Return to the Main Document

1. INTRODUCTION

A crucial element of the long-term postclosure safety strategy for the potential monitored geologic repository at Yucca Mountain is to contain high-level radioactive waste (HLW), and to keep that waste and its container as dry as possible. There are several degradation processes that could impact the performance of the engineered barrier system (EBS). The role of this PMR is to describe one of the process-level models (Waste Package Degradation) utilized to evaluate these degradation processes. To evaluate the postclosure performance of the monitored geologic repository proposed for construction at Yucca Mountain, a Total System Performance Assessment (TSPA) will be conducted. A set of nine Process Model Reports (PMRs), of which this document is one, is being developed to summarize the technical basis for process-level models supporting the TSPA model. These reports cover the following areas:

These PMRs are supported by several Analysis and Model Reports (AMRs) that contain more detailed technical information. This technical information consists of data, analyses, models, software, and other documentation necessary to defend the applicability of each model for its intended purpose. The PMR is intended to ensure the traceability of waste package degradation information from its various sources through the AMRs, PMRs, and eventually in TSPA.

As described in the Monitored Geologic Repository Project Description Document, the recommended waste package design is Enhanced Design Alternative II (CRWMS M&O 1999a, Section 1.1.2).

The Enhanced Design Alternative II (EDA II) waste package design is the reverse of the Viability Assessment (VA) design. In EDA II, the corrosion-resistant material protects the underlying structural material from corrosion, while the structural material supports the thinner, more expensive corrosion-resistant material. As shown in Figure 1-1, this design includes a double-wall waste package (WP) made from Alloy 22 and 316 nuclear grade (NG) stainless steel. The WP is placed underneath a protective drip shield (DS) made of a titanium-based alloy, as shown in Figure 1.2.

EDA II uses thermal management features (line loading, ventilation, and blending) to limit peak temperatures of cladding, the waste package shell, and the drift wall. This produces more uniform temperature along the drifts, and margin in meeting requirements for cladding integrity, drift stability, and avoidance of phase instability in the waste package materials. Figure 1-2 shows a typical layout of an in-drift placement of three types of waste packages. Note that DSs will be placed over the waste packages just before repository closure to provide protection from dripping water and rock fall.

The purpose of this PMR is to account for the degradation of WP materials under conditions expected at the Yucca Mountain site, which is being evaluated as a possible geologic repository for the disposal of HLW waste. Data pertaining to WP material degradation are presented in supporting AMRs. Functional summaries of the component models and their respective output are provided in Section 1.4. This report was developed in accordance with the technical product development plan Development Plan for Waste Package Degradation Process Model Report (CRWMS M&O 1999b).

Each of the process-level models described in the Waste Package Degradation PMR accounts for a different degradation process, all of which are integrated in the waste package degradation code (WAPDEG) (CRWMS M&O 2000g). Each model was developed using unique methodologies and inputs, with the determination of that model dependent on the functional requirements of the waste package component being represented.

In January 2000, the Project modified the repository design. The changes instituted involve removing backfill from the reference design and reorienting the drifts to minimize the impacts of the rockfall. Preparation of this PMR and the supporting AMRs preceded this design change and do not, therefore, include the impact of that change. However, a qualitative assessment has been made to evaluate the effect of no backfill on the results presented in this PMR and the supporting AMRs. AMRs impacted by the removal of the backfill include those that describe:

At this time, the AMR on the SCC assumes that the DS is protected by backfill from rockfall-induced stress and therefore, no SCC. Removal of the backfill may subject the DS to localized stress due to rockfall, thereby increasing the susceptibility to SCC. The abstraction AMR on SCC and the WAPDEG analysis AMR are similarly affected. While a more detailed analysis is required to quantify the impact of the design change, preliminary analyses indicate that any stress corrosion cracks in the DS will not result in the direct dripping of water on the waste package since it is believed that the cracks will become plugged with corrosion products. While this cracking constitutes failure of the DS, it is expected to continue to maintain its function of keeping the water away from the WP. These issues are planned to be addressed in the next revisions of the specific AMRs.

In addition to the increased susceptibility to SCC, the DS may also be subjected to increased corrosion due to rockfall-induced cold-worked regions. Preliminary review of the literature indicates that this is not a significant issue.

It is planned to address these issues in the next versions of the AMRs.

1.1 OBJECTIVES

1.1.1 Objectives of this Report

The objectives of this report are to document degradation models for the waste package and DS material with specific regard to the data input methodologies used to construct the model, uncertainties and limitations of the modeling results, and validation of the model. This report summarizes the following:

Assumptions that are specific to the individual models are listed in Chapter 3. Additional details of model assumptions can be found in Chapter 5 of the individual AMRs.

1.1.2 Purpose of the Analysis and Model Reports

The primary purpose of the supporting AMRs is to provide detailed documentation of the process-level models necessary for predicting the performance of the WP and DS materials in environments relevant to Yucca Mountain. These models enable engineers and scientists to predict the release of radionuclides from the WP, and their transport in the saturated and unsaturated zones. Figure 1-3 shows the model inputs, outputs and the laboratory test information that forms the bases for the confidence in the model results.

At the base of the model confidence foundation are the data generated from various testing programs; then, data along with other input parameters related to the repository design and expected environment are used in the development of degradation process models. The output from these models are calculated lifetimes for the waste package and drip shields.

1.2 SCOPE

The Waste Package Degradation PMR describes processes that will lead to degradation of the waste package components within the near-field environment (NFE). Specific technical information contained in the Waste Package Degradation PMR consists of data, analyses, models, software, and supporting documents. This report also provides a technical basis for the applicability of the overall integrated model for its intended purpose of evaluating postclosure performance of the Yucca Mountain repository system.

The Waste Package Degradation PMR provides information about important factors that affect WP and DS lifetimes. This PMR uses inputs from companion documents, including the license application design selection (LADS) report, which describes the EDA II design and expected temperature history for the waste package in the repository.

Chapter 2 of this PMR describes the evolution of the waste package degradation model. Details of the individual models and analyses are provided in Chapter 3. Chapter 4 describes the NRC IRSR, which serves as a driver for much of the work that is discussed. Acceptance criteria and responses to these criteria are addressed in this chapter. Summary and conclusions are provided in Chapter 5.

1.3 Principal Factors and Other Factors Considered

The magnitude of the Yucca Mountain Site Characterization Project (YMP) and the complexities associated with both the natural and engineered barrier systems dictate that the YMP prioritize its activities and focus on the factors most important to performance, hereafter named the Principal Factors. The Repository Safety Strategy: U.S. Department of Energy's Strategy to Protect Public Health and Safety After Closure of a Yucca Mountain Repository (CRWMS M&O 2000w) has identified seven Principal Factors and twenty Other Factors of second-order importance. The selection of the Principal Factors was based on preliminary TSPA analyses and expert judgment, which showed that these factors significantly affect the performance of the potential repository. The Other Factors were deemed to have minimal impact on the repository performance in terms of dose to the accessible environment. Table 1-1 lists the seven Principal Factors, the twenty Other Factors, and the PMRs that address each factor. Specific Principal Factors discussed in this report are:

Performance of the DS is a principal factor since it represents the diversion of seepage away from the WP. This factor defines the timing and amount of water transmitted through the DS.

Performance of the WP barriers is a principal factor for the postclosure safety case because it defines the timing and amount of water transmitted into the WP, and thereby controls the rate of release of radionuclides.

General guidelines dictate that the YMP bound the effects of the Other Factors, when possible, and perform analyses that are conservative. Principal Factors are to be studied and evaluated more thoroughly, using both realistic evaluations and bounding analyses.

1.4 QUALITY ASSURANCE

The Quality Assurance (QA) program applies to this analysis. All types of WP designs were classified as per Classification of Permanent Item, QAP-2-3 as Quality Level 1. This report applies to all of the WP designs included in the Monitored Geologic Repository Classification Analyses. Classification of the MGR Uncanistered Spent Nuclear Fuel Disposal Container System (CRWMS M&O 1999c) is cited as an example of a waste package type. The development of this report is conducted under the activity evaluation 1101213PM7 Waste Package Analyses & Models - PMR (CRWMS M&O 1999d), which was prepared per QAP-2-0, Conduct of Activities. The results of this evaluation indicate that the activity is subject to the Quality Assurance Requirements and Description (DOE 2000) requirements.

The Waste Package Degradation PMR was prepared in accordance with AP-3.11Q, Technical Reports, and reviewed in accordance with AP-2.14Q, Review of Technical Products. The AMRs that support this PMR were prepared in accordance with AP-3.10Q, Analyses and Models.

