for simulation 12 \number simulated nodes t0 2.5 \0=full IK, 1=med approx(c 0 \0=SK, 1=OK 8 \number cutoffs -0.725 0.066 1 0.02 \cutoff, global cdf, nst, nugget 1 0.96 1.82 \ 0.0 0.0 0.0 1.0 1.0 \ ang1,a-0.050 0.197 1 0.02 \cutoff, global 1 0.96 1.82 \ it, aa, 0.0 0.0 0.0 1.0 1.0 \ ang1,ang2,ang3,anis1,2 0.625 0.443 1 0.02 \cutoff, 0.0 0.0 0.0 1.0 1.0 \ ang1,ang2 1.300 0.623 1 0.02 \cutoff, global cdf 1 0.96 1.82 \ it, aa, cc 0.0 0.0 0.0 1.0 1.0 \ ang1,ang 1.9 7 1 0.96 1.82 0.0 0.0 0.0 1.0 1 2.650 0.852 1 0.02 \cutoff, global cdf, nst, nugget 1 0.96 1.82 0.0 0.0 0.0 1.0 1.0 \ ang1,ang2,ang3,anis1,2 3. 32 1 0.96 1.82 0.0 0.0 0.0 1.0 1 4.000 0.999 1 0.02 1 0.96 0.0 Figure E-4. Input File Pa Generating Software SISIM N5-air K data 4 y z log-k+12 8851 14.25626 5.63022 0.43136 -0. -0. 983-0.981 48 15.16481 5.73326 3.6-0.97913 15.46766 5.76761 0.6 -0.97444 16.07336 5. -0.97210 16.37621 5.87065 0.51851 -0.96976 16.67906 5.90500 1.44560 -0.96741 16.98191 5.93934 1.58320 -0.96507 17.28476 5.97369 2.21748 -0.96272 17.58761 6.00804 2.26007 -0.96038 17.89046 6.04239 1.37107 -0.96330 14.56905 5.13219 1.20683 -0.96067 14.87362 5.14378 0.50243 -0.95804 15.17818 5.15538 1.19312 -0.95541 15.48275 5.16698 0.16732 -0.95278 15.78732 5.17858 2.61066 -0.94751 16.39645 5.20177 2.79518 -0.94488 16.70102 5.21337 2.74194 -0.9493 -0. -0.93435 17.91929 5.25976 0.348301 -0.93172 18.22386 5.27135 3.6053.28295 3.2833 -0.92909 18.52843 5 0.02100 14.69933 5.11511 3.97818 0.02100 15.00408 5.12069 0.96942 0.02099 15.30883 5.12627 0.94694 0.02098 15.61358 5.13185 0.98588 0.02097 15.91833 5.13743 -0.38091 0.02097 16.22307 5.14301 0.27184 0.02095 16.83257 5.15417 -1.31247 0.02095 17.13732 5.15975 -0.27984 0.02094 17.44207 5.16533 1.34242 0.02093 17.74682 5.17091 1.56348 0.02092 18.05157 5.17649 -1.31695 0.02092 18.35632 5.18207 -0.22841 091 18.66107 5.18765 1.12710 0.0202 0. 0.02089 19.27056 5.19881 -0.07624 0.02089 19.57531 5.20439 0.39445 97 0.26007 0.02088 19.88006 5.20902087 20.18481 5.215 0. 0. 0. 020 0.020 0.02084 21.40381 5.23787 -0.67572 0.02084 21.70855 5.24345 -0.53018 0.02083 22.01330 5.24903 1.82672 1 1.83696 0.02082 22.31805 5.2546 0.02081 22.62280 5.2601 0.02081 22.92755 5.2657 0.02080 23.23230 5.2713 0.02079 23.53705 5.2769 0.02079 23.84180 5.2825 0.02078 24.14655 5.2880 0.02077 24.45129 5.2936 Figure E-5. Measured Conditioning variables = x y z var zon e -0.44 50000E+01 0.2095000E+02 0.2050000E+01 0.685333 -0.4350000E+01 0.2095000E+02 0.2050000E+01 0.1218344 -0.4250000E+01 0.2095000E+02 0.2050000E+01 0.714998 ... 0.3250000E+01 0.2285000E+02 0.6950000E+01 0.2682163E+01 0.3350000E+01 0.2285000E+02 0.6950000E+01 0.3089783E+01 0.3450000E+01 0.2285000E+02 0.6950000E+01 0.2990684E+01 Figure E-6. Excerpt from the Generated Permeability Field Niche 1620 left batwing x y z -4.450000E+000 1.455000E+001 0.000000E+000 -4.350000E+000 1.455000E+001 0.000000E+000 -4.250000E+000 1.455000E+001 0.000000E+000 .... -2.850000E+000 1.965000E+001 3.026709E+000 -2.650000E+000 