The status of the data supporting this PMR is included in the supporting AMRs and in the Document Input Reference System (DIRS) database. The data are incorporated in the Technical Data Management System (TDMS). Data verification and qualification were carried out in accordance with procedures AP-3.15Q, Managing Technical Product Inputs and AP-SIII.2Q, Qualification of Unqualified Data and the Documentation of Rationale for Accepted Data, respectively.

No software codes were used directly in the development of this PMR. However, this report does include the results from software codes used in the supporting AMRs.

ANSYS, Version 5.3, which is a finite element analysis code used for thermal and stress analyses, was used to develop data cited in SCC AMR (CRWMS M&O 2000f).

pcCRACK, Version 3.1, is a fracture mechanics code used for stress intensity and crack growth simulation analyses. This code was also used to develop data cited in the SCC AMR (CRWMS M&O 2000f).

WAPDEG, Version 4.0, is used to determine waste package and DS failure fractions as a function of time. Details of the use of this code are provided in the AMR on WAPDEG Analysis (CRWMS M&O 2000g).

1.5 WASTE PACKAGE DEGRADATION PROCESS AND ABSTRACTION MODELS

The integrated model for WP and DS degradation consists of several individual process-level and abstraction models, as well as some numerical analyses.

In all, eight (8) process-level models, six (6) abstraction models, and two (2) engineering calculations were developed and documented in individual AMRs or Calculations. These are listed below.

Process-Level Model Analyses:

 Abstraction Models and Calculations:

Results from the last two AMRs (CRWMS M&O 2000o and CRWMS M&O 2000p) are not being used as input to the Waste Package Degradation Nominal Case Analysis. These two reports are omitted as input since the stainless steel inner shell is not considered to be a corrosion barrier. Thus, it is not assigned any performance credit. The AMR on uncertainty and variability requires additional data to further develop the model, and will be included in the next major revision of the PMR. For this PMR, uncertainty and variability are included in the bounding approach used for process-level and abstraction models.

Figure 1-4 shows the elements of the integrated model, and interrelationships among the various process-level models and the AMRs that comprise it. Related phenomena are logically grouped in the process-level models. For example, the process-level model for general corrosion (GC) and localized corrosion (LC) of the waste package outer barrier (WPOB) includes dry oxidation (DOX), humid air corrosion (HAC), and expected environment on the surface. A brief overview of each of these models is presented below. Details of the process-level and abstraction models are presented in the AMRs and in Chapter 3.

1.5.1 Environment on the Surface of Drip Shield and Waste Package Outer Barrier

Information on the surface environment provided below is based on the parent AMR (CRWMS M&O 2000a, Section 1.2) and is discussed in greater detail in Section 3.1.3. This process-level model addresses the evolution of the chemistry of the water film on the DS and WPOB as a function of temperature and relative humidity (RH). The concentrations of aqueous salt solutions that can form on the hot waste package surface are being determined experimentally and theoretically (based upon chemical thermodynamics). These concentrations define the medium for testing WP materials under a worst-case scenario. An example is the development of a simulated saturated J-13 water (SSW) with an elevated boiling point (120-127°C). Hygroscopic salts may be deposited on surfaces by seepage water and episodic water flow, as well as by dust and aerosols entrained in ventilation air. The deliquescence point of these salts determines the RH at which humid-air and aqueous-phase corrosion commences.

Abstracted models are developed for the evolution of environments on the exposed surfaces of the DS and WPOB as a function of time and location within the repository. These abstracted models are in forms that are suitable as inputs to the WAPDEG analysis and include the uncertainty and variability of exposure conditions. Additional information on the WAPDEG code is presented in Section 2.2.

1.5.2 Mechanisms for Early Failures and Manufacturing Defects

Information on early failures provided below is based on the parent AMR (CRWMS M&O 2000m, Section 6) and is described in greater detail in Section 3.1.2. This analysis addresses the potential for early failures of the waste package due to material defects, as well as defects from waste package fabrication processes. These fabrication processes include welding. The probability of waste package fabrication defects, their uncertainty and variability, and the consequences of the defects on waste package failure times (e.g., number of potential failure sites and flaw size distribution) are discussed.

Abstracted calculations are developed for the occurrence and size distribution of defect flaws from material and manufacturing defects in the waste package. Abstracted calculations include uncertainty and variability of the above properties.

1.5.3 Aging and Phase Stability of Waste Package Outer Barrier

Information on aging and phase stability provided below is based on the parent AMR (CRWMS M&O 2000b, Section 6) and is described in greater detail in Section 3.1.4. This process-level model addresses degradation of the WPOB resulting from exposure to elevated temperatures. In the case of Alloy 22 and related materials, such thermal histories can result in the formation of precipitates (µ, P, ) and other undesirable phases. The precipitates can form within the individual grains or at grain boundaries. Precipitation can cause embrittlement, thereby increasing susceptibility to damage by rockfall and impact. Since these precipitates may be rich in molybdenum and chromium, two of the alloying elements responsible for the high degree of passivity of Alloy 22, aging can also result in increased susceptibility to general and LC, as well as to SCC. The time-temperature-transformation (TTT) curve and an expression for estimating the volume fraction of precipitates in the grain boundary (GB) have been developed for Alloy 22. Estimates of uncertainty are made. The effect of aging on corrosion has been addressed, and has been determined to be a corrosion enhancement factor of approximately 2.5 for the fully aged material.

1.5.4 General Corrosion and Localized Corrosion

Information on corrosion provided below is based on three separate AMRs (CRWMS M&O 2000c Section 6, 2000d Section 6, and 2000e Section 6) and is described in greater detail in Sections 3.1.5 and 3.1.6. Three separate process-level models were developed to address general corrosion and localized corrosion of the DS, WPOB, and stainless steel structural support respectively. The current design uses Titanium Grade 7 as the DS, Alloy 22 as the WPOB, and 316NG stainless steel as the inner structural material. Each of these models includes sub-models for DOX, HAC, GC in the aqueous phase, and LC in the aqueous phase. Note that "dry oxidation" and "dry air oxidation" are synonymous terms. While the stainless steel structural material is not specifically intended to be a corrosion barrier, it may affect the chemistry of water entering the waste package and retard the rate of radionuclide release from the breached waste package. Given the limited availability of data for 316NG stainless steel, data for 316L stainless steel are used as representative. This is appropriate since the compositions of these two materials are very similar, and since their susceptibilities to corrosion are known to be similar. Microbial corrosion is addressed in the section devoted to LC (Section 3.1.6.8).

1.5.4.1 Dry Oxidation

The process-level model for corrosion of the DS and WP materials includes a sub-model for DOX. It is assumed that DOX can be treated as a type of GC, with uniform attack. The rates of DOX are estimated as a function of temperature, to the extent possible.

1.5.4.2 Humid Air Corrosion & Vapor Phase Corrosion

The process-level model for corrosion of the DS and waste package materials accounts for humid air and vapor phase corrosion. These modes of corrosion are also treated as a type of GC. To the extent possible, rates of humid air and vapor phase corrosion (VPC) are estimated as a function of temperature.

1.5.4.3 Aqueous Phase Corrosion

The process-level model for aqueous phase corrosion (APC) of the DS and WP accounts for both general and localized attack. Two models for the initiation of LC are considered (Methods A and B). In Method A, the threshold potential for localized attack of the material is determined from experimentally determined cyclic polarization (CP) data obtained with relevant test environments. These environments include simulated dilute water (SDW), simulated concentrated water (SCW) and simulated acidic concentrated water (SAW) at 30, 60, and 90°C, as well as SSW at 100 and 120°C. The compositions of SDW and SAW are 10X and 1,000X concentrations of J-13 well water, respectively. More recently, basic saturated water (BSW) has been included in the set of standard test media. Published rates for LC are invoked when the expected open circuit corrosion potential exceeds the threshold potential. Since pitting has not been observed in laboratory experiments at LLNL, the primary mode of LC is expected to be crevice corrosion. If the threshold for LC is not exceeded, it is assumed that the mode of attack is GC. GC rates are estimated from various sources of available test data, including weight loss measurements from the Long Term Corrosion Test Facility (LTCTF). In Method B, a threshold temperature for localized attack at the open circuit corrosion potential is determined from published literature data or from tests at elevated temperature and pressure. The same rates of LC are used in Method B as used in Method A. This APC model will be applied to each unit area of the WP exterior surface.

Abstracted models have been developed to account for GC of the DS and WP materials. These abstractions are very similar to the process-level models. The abstracted models include thresholds for initiation of various modes of corrosion, as well as the corresponding rates of penetration. The abstracted models are in forms that are suitable for input to the WAPDEG analysis. The RH threshold for initiation of HAC and APC are included, as well as the electrochemical potential for initiation of localized attack during APC. In the case of the DS and WPOB, distributions of GC rates are based upon data from the LTCTF, while published data are used as the basis of estimating LC rates. Both general and localized corrosion rates of 316NG stainless steel are based upon published data. Estimates of the uncertainty and variability of each quantity are provided. When LC occurs, it is assumed to occur over an entire WAPDEG patch (element).

1.5.5 Stress Corrosion Cracking

The process-level model for the SCC is documented in the corresponding AMR (CRWMS M&O 2000f, Section 6) and is described more fully in Section 3.1.7. SCC of materials may occur when an appropriate combination of material susceptibility, tensile stress and environment are present. This process-level model accounts for the possibility of SCC of the DS, the WPOB, and the stainless steel structural material. This model also evaluates two alternative methods: Method A which is based on threshold stress intensity factor criterion for initiation of SCC (KI > KISCC); and Method B which is based upon a threshold stress and a finite rate of SCC propagation. The rate of SCC propagation is dependent upon the local environment and the stress intensity factor at the crack tip. The stresses for initiation and propagation of SCC are due to unannealed closure welds, deformation caused by rock fall, and the weight of the WP. This particular analysis requires appropriate stress analysis models and measurements for calculation of the through-wall stress distribution for representative cross sections of the WP, including unwelded base metal and unannealed welds. This stress distribution is used to calculate a corresponding stress intensity factor distribution for flaws that range in size from zero to the entire thickness of the wall including the welded region. This stress intensity factor distribution is used as input for both Methods A and B. In Method A, SCC initiates at pre-existing flaws that develop during fabrication of the waste package, or at flaws that develop during LC. Values of KISCC are based upon data published in the technical and scientific literature, or measurements made with the double cantilever and compact tensions beam techniques. In Method B, SCC is initiated if the threshold stress is exceeded on a smooth surface. Then, the SCC propagation rate is calculated as a function of local environment and stress intensity factor. The time-to-failure is determined by integrating the calculated propagation rate. As previously discussed, relevant test environments include SDW, SCW, and SAW at 30, 60, and 90°C, SSW at 100 and 120°C, and BSW at 110°C.

Abstracted models have been developed for SCC of the WPOB (Alloy 22). These abstracted models include: (1) a threshold stress for initiation; (2) a crack growth rate as a function of stress and local exposure conditions, including temperature; (3) crack density; and (4) crack morphology. Crack morphology includes a description of the size of openings. The abstracted models are in a form that is suitable for input to the WAPDEG analysis, and include the uncertainty and variability of the above processes.

Post-weld stress mitigation techniques are being employed to delay the initiation of SCC. The techniques under consideration include laser peening and localized induction annealing. These processes are accounted for in the model presented here.

1.5.6 Hydrogen-Induced Cracking of Drip Shield

This process-level model establishes the conditions under which the DS (Titanium Grade 7) will experience hydrogen uptake, thereby leading to the threat of hydrogen embrittlement and hydrogen-induced cracking (HIC). This mode of failure is not believed to be a credible threat to Alloy 22 and has been, excluded as shown in Section 1.6 below. This is based on the design, which includes backfill, and the rate of hydrogen pick up is very low. HIC may be a greater threat without backfill, since the titanium DS can be galvanically coupled to carbon steel rock bolts and mesh. Analysis of the galvanic effects and hydrogen pick up is planned to be included in the next version of the AMR.

1.6 Screening of Features, Events, and Processes

The initial set of features, events and processes (FEPs) has been developed for the YMP TSPA by combining lists of FEPs identified as relevant to the YMP. This combined list consists of 1,261 FEP entries from the Nuclear Energy Agency working group, 292 FEPs from YMP literature and site studies, and 82 FEPs identified during YMP project staff workshops. The FEPs have been identified by a variety of methods including expert judgement, informal elicitation, event tree analysis, stakeholder review, and regulatory stipulation. All potentially relevant FEPs have been included, regardless of origin. The compilation included FEP entries mentioned above and 151 layers, categories and headings. It resulted in a FEP list of 1786 entries. This approach has led to considerable redundancy in the FEP list, because the same FEPs are frequently identified by multiple sources, but it also ensures that a comprehensive review of narrowly defined FEPs will be performed.

Each FEP has been classified as either a Primary or Secondary FEP. The classification resulted in the identification of 310 Primary FEPs. Primary FEPs are those for which detailed screening arguments are developed. The classification and description of Primary FEPs strives to capture the essence of all the Secondary FEPs that map to the primary. Secondary FEPs are either FEPs that are completely redundant or that can be aggregated into a single Primary FEP. The Primary FEPs have been assigned to associated PMRs. The assignments were based on the nature of the FEPs so that the analysis and resolution for screening decisions reside with the subject-matter experts in the relevant disciplines. The resolution of other system-level FEPs are documented in AMRs prepared by the responsible PMR groups. This section summarizes the screening decisions associated with the FEPs for the waste package and DS PMR group. Details of the screening processes are documented in the associated AMR.

The scope of the FEPs screening is to identify the treatment of the Primary FEPs affecting waste package and DS degradation. The FEPs that are deemed potentially important to repository performance are evaluated, either as components for the TSPA or as separate analyses in the AMR. The scope for this activity involves the following two tasks:

Task 1: Identify FEPs that are considered explicitly in the TSPA (called included FEPs) and the AMRs in which these FEPs are addressed
Task 2: Identify FEPs not included in TSPA (called excluded FEPs) and provide justification for why these FEPs do not need to be a part of the TSPA model.

Of the original list of FEPs, twenty-eight have been identified as Primary FEPs in relationship to waste package and DS degradation. The approach used for this analysis is a combination of qualitative and quantitative screening. The analyses are based on the criteria provided by the NRC in the proposed 10 CFR Part 63 (Dyer 1999) and by the U.S. Environmental Protection Agency in the proposed 40 CFR Part 197 (64 FR 46976) to determine whether or not each FEP should be included in the TSPA. For FEPs that are excluded from the TSPA based on NRC or U.S. Environmental Protection Agency criteria, the screening argument includes a summary of the basis and results that indicate either low probability or low consequence. As appropriate, screening arguments cite work performed outside this activity, such as in other AMRs. For FEPs that are included in the TSPA, the TSPA disposition includes a reference to the AMR that describes how the FEP has been incorporated in the process models or the TSPA abstraction models.

The FEPs screening analysis results for the twenty-eight Primary FEPs relevant to waste package and DS degradation processes are summarized in Table 1-2. This table shows the FEP number, FEP name, screening decision (include/exclude), and a synopsis of the screening argument. Details of the screening processes and arguments are planned to be included in the next revision of the FEPs AMR.

1.7 RELATIONSHIP TO OTHER PROCESS MODEL REPORTS AND DOCUMENTS

This PMR provides information about important factors affecting waste package lifetime such as the thermal, hydrologic, and geochemical processes acting on the waste package surface. The PMR uses inputs from other documents such as LADS report (EDA II design) and thermal analyses. The emphasis of the discussion of model inputs and outputs is on information needed for the assessment of postclosure performance. The waste package degradation PMR supports the TSPA and other major Project milestones, such as the Site Recommendation (SR) and the License Application (LA).

While the scope of this PMR is to address degradation of the WP and the DS, other PMRs (such as the EBS PMR) address other aspects of DS and EBS component performance.

 

 

2. EVOLUTION OF THE PROCESS MODEL

2.1 BACKGROUND

The overall performance of the WP and DS materials has been identified as a principal factor in the performance of the repository. It is expected that the lifetime goals for the DS and WP may be increased in the future. Therefore, materials and designs that prolong service life are sought continuously.

The reference design used in the viability assessment (VA) has been estimated to experience first through-wall failure by pit penetration in about 2,700 years, with about 1% of the packages failing in approximately 10,000 years. This estimate is based upon the waste package degradation model (WAPDEG) used as input to Total System Performance Assessment-Viability Assessment (TSPA-VA) Analyses Technical Basis Document (CRWMS M&O 1998a, TSPA-VA, Chapter 5). However, the goals of the YMP have been continuously pushed towards longer WP lifetimes, thereby requiring that the repository exceed performance requirements of the VA design by a significant margin. Accordingly, a selection process for alternative materials and designs intended to provide higher levels of confidence for an extended WP lifetime was undertaken. As mentioned in the LADS report, this process resulted in the selection of a double-walled WP placed under a protective DS made of Titanium Grade 7 (CRWMS M&O 1999a). The outer wall of the WP is corrosion-resistant Alloy 22, with an inner wall of stainless steel (316NG) that serves as a structural support. This new design is known as EDA II (CRWMS M&O 1999a). The selection of this new design configuration, coupled with the selection of new materials, has necessitated the development of new models to predict penetration rates. Individual component models (sub-models) are documented in the individual Analysis and Model Reports (AMRs) (see Section 1.5 for the topics in each AMR). Each AMR is provided as an input to the WP Degradation PMR, as well as to the TSPA for Site Recommendation.