1.965000E+001 2.910188E+000 -2.550000E+000 1.965000E+001 2.883803E+000 .... -1.650000E+000 2.355000E+001 0.000000E+000 -1.550000E+000 2.355000E+001 0.000000E+000 -1.450000E+000 2.355000E+001 0.000000E+000 Figure E-7. Excerpts from Interpolated Left Slot Profile of Niche 5 Niche 1620 right batwing data x y z 2.050000E+000 1.455000E+001 0.000000E+000 2.150000E+000 1.455000E+001 0.000000E+000 2.250000E+000 1.455000E+001 0.000000E+000 2.250000E+000 2.005000E+001 2.991148E+000 2.350000E+000 2.005000E+001 2.904599E+000 2.450000E+000 2.005000E+001 2.835122E+000 2.750000E+000 2.355000E+001 0.000000E+000 2.850000E+000 2.355000E+001 0.000000E+000 2.950000E+000 2.355000E+001 0.000000E+000 Figure E-8. Excerpts from Interpolated Right Slot Profile of Nich #! /bin/sh # # Unix shell script file sh.N5BH5_28-29ft_mesh # Usage: sh.N # # # Generates TOUGH2 mesh N5BH5_28-29ft.mes # uses # airK_N5_3.dat ceiling_N5_3.dat # # l # righ # 1 # TA Ghezzehei (Sept 10, 2002), Version # # modified from S. Finsterle, August 6, 1999, Ve # echo cript sh.N5B echo ' Start shell scho ' mesh generator e echo '= echo # echo echo 1. Generate 3d mesh echo -------------------- ft 9 # general tough2 -mesh N5BH5_28-29 # echo echo 2. Center mesh echo -------------- esh << eof xMoveM N5BH5_28-29ft.mes temp01.mes -4.50 20.90 7.00 eof # echo echo 3. Map correlated permeability fiecho ------------------------------------ cei 2 .0 100 1.0 1.0 # cosine multiplicatio -2.00 # Xmin # Xmax 2.10 80 20. 23.00 0.0 4.7 eof # eplace all NIC98 and N # R # will be NIC98 and NIC99 (TA Ghezzehei June 1 # echo echo 5. o - ech sed 's/NIC98/NIC78/g' temp04.mes | \ /NIC99/NIC79/g' > temp05.mes sed 's # echo echo 6. Cut out left batwing -- echo ------ temp0 temp06.mes batwing_N5_3.dat left 2 100.0 1.0e-10 1.0 -4.10 -1.90 # Xmax # Ymin N5_3 20.80 0.0 # Zmin 3.6 # Zmax eof # 8 and NIC99 eleme # Replace all NIC9 # left batwing is now NIC88l be NIC98 and NIC99 # wil # o ech echo 7. Replace NIC9* by NIC8* --- echo ---------------------- sed 's/NIC98/NIC88/g' temp06.mes | \ 's/NIC99/NIC89/g' > temp07.mes sed # o ech echo 8. Cut out right batwing ----------------------- echo - xCutNiche1.2 << eof p07.mes tem temp08.mes 0.0 # niche volume emp09.mes # input mesh file emp10.mes # output mesh file # boundary material type sed 's/A5C44A6C44.*/A5C44A6C44 30.0000E-010.5000E-010.1000E-010.1000E+01 /g'| \ sed 's/A4A44A5A44.*/A4A44A5A44 30.5000E-010.0000E-010.1000E-110.1000E+01 /g'| \ sed 's/A4B44A5B44.*/A4B44A5B44 30.5000E-010.0000E-010.1000E-110.1000E+01 /g'| \ sed 's/A4C44A5C44.*/A4C44A5C44 30.5000E-010.0000E-010.1000E-110.1000E+01 /g' \ > temp12.mes # echo echo 13. Edit volume of niche elements echo --------------------------------- sed 's/NICHE........../NICHE0.5000E+52/g' temp12.mes > temp13.mes # # echo echo 14. Add evaporation elements echo ---------------------------- cat temp13.mes | sed -n '1,/BOT99/p' > eleme cat temp13.mes | sed -n '/TOP99A21 1/,$p' > conne grep NIC temp13.mes | sed 's/NIC/EVP/' | sed 's/NICHE/EVAPP/' > elemeconne cat << eof >> eleme NE p -v EVAPP elemeconne | \ ed 's/EVP\(.......