2.2 PRevIOUS TREATMENT OF WASTE PACKAGE DEGRADATION MODELING

Modeling of WP degradation has evolved over the past decade along with changes in materials selected for its containment barriers. Early WP designs consisted of thin-walled stainless steel canisters with heavy-walled carbon steel overpacks, and were designed for emplacement in salt, basalt or tuff repositories. The design thickness of the overpack component was the sum of the required structural thickness, plus the corrosion allowance necessary to assure that the required structural thickness will survive the required containment period. The corrosion allowance was established on the basis of the calculated temperature profile of the WP surface during the containment period, the unexpected presence of an unlimited quantity of anoxic brine, and the resulting corrosion rate.

In late 1987, the U.S. Congress passed The Nuclear Waste Policy Amendments Act, As Ammended. With Appropriations Acts Appended (DOE 1995), which reduced the number of potential repository sites to be characterized to one: Yucca Mountain. Prior to this event, work on the WP design for Yucca Mountain had followed the same rationale as that for salt and basalt repositories. In fact, the initial conceptual design of the WP developed in early 1983 was the same as that for the other two repositories. However, it was recognized that the expected conditions in Yucca Mountain were quite different from the other two in that the repository was located in the unsaturated zone. The environment in this zone, although not firmly established, was expected to be oxidizing, but dry with low humidity most of the time. The design temperature for the WP surfaces was not expected to exceed 250°C. Liquid water was expected to be present only under transient conditions, and its composition was not expected to be very aggressive from a corrosion standpoint.

The limitations of 304L stainless steel, from the standpoint of SCC and LC had been recognized from early on. Consequently, alternative materials were being sought. In 1993 (CRWMS M&O 1994), the design was changed from thin-walled containers emplaced in bore holes to drift-emplaced robust multibarrier WP. The proposed materials were Alloy 825 as the inner barrier and a corrosion allowance material (CAM) "such as weathering steel" as an outer barrier. Degradation modeling of this WP design included only APC. It was assumed that no corrosion would occur above the temperature at which liquid-phase water could not exist on the WP surface.

Model enhancements were incorporated in 1995 (CRWMS M&O 1995) and were based upon the same double-wall WP design that was used in 1993. This model was an initial attempt to account for HAC and APC. The APC model had components that simulated pitting corrosion and galvanic coupling of the CAM and corrosion resistant materials (CRM). This model included estimates of the variability in WP corrosion, including package-to-package and patch-to-patch variability. Estimates of the uncertainty in the threshold RH for initiation of HAC and APC were also made. An empirical model for GC and LC of the CAMs was developed based upon published literature data. Rates of GC and LC of the CRM were based upon the collective opinion of a panel of experts (expert elicitation) (CRWMS M&O 1995). The model assumed that galvanic protection would delay LC (pitting) of the CRM until a specified percentage of the CAM thickness had been consumed by GC.

Exposure conditions included temperature, RH, presence of liquid-phase water (dripping), water chemistry, backfill, and rock fall (CRWMS M&O 1995).

Along with the improvements in the degradation modeling, the evaluation of materials selection for the WP barriers continued during the following several years. The corrosion resistant material, Alloy 825, was replaced with a more corrosion resistant nickel-based Alloy 625 during this period (CRWMS M&O 1996). The lifetime goals for the WP was increased further and resulted in the selection of Alloy 22 as the corrosion resistant barrier for the Viability Assessment design. The superiority of Alloy 22 is well known and generally accepted.

Additional improvements in the WP corrosion model were made for the Viability Assessment (VA) design (CRWMS M&O 1998a, Chapter 5). This model accounted for humid-air GC and LC of the CAM; aqueous-phase GC and LC of the CAM; and aqueous-phase GC and LC of the CRM. Degradation of the WP was modeled by dividing the surface into patches and populating the corrosion rates stochastically over the patches. The concept of a localization factor was used. Pitting of the CAM was assumed to occur under alkaline conditions (pH 10).

Microbiologically influenced corrosion (MIC) and SCC were not accounted for in the TSPA models used for VA. The effects of salt deposition and evaporative concentration of dripping water on the WP surface were also neglected. These models were not based upon qualified experimental data from materials testing in repository-relevant environments, but relied heavily on the opinion of experts and other published data. For this PMR, a broad range of laboratory data and associated process models have been developed to more realistically approximate the degradation processes of potential significance to repository performance. Changes to the degradation models have been necessitated by the changes in WP design and inclusion of the DS for Site Recommendation (SR).

2.3 Total System Performance Assessment-Site Recommendation APPROACH

The approach in the TSPA-SR WP degradation analysis is greatly enhanced version of that used in TSPA-VA. (For convenience of discussion in this section, the DS is modeled as an integral part of WP, and no separate discussion is given for DS.) The WAPDEG model, which is based on a stochastic approach, has been upgraded to include SCC and is used for the SR WP degradation analysis. The motivations for the stochastic approach used in the WAPDEG model are three-fold:

Abstractions of the process-level models were developed for WAPDEG in a manner that allows important features of the process-level models to be captured as explicitly as possible, and in a manner that allows the degradation processes and their characteristics to be properly represented in the WP degradation analysis. More details of the TSPA-SR approach to WP and DS degradation analysis are given in the supporting AMR entitled WAPDEG Analysis of Waste Package and Drip Shield Degradation (CRWMS M&O 2000g).

As in the TSPA-VA analysis, effects of spatial and temporal variations in the exposure conditions over the repository were modeled by explicitly incorporating relevant histories of exposure condition into the WP degradation analysis. The parameters that represent exposure conditions were considered to be varying over the repository. These include RH and temperature at the WP surface, seepage into the emplacement drift, chemistry of seepage water, and rockfall. In addition, potentially variable corrosion processes within a single WP were represented by dividing the WP surface into unit areas called "patches" and stochastically populating the corrosion model parameter values and/or corrosion rates over each patch. The model parameter values and corrosion rates were sampled from their distribution over the range of the expected local exposure conditions.

In the nominal case analysis, the WPOB and DS were included in the WP degradation analysis. The stainless steel inner container, which is to provide structural support to WP, is not included in the degradation analysis. Although, this inner container would actually provide some performance for waste containment after the outer barrier breach, and would also provide a barrier to radionuclide transport after the WP is breached, the potential performance credit was ignored in the nominal TSPA-SR analysis. This is a conservative approach. However, performance credit for the stainless steel is being evaluated for future consideration.

In summary, the TSPA-SR WP degradation analysis includes the following potentially important degradation processes:

As previously discussed, significant improvements have been made to the GC and LC models, making them superior to those used in TSPA-VA. In the analysis, the DS was considered to be immune to SCC because it will be fully annealed before it is placed in the emplacement drift. Likewise, all the fabrication welds in the WP, except the welds for the closure lids, will be fully annealed and therefore not subject to SCC. Only the WP closure weld is considered in the SCC analysis. Two alternative SCC models were considered, the slip dissolution model and the threshold stress intensity factor model. The effect of radiolysis on corrosion is expected to be insignificant under the conditions expected in the repository (see Section 3.1.6.6), and was therefore not included in the nominal case analysis. The DS was assumed to be immune to the MIC. Since the bounding analyses have shown that the hydrogen uptake by the DS is much less than the threshold hydrogen concentration for HIC (CRWMS M&O 2000h, Sections 6.1.3 and 6.2.4), this mode of failure was not included in the DS degradation model.

WP failure requires through-wall penetration. The WAPDEG analysis tracks degradation of WP for three penetration modes: SCC (crack penetration), LC (rapid crevice penetration), and GC (slower uniform penetration). Here, localized attack is assumed to be crevice corrosion over an entire patch, which is conservative. The analysis provides, as output, the cumulative probability of WP failure by one of the three penetration modes as a function of time and also provides the number of penetrations for each of the penetration modes as a function of time. The WP failure time and number profiles for penetration are used as input to other TSPA analyses, such as the WF degradation and the radionuclide release rate from WPs.

The TSPA-SR analysis yields a more explicit representation (than previous TSPA analyses) of the uncertainty and variability in WP degradation (i.e., WP failure and penetration number profiles). For the corrosion models and parameters for which data and analyses are available, their uncertainty and variability were quantified and incorporated into the WAPDEG analysis. For other models and parameters for which the uncertainty and variability are not quantifiable, the variance in their value was assumed and used as uncertainty. In the TSPA-SR analysis, WP degradation was analyzed with multiple realizations of WAPDEG for the uncertainty analysis of the uncertain corrosion parameters—each WAPDEG realization corresponding to a complete WAPDEG run to account for the WP degradation variability for a given number of WPs. Accordingly, each of the WAPDEG analysis outputs (i.e., WP failure time, crack penetration number, pit penetration number, and patch penetration number) is reported as a group of curves that represents the potential range of the output values.

2.4 Issues Related to Waste Package Degradation

This PMR also addresses related issues identified based on a review of the past two years Advisory Committee on Nuclear Waste (ACNW) and Nuclear Waste Technical Review Board (NWTRB) meeting summaries and correspondences, Viability Assessment (VA) Volume 3, Sections 6.5.2 and 6.5.3 and Volume 4 Section 4.3, TSPA peer review documentation, NRC comments on TSPA-VA, expert elicitation recommendations, and licensing correspondence files for the NRC, the State of Nevada, and the Nevada counties. Table 2-1 provides a summary of all identified issues and describes how each issue is addressed in this PMR. In addition, acceptance criteria from the Issue Resolution Status Reports Key Technical Issue: Container Life and Source Term (NRC 1999a), the Issue Resolution Status Report Key Technical Issue: Total System Performance Assessment and Integration (NRC 2000), and the Repository Design and Thermal-Mechanical Effects Report (NRC 1999b) are separately addressed in Table 4-1 in Chapter 4 of this document.

 

 

3. Models and ABSTRACTED MODELS

The WP degradation process model consists of several different models: dry oxidation, humid air corrosion, aqueous phase corrosion, general corrosion, localized corrosion, microbiologically induced corrosion, stress corrosion cracking, and hydrogen-induced cracking. A generic integrated model containing the above component models is illustrated by Figure 1-4. This model can be applied to the DS and WP materials of interest.

3.1 Model Descriptions

As stated above, the WP degradation process model includes a number of component models. These component models are discussed in the following sections. Model parameters and definitions for the individual models are provided in the respective AMRs.

3.1.1 Overview of Waste Package and Drip Shield Design

As described in the LADS report, the recommended WP design is EDA II (CRWMS M&O 1999a, Section 1.1.2). This design includes a double-wall WP underneath a protective DS. The DS is to be fabricated from Titanium Grade 7. The EDA II corrosion resistant WPOB is to be fabricated from nickel-based Alloy 22. Stainless steel 316NG is to be used for construction of the structural support container within the WPOB. The 316NG inner cylinder will increase the overall strength of the WP.

3.1.1.1 Titanium Drip Shield

Titanium alloys (1.5-cm thick) have been considered for construction of the DS. The current recommendation is to use Titanium Grade 7 [Unified Numbering System for Metals and Alloys (UNS) R52400]. The composition of this alloy is 0.03% N (max), 0.10% C (max), 0.015% H (max), 0.25% O (max), 0.30% iron (max), 0.12-0.25% Pd (max), and 0.4% Residuals (total), with the balance being Ti (approximately 98.7 to 98.8%). The nominal thickness of the DS is 15 mm. Properties and performance of these materials are reviewed elsewhere and cited in the respective AMRs. The unusual corrosion resistance of titanium alloys is due to the formation of a passive film of TiO2, which is stable over a relatively wide range of electrochemical potential and pH. A similar material, Titanium Grade 16 with 0.04 to 0.08% Pd, is used as an analog for Titanium Grade 7 in some parts of the testing program. The rates of general corrosion and dry oxidation (or dry air oxidation) of this material have been shown to be very low. Corrosion testing of Titanium Grade 16 has been conducted in the Long Term Corrosion Test Facility (LTCTF) of the YMP (CRWMS M&O 2000d, Section 6.5).

3.1.1.2 Nickel-Based Waste Package Outer Barrier

Alloy 22 [UNS N06022] (2.0 cm thick) is now being considered for construction of the WPOB. This alloy consists of 20.0-22.5% chromium, 12.5-14.5% molybdenum, 2.0-6.0% iron, 2.5-3.5% W, and 2.5% Co (max), with the balance being nickel (approximately 50-60%). Other impurity elements include P, Si, S, Mn, Nb, and V. Alloy 22 is less susceptible to LC in environments that contain Cl- than Alloys 825 and 625, materials of choice in earlier WP designs. Corrosion testing of Alloy 22 has been and continues to be conducted in the LTCTF of the YMP (CRWMS M&O 2000c, Section 6). Nominal thickness of the Alloy 22 shell is 20 mm for the commercial spent fuel packages and 25 mm for the packages containing navy waste.

3.1.1.3 Stainless Steel Structural Material

316NG stainless steel (5-cm thick) is to be used for construction of the structural support inside the WPOB. This inner cylinder of 316NG stainless steel will increase the overall strength of the WP. 316L stainless steel is considered to be a good analog for 316NG stainless steel because the chemical composition of the two alloys is essentially the same, except that 316NG stainless steel has better mechanical properties than 316L. 316 stainless steel [UNS S31603] has a composition of 16-18% chromium, 10-14% nickel, 2-3% molybdenum, 2% Mn (max), 1% Si (max), 0.03% C (max), 0.045% P (max), 0.03% S (max), 0.10% N (max) and the balance being iron (65-69%). 316L stainless steel is less susceptible to LC in environments that contain Cl- than stainless steel 304, but more susceptible than other corrosion resistant materials such as Alloys 22, 625 and 825 that have been considered in various WP designs. The superior LC resistance of 316L stainless steel in comparison to 304 stainless steel is apparently due to the addition of molybdenum, which helps to stabilize the passive film at low pH values. Molybdenum oxide is very insoluble at low pH. Consequently, 316L stainless steel exhibits relatively high thresholds for localized attack (CRWMS M&O 2000e, Section 1.2).

3.1.2 Manufacturing Defects (Early Failure AMR)

Manufacturing defects and failure modes that might lead to early failure of a WP are accounted for in an AMR (CRWMS M&O 2000m, Section 6) that supports this PMR. The AMR on early failure includes a literature review directed towards obtaining information on the rate of manufacturing defect-related failures in various types of welded metallic containers, the types of defects that produce these failures, and the mechanisms that cause defects to propagate to failure. Types of defects applicable to the current WP design are identified. For each applicable type of defect, the probability of its occurrence on a WP is estimated. Potential consequences to the long-term performance of the WP if the defect is present are discussed. Specific details on how the defect will affect WP materials are provided in separate AMR on SCC (CRWMS M&O 2000f, Section 6). Defects or flaws may serve as initiation sites for SCC.

3.1.2.1 Analysis Assumptions in AMR

The following assumptions support the development of probabilities for various size flaws in the welds of the WP shell and lids. Based on the similarity of the processes used for welding Alloy 22 and stainless steel, they are predicted to have the same frequency and size distributions for flaws. Information on the reliability of radiographic, ultrasonic, and dye-penetrant testing is assumed to be applicable to the materials and inspection methods that will be used for the WP. This information is based on older reliability studies of these non-destructive examination methods, and the assumption that future improvements in the inspection technology will result in increases in the probability of flaw detection. It was assumed that flaws detected by post-weld inspections will be repaired, whenever the flaw size is larger than the flaw size of concern for postclosure performance. Embedded weld flaws are not considered to be a concern for postclosure performance in the supporting AMR (CRWMS M&O 2000m, Section 6), since the WP is not a pressure vessel and will not be subjected to cyclic fatigue (the primary mechanism for causing such flaws to grow through-wall in pressure vessels). However, as the weld undergoes GC, subsurface flaws may eventually be exposed.

The probabilities of human error have not been quantified for the specific actions associated with the fabrication of the WP, but the information used represents human error probabilities for similar types of actions.

In developing the probability of the use of improper material in the WP shell or lid welds, it is assumed that a field verification of the chemical composition of weld wire will be performed prior to its use in fabricating any weld and that material controls required in nuclear quality programs will be used. It is further assumed that such field verification will use state-of-the-art instrumentation. This assumption is based on the expected administrative requirements on the process qualification program.

Assumptions are used to support the development of the probability of having corrosion-enhancing surface contamination on the WP or improper heat treatment of the WP. These assumptions are based on the general descriptions of these activities. The assumptions support the development of event sequence trees for quantifying the probabilities of improper heat treatment or a failure in the cleaning process. The assumptions involve the number of operators involved in each process, the QA procedures and inspections governing the processes, and the reliability of the equipment used.