\).*0.1000E-090.5000E-01\(..........\).*$/EVP\1 \ echo ---------------------------- echo mv temp14.mes N5BH5_28-29ft.mes rm temp*.* rm t2.msg rm fort* rm *airK.dbg echo echo =========================== # # # Unix script to perform multiple inv # Niche5, Borehole 5 (28-29ft) # # ceiling_N5_3.dat # leftbatwing_N5_3.dat rightbatwing_N5_3b.dat ## # TA Ghezzehei 09/09/2002 # Adapated from S. Finsterle, V1.0, 8/20/02 clear # ing echo Copy air-K and Niche ceil # echo 'Generate meshes for multi 5, 28-29ft)' echo " " i=0 j=0 while test $j -lt $runs do # calculate new seed number j=`expr $j + 1` i=`expr $j + $j` i=`expr $i + 59067` echo "Run $j of $runs" echo "============================================================================" echo "Create \\\seed/" \ > N5BH5_28-29ft_sisim.par$j xSisim << eof N5BH5_28-29ft_sisim.par$j cp 5BH5_28-29ft_airK.d# echo `date`: Mesh generation sh.N5BH5_28-29ft_mesh cp N5BH5_28-29ft.mes N5BH5_28-29ft.mes$j # # echo remove unnecessary files echo rm N5BH5_28-29ft_sisim.pa # ech sh.N5BH5_28-29ft_r APPENDI PREPARATION OF SEEPAGE RATE AND RELATIVE-HUMIDITY DATA FOR THE _sep_193259.txt 9. Translate date and time to seconds after February 28, 2001 13:46: Col. 2 = (RC[-1]-“2/28/2001 13:46”)*86400 10. Save as space delimited text file RH.txt. Open RH.txt using text editor vim. 13. Add first dummy data point (, ) to provide historic relative humidity. Add last dummy data point (1E20, 0.252) to cover potential prediction time frame. Duplicate all 4141 lines twice. Remove second column from Line 2 to Line 4142. Remove first column from Line 4143 to end of file. 19. Add new line “EVA98”, Line 4143. Add new line “EVA99”, Line 8285. 1200.0 2400.0 ......... 45380.0 52 52 46580.0 5247780.0 1.0E20 0.190 0.190 0.190 88 0.1 ..... 0.275 0.260 0.252 0.252 99 EVA 0.190 90 0.1 0.190 0.188 ..... 0.275 0.260 0.252 2 0.25 Figure INTENTIONALLY LEFT BLANK APPENDIX G EXECUTION OF MULTIPLE INVERS Cross-Drift were performed, based on multiple realizations of the underlying permeability field. he following steps were performed ( T 1. Go to the appropriate number, Y represents the injection zone, ed in the same interval). sequences were perform 2. Select the appropriate mesh c a steady-state sim 3. Perform eff input to iTO 4. with evaporation effe 2 2.13×10-5 m/s). A dry-out zone develops around the drift. e inv 5. Take saturation distribution from previous simulation as initial conditions for th A representative TOUGH2 input file LAx_zoneY_setZ (as input to iTOUGH2 V 5. is shown in Figure G-3; an excerpt from a representative iT 2002 [DIRS 160106])) V5.0 (LBNL 2002 [DIRS 160106]) input file L 6. Figure 0106 16 obtained from a single inversion. #! /bin/sh # # sh.run LA ZONE SET [RU # # LA = Boreho # ZONE = Zone number Test set number # SET = S = N # RUN ltiple inversions of seepage data # cd $HOME/ym/Seepage/ST/LA$1/Zone$2/Set$3 # if test "$4" = "" hen t se #