It is assumed that the probability of damaging a WP during transport or handling at the repository is equivalent to the probability of damaging spent nuclear fuel (SNF) assemblies during transport or handling. The basis for this assumption is that a WP will be subjected to about the same number of handling steps as a SNF assembly. It is assumed that both are handled with about the same amount of care. It is expected that the WP will be inspected for handling damage upon arrival at the repository and before final emplacement in the drift. It is further expected that the WP will be completely repaired or scrapped if such handling damage occurs.

3.1.2.2 Analysis Description in AMR

The AMR presents the results of a literature review performed to determine the rate of manufacturing defect-related failure for various types of welded metallic containers. In addition to providing examples of the rate at which defective containers occur, this information provides insight into the various types of defects that can occur and the mechanisms that cause defects to propagate to failure. In summary, eleven generic types of defects were identified. These are:

For dry storage casks, all of the defects were identified by post-weld inspection prior to commencement of the storage phase, and thus do not represent true instances of early failure as it is defined in the AMR. The eleven types of defects were reviewed for their applicability to the WP. From this review, the following generic defect types are considered not applicable to the WP: improper weld flux material, poor weld joint design, missing welds, and mislocated welds. This determination is based on the fact that the welding process for WP fabrication does not use flux as noted in the AMR. Poor joint design is unlikely because of extensive development and testing. Missing welds and mislocated welds are easily detected and controlled by process qualification. The probability of occurrence and the effect on postclosure performance of the WP are assessed for the remaining defects.

Using information on linear flaw density, flaw size distribution, inspection reliability, and information on various weld lengths, frequencies of weld flaws of various size that break the outer surface have been estimated in the supporting AMR (CRWMS M&O 2000m). The procedure is essentially the same for all cases. First, the total flaws per type of WP weld were calculated by multiplying the weld length by the linear flaw density and by an adjustment factor for the weld thickness. The base linear flaw density with credit for radiographic and dye-penetrant test inspections was used for the shell and bottom lid welds, and the uninspected flaw density was conservatively used for the top lid closure weld. Next, the flaw size distribution for that weld thickness was used to determine the probability that a flaw would have a size within a given range. A range size of 0.5% of the weld thickness was used. This was the largest size range that could be used without introducing any significant (within two significant figures) amount of numerical error associated with discretizing a continuous size distribution. The probability for each size range was then multiplied by the total number of flaws per weld to determine the expected number of flaws within that size range. For welds subjected to an ultrasonic (UT) inspection, the expected number of flaws within each size range was then reduced by multiplying by the probability of nondetection (PND) for the lower end of the size range. This is conservative because the PND is higher for smaller flaws and ultrasonic inspection identifies small flaws. Since the UT PND is based on a single angle UT examination and a multi-angle examination is planned for the lid welds (possibly four angles), the square of the PND was used for the lid welds. This effectively treats a multi-angle exam as two independent examinations. For all cases, each range was then multiplied by 0.34% to yield the expected number of outer surface breaking flaws within that range. Finally, the expected number of outer surface breaking flaws in each size range were summed to determine a new value for total flaws per weld which accounts for the UT inspections. A complementary cumulative distribution of outer surface breaking flaw size was also determined. These results are summarized in Figure 3-1 for the Alloy 22 barrier shell welds, and in Figure 3-2 for the Alloy 22 barrier lid welds.

3.1.2.3 Uncertainties in AMR

The inputs used to estimate the probability of various defects that can potentially lead to early failure are open to interpretation and uncertainty. An uncertainty analysis was performed to develop an upper bound for an event sequence probability based on the uncertainty of modeled human actions. The analysis applies to those defects for which probabilities are estimated using event sequence trees, namely: DS emplacement error, WP handling error, WP surface contamination, thermal misload, and improper heat treatment. The method used to establish an upper bound value for event sequences combines the human error rates probabilistically. Uncertainties are considered only for human error probabilities related to failures. Probability components for success are treated at their nominal level (i.e., without uncertainty), which produces conservative results. No upper bounds were estimated for other failure probabilities related to mechanical failure or based on historical data. Accordingly, the upper bound for an event sequence probability is adjusted for human error probability uncertainties only (CRWMS M&O 2000m).

3.1.2.4 Analysis Conclusions in AMR

The AMR on early failure of the WP reviewed available literature on defect-related early failures of welded metallic components. Types of components examined include boilers and pressure vessels, nuclear fuel rods, underground storage tanks, radioactive cesium capsules, dry-storage casks for SNF, and tin-plate cans. The fraction of the total population that failed due to defect-related causes during the intended lifetime of the component is generally in the range of 10-3 to 10-6 per container. In most cases, defects that lead to failure of the component require an additional stimulus to cause failure (i.e., the component was not failed when it was placed into service). There were several examples that indicate that even commercial standards of quality control could reduce the rate of initially failed components well below 10-4 per container. The literature review identified eleven generic types of defects that could cause early failures in the components examined: weld flaws; base metal flaws; improper weld material; improper heat treatment; improper weld flux material; poor weld joint design; contaminants; mislocated welds; missing welds; handling and installation damage; and administrative error resulting in an unanticipated environment. The following defect types are considered "not applicable" to the WP: improper weld flux material, poor joint design, missing welds, and mislocated welds. The analysis estimates the probability that specific defect types will occur on a given WP, despite a set of quality controls designed to prevent their occurrence. Results of the analysis for the remaining seven types of defects are shown in Table 3-1.

3.1.2.5 Accounting for Embedded Flaws

While the AMR based the determination of flaw density on surface-breaking flaws, a more conservative approach is to base such determinations on embedded flaws. The recent work by Khaleel et al. (1999) is cited. The TSPA analysis that will be discussed in subsequent sections uses estimates of flaw density based upon data given in Table V of this reference. Specifically, the values for embedded flaws at depths equivalent to the outer quarter of the wall thickness are used. In the dual-lid WP design, cumulative distribution functions are needed for the closure welds of both the inner and outer lids. The flaw density for the outer lid weld is 18 defects per WP at the 50th percentile, and 40 defects per WP at the 100th percentile. The flaw density for the inner lid weld is 15 defects per WP at the 50th percentile, and 40 defects per WP at the 100th percentile (same as outer lid at 100th percentile). Note that each closure weld is represented by thirty two (32) WAPDEG patches.

3.1.2.6 Accounting for Flaw Orientation

In considering the potential effects of weld defects on SCC, the presence of planar defects in the region of the weld and heat-affected zone (HAZ), where weld-induced residual tensile stress exists, can lead to SCC initiation and growth. The two principal attributes of such weld defects that foster SCC are the stress concentration effect at the base of the defects, which can generate a stress intensity, and the occluded nature of such defects that may lead to the development of more aggressive crevice chemistry within the defect volume. However, as described in the SCC AMR, only defects oriented normal to the direction of the weld centerline (radially oriented defects) have sufficient calculated stress intensity to drive a stress corrosion crack through wall.

Weld defect types and expected defect orientation for the closure weld case are described briefly below.

The fabrication welds will be performed at the contractors’ facilities. Currently, only two weld methods are being considered for the fabrication process, gas metal arc and tungsten inert gas methods. This automatically eliminates slag inclusions, the most commonly found defect when the autosubmerged arc welding process is used. The most common defects for gas metal arc and tungsten inert gas are lack of fusion. This occurs because of missed sidewall or lack of penetration in the sidewall. This generally produces large defects that are readily found by ultrasonic and radiographic inspection. The other defect types are tungsten inclusion, silicon, and porosity. Because both ultrasonic and radiographic methods will be used for post-weld inspections, there should be no undiscovered defects for these welds. Additionally, dye penetrant inspection will be performed on the surface of the weld to detect and repair any surface-breaking defects. The lack of fusion defect is, by definition, oriented in the direction of the weld bead. The silicon, porosity, and tungsten are rounded defects that have no direction.

The closure weld will be made in the hot cell facility using the narrow groove tungsten arc welding process. This, by definition, eliminates the lack of fusion defects between beads since this is a single-pass process. The other defects such as tungsten inclusion, caused by the flaking of the tungsten electrode, and porosity, caused by the loss of gas coverage, are easily detectable by monitoring systems that will be built into the welding system. This leaves only nonfusion defects, which are detectable by ultrasonic testing (UT). All of the above defects are either rounded or in the direction of the weld seam; none is oriented in the radial direction.

The defect description discussed above is consistent with the brief comment on flaw orientation in the Early Failure AMR, "No information was found in the literature regarding angle of the flaw from a line parallel to the direction of the weld. However, most planar defects, such as lack of fusion and slag inclusions, would logically be expected to be oriented within a few degrees of the same direction in which the weld head is moving."

This flaw description is also consistent with the relevant literature paper (Shcherbinskii and Myakishev 1970) that describes a statistical treatment of weld-flaw orientations based on analysis of a significant data set of ultrasonic flaw orientation measurements. This paper concludes that planar-type weld defects detected ultrasonically tend to be predominately oriented in the direction of the weld centerline. It appears that more than 98% of the defects fall within 16 degrees of the weld centerline in the case of steam-pipe welds. A similar conclusion is drawn from the data for sheet-structure welds. Statistical distribution of the defects with respect to the orientation angle yields a probability of 99% that the defects are located within about 13 degrees. This suggests that much less than 1% of these flaws have a potential to undergo SCC.

Above discussions indicate that a correction factor for weld-flaw orientation for the embedded flaw density in the outer quarter of the thickness should be applied and used in waste package lifetime calculations. Based on the welding process and the inspection techniques to be employed for the closure welds, and the narrowness of the flaw orientation distributions presented in the subject paper, it is recommended that a conservative multiplication factor of 0.01 (1%) be used on total number of flaws of any given size for the subsurface flaws.

Ultrasonic examinations have now been performed on three actual WP welds on two mock-ups. These were unannealed closure welds, one on Alloy 625 and two on Alloy 22. The total length of weld was approximately 45 feet, and no defects were detected. Therefore, the probability of defects with actual welds is not inconsistent with low-defect densities that will result from this recommendation.

3.1.3 Environment on the Surface of the Waste Package and Drip Shield

The WP will experience a wide range of environments during its service life. Initially, it will be hot and dry due to the heat generated by radioactive decay. However, the temperature will eventually drop to levels where both HAC and APC will be possible. A companion AMR Environment on the Surface of the Drip Shield and Waste Package Outer Barrier (CRWMS M&O 2000a, Section 6) defines the detailed evolution of the environment on the WP surface. Input for this AMR includes bounding conditions for the local environment on the WP surface, which include temperature, RH, presence of liquid-phase water, liquid-phase electrolyte concentration (chloride, buffer, and pH), and oxidant level. This AMR has been used to define the threshold RH for HAC and APC, as well as a medium for testing WP materials under what is now believed to be a worst-case scenario. These test media are the neutral-pH SSW and the high-pH BSW, with nominal boiling points of 112 and 120°C, respectively.

Crevices will be formed between the WP and supports, beneath mineral precipitates, corrosion products, dust, rocks, cement, and biofilms. After the WP fails, the gap between the WPOB and the stainless steel structural support can form crevices where the environment may be more severe than the NFE. The hydrolysis of dissolved metal can lead to the accumulation of H+ and a corresponding decrease in pH. Electromigration of Cl- (and other anions) into the crevice must occur to balance cationic charge associated with H+ ions. These conditions can exacerbate subsequent attack of the WPOB and stainless steel structural material by general and LC, SCC, and other mechanisms. Crevices might also form with the DS. These are addressed in the general and LC discussions in Sections 3.1.5 and 3.1.6.

3.1.3.1 Threshold Relative Humidity

As represented by Equation 3-1, HAC can occur at any RH above the threshold (CRWMS M&O 2000c, 2000d, 2000e):

RH RH critical (Eq.3-1)

Rates of HAC and APC are represented by the same cumulative distribution function. HAC is assumed to occur uniformly over each patch used in the WAPDEG code. Each patch is comparable in size to that of a LTCTF test sample.

As discussed in the AMR Environment on the Surface of the Drip Shield and Waste Package Outer Barrier (CRWMS M&O 2000a, Section 6.4.2), hygroscopic salts may be deposited on the EBS components by aerosols and dust entrained in ventilation air, backfill, seepage water that enters the drifts, and the episodic water that flows through the drifts. Hygroscopic salts enable aqueous solutions to exist at relative humidities below 100%. The threshold RH (RHcritical) at which an aqueous solution will form for a particular salt is defined as the deliquescence point. This threshold RH defines the condition necessary for aqueous electrochemical corrosion of the metal to occur. The deliquescence point of NaCl is relatively constant with temperature, and is in the range 74-76% RH. In contrast, the deliquescence point of NaNO3 has a strong dependence on temperature, ranging from an RH of 75.36% at 20°C to 65% at 90°C. The equilibrium RH is 50.1% at 120.6°C, which is the boiling point of the saturated solution at 101.32 kPa. The evaporative concentration of well J-13 water, which is assumed to be typical of waters contacting the EBS components, results in a solution of Cl-, , and K+ ions. Other ions that could form salts with lower deliquescence points, such as Ca2+ and Mg2+, are precipitated. It is therefore conservatively assumed that the deliquescence point of NaNO3 determines the threshold RH. The equilibrium RH for a saturated solution of NaNO3 as a function of temperature is shown in Figure 3-3. The experimental data fit the following polynomial in temperature:

(Eq. 3-2)

The goodness of fit is characterized by

R2 = 0.9854

where R2 is the coefficient of determination and where R is the coefficient of correlation. This correlation is shown in Figure 3-3 below. The uncertainty in RHcritical is discussed in the AMRs on surface environment.

The evaporation of J-13 water results in high concentrations of Na+, K+, Cl-, NO3-, and CO32-. The concentrations of F- and SO42- initially increase, but eventually fall due to precipitation. The SSW used for testing is an abstract embodiment of this observation (Section 1.5.4.3). This formulation is based upon the assumption that evaporation of J-13 water will eventually lead to a sodium-potassium-chloride-nitrate solution. The elimination of carbonate in this test medium is believed to be conservative, in that carbonate would help buffer pH in any occluded geometry such as a crevice.

3.1.3.2 Aqueous Phase Environments

At a given surface temperature, the existence of liquid-phase water on the WP depends upon the nature of the hygroscopic salt either present on the surface or contained in water dripping on the surface. Two conditions must exist for APC: dripping water, and RH above the deliquescence point of the hygroscopic salts in the dripping water. While dripping can occur without the latter condition being met, both conditions are necessary for APC. Without this level of RH, no aqueous phase could be sustained on the surface. However, this requires that the evaporation rate of water from the surface exceeds the rate of dripping so that equilibrium conditions exist (CRWMS M&O 2000a).

This model uses Equation 3-2 to conservatively estimate the threshold RH for APC (RHcritical). The composition of the electrolyte formed on the WP surface is assumed to be that of SCW below temperatures of 100°C, and that of SSW above temperatures of 100°C. These media are defined in Section 1.5.4.3. Their compositions are shown in Table 3-2 (CRWMS M&O 2000a, Section 6.12). General APC is assumed to occur uniformly over each WAPDEG patch, which is the same size as a standard LTCTF weight-loss sample. Effects of backfill are not considered on the APC threshold and rate.

3.1.3.3 Condensation Underneath Drip Shield

Moist air and liquid water flow into and within the drift over time. Although the RH underneath the DS increases with time, conditions for condensation on the DS can only occur if the DS is cooler than the top of the invert and the invert moisture content produces nearly 100% RH. This is unlikely since the surfaces of the WP and DS are at a higher temperature than the invert.

3.1.3.4 Composition of Water on Exposed Surfaces of Drip Shield and Waste Package

The YMP has used test media relevant to the environment expected in the repository. Relevant test solutions are assumed to include SDW, SCW, and SAW at 30, 60, and 90°C, as well as SSW at 100 and 120°C (Section 1.5.4.2). The compositions of all of the environments are given in Table 3-2. While most of the solutions have been used for several years, the SSW has been recently developed. In general, anions such as chloride promote LC, whereas other anions such as nitrate tend to act as corrosion inhibitors. Thus, there is a very complex synergism of corrosion effects in the test media.

BSW represents another plausible extreme in water chemistry. The BSW composition was established on the basis of results from a distillation experiment. Tables 3-3 and 3-4, show the corresponding water chemistry. The total concentration of dissolved salts in the starting liquid was approximately five-times (5) more concentrated than that in the standard SCW solution. It was prepared by using five-times the amount of each chemical that is specified for the preparation of SCW. After evaporation of approximately ninety percent (~90%) of the water from the starting solution, the residual solution reaches a maximum chloride concentration and has a boiling point of ~112°C. The resultant BSW solution contains (sampled at 112°C) 9% chloride, 9% nitrate, 0.6% sulfate, 0.1% fluoride, 0.1% silicate, 1% (total inorganic carbon from carbonate and bicarbonate), 5% potassium, ion, and 11% sodium ion.

In order to add some soluble silica to the solution, the initial BSW solution recipe was later revised to contain ~1% metasilicate by adding sodium metasilicate (Na2SiO3•9H2O). This solution is designated as BSW-SC where SC indicates the presence of silicate and carbonate in the solution.

The pH of aqueous solutions is affected by the partial pressure of CO2 in the gas phase. The implication of this is that unless an effort is made to control the pH of the BSW solution, the pH may vary with test conditions and time. In order to conduct long-term testing (months to years), the test environments should be stable. Stable test solutions require that carbonate and silicate not be added. Both of these species can affect pH. Furthermore, gaseous CO2 must be removed from the air passing above the solution. With no gaseous CO2 in contact with the solution, and with no carbonate-bicarbonate or silicates in solution, the test environments will be stable. Sodium hydroxide is used to maintain the higher pH of the solution.

In order to maintain constant pH conditions, the BSW solution was modified for corrosion tests, yielding BSW-11, BSW-12, and BSW-13. The three solutions have pH values of approximately 11, 12, and 13, respectively. The recipes of these solutions are given in Table 3-4.

3.1.4 Phase Stability and Aging

Exposure of materials like Alloy 22 to elevated temperatures can result in the formation of undesirable phases. The phases which form in Alloy 22 are often rich in molybdenum and chromium, the two elements that are responsible for the high degree of corrosion resistance of this material. The formation of precipitates depletes these alloying elements from the surrounding areas, therefore increasing susceptibility to general and LC, as well as SCC. Formation of brittle molybdenum- or chromium-rich intermetallics can also lead to embrittlement of the material and degradation of its mechanical properties. LRO in alloys similar to Alloy 22 has been linked to an increased susceptibility to SCC and hydrogen embrittlement.

The aging of Alloy 22 is dependent on both time and temperature. While the effects of aging have been observed for exposures to elevated temperatures (>600°C) for short time periods, it is important to know the kinetics of this process to enable prediction of aging effects for lower temperatures (200-300°C) and much longer times (10,000 years).

This section discusses the process-level model developed to account for aging and phase stability in Alloy 22. The development of this model is presented in detail in the corresponding AMR (CRWMS M&O 2000b). Only the highlights will be presented here.

3.1.4.1 Phase Identification in Alloy 22

The long-term aging of Alloy 22 at elevated temperature can cause the precipitation of undesirable intermetallic phases, if the temperature is sufficiently high. In order to provide a technical basis for the development of a model for aging effects in Alloy 22, samples were aged for a variety of times at different temperatures: for 40,000 hours at 260, 343, and 427°C; for 30,000 hours at 427°C; for 1000 hours at 482, 538, and 593°C; and for 16,000 hours at 593, 649, 704, and 760°C. Samples were then examined with transmission electron microscopy (TEM). A weld sample aged at 427°C for 40,000 hours was also examined in the weld metal, in the HAZ, and in the base metal removed from the weld. Several phases were observed to form in Alloy 22: P, µ, , carbide, and Ni2 (Cr, Mo) LRO. At 593°C, P phase was observed only on the GB. At the higher aging temperatures (649, 704, and 760°C), both µ and P phases precipitated on grain boundaries. As the aging temperature increased, more µ and P phase precipitation occurred within the grains. GB carbide precipitation was observed in samples aged at 593 and 704°C. Because of the small amount of carbide present in these samples and the small volume examined in TEM, it is likely that carbides also form at 649°C. A phase was observed in the samples aged at 704 and 760°C. The amount of phase observed in these samples was small compared to the amount of µ and P phases. Long range order (LRO) was observed in the samples aged at 593°C for 16,000 hours and for 1,000 hours, in the sample aged at 538°C for 1,000 hours, and in the samples aged at 427°C for 40,000 hours and for 30,000 hours. These observations are summarized in Table 3-5.

3.1.4.2 Kinetics of Intermetallic Precipitation in Alloy 22 Base Metal

Table 3-6 shows the aging times for the various stages of intermetallic precipitation in Alloy 22 base metal as a function of temperature. These times were approximated from the examination of aged samples after approximately 1, 10, 100, 1000, and in some cases 16,000 hours. The errors noted are due to the uncertainty associated with the coarse time intervals of examination, and are not due to any measurement and test equipment uncertainties, which are much smaller. For example, if precipitation was observed on twin boundaries at 100 hours, it could have begun at any time between 10 and 100 hours. In that case, the time noted for the start of precipitation on the twin boundaries would be 55 hours (halfway between 10 and 100), with upper and lower error bars of 45 hours. Because of the coarse examination intervals, there is some judgment involved in choosing the times noted in Table 3-6. These measurements are only intended as an initial estimate of the precipitation kinetics. These measurements were also used to generate the isothermal time-temperature-transformation (TTT) diagram for Alloy 22 base metal shown in Figure 3-4. Some of the data presented in Table 3-5 are also shown in Figure 3-4. The curve associated with LRO came from TEM observations. Only a limited number of samples were examined in TEM; therefore, it is likely that ordering occurs at shorter times than indicated in Figure 3-4. The precipitation of intermetallic phases at grain boundaries in Alloy 22 is shown in Figure 3-5 as the white phase surrounding the grains.

Nucleation and growth kinetics can be represented by an equation of the form:

(Eq. 3-3)

where f is the volume fraction of the precipitating phase, t is time and k and n are constants. The value of k depends on nucleation and growth rates, and, therefore, depends very strongly on temperature. This dependence is shown in Equation 3-4:

(Eq. 3-4)

where C1 and C2 are constants, and T is the absolute temperature in Kelvins. Combining Equation 3-3 and Equation 3-4 at constant volume fraction yields:

(Eq. 3-5)

where tf is the time to reach a given volume fraction of GB precipitation. Plots of logarithm of time versus reciprocal temperature for the various stages of precipitation in Alloy 22 base metal are shown in Figure 3-6. At the higher temperatures, GB precipitation is predicted to start after 1 hour, which is the shortest aging time investigated thus far.

If it can be assumed that the precipitation mechanism does not change, the lines in Figure 3-6 can be extrapolated to give the times required for the various stages of precipitation at lower temperatures. The measured times are based on examination of micrographs of samples with widely spaced aging times. Extrapolation to lower temperature is difficult since the precipitation rate is very sensitive to temperature. A small change in slope can make a very large change in the time obtained from extrapolation to low temperature. In order to make a bounding argument, however, the curves associated with GB coverage and bulk precipitation in Figure 3-6 are graphically extrapolated to 10,000 years in Figure 3-7. The start of GB precipitation is not plotted because of the limited amount of available data. It must be noted that all data is regarded as preliminary at the present time. Additional work is needed.

The horizontal axes in Figure 3-7 are reciprocal temperature; temperature increases to the left. If an extrapolation of the data intersects the horizontal line corresponding to 10,000 years to the left of the vertical line corresponding to 300°C, then the temperature must be held higher than 300°C to get bulk precipitation in 10,000 years. In both cases, the data indicate that intermetallic precipitation will not occur in less than 10,000 years, even if the temperature is held at 300°C. Also plotted in Figure 3-7 are lines with the minimum possible slope allowed by the error bars on the data. Even accounting for the rather large uncertainty, bulk precipitation does not appear likely in 10,000 years at 300°C. However, GB precipitation might occur if the Alloy 22 stayed at 300°C for 10,000 years. Therefore, a good bounding argument would be to assume that precipitation covers the grain boundaries. As can be seen in the TTT diagram of Figure 3-4, samples aged in the laboratory for more than 100 hours at 700°C would produce a microstructure with precipitation covering the grain boundaries. Complete GB coverage is taken to represent a fully aged material (worst case). Corrosion data obtained from Alloy 22 base metal aged in such a way should represent a worst case condition in regard to phase stability. Corrosion data obtained with aged samples are presented in Section 3.1.4.5 and are presented in greater detail in the AMR on the WPOB (CRWMS M&O 2000c, Section 6.7).

Figure 3-7a shows calculated WP surface temperature as a function of repository storage time (CRWMS M&O 2000r). The curves shown are for the hottest (design basis) WP with and without backfill option. It can be seen that, for the case of "no backfill", WP temperatures are sufficiently low that phase stability of Alloy 22 should not be an issue.

As a measure of the reasonableness of the data plotted in Figure 3-7, the activation energy can be calculated. The slopes of the lines in Figure 3-7 (after accounting for the log(e) factor) are equal to C2/n in Equation 3-5. If these slopes are averaged and n is assumed to be equal to one, then the activation energy is 280 kJ mol-1 (68 kcal mol-1 using a gas constant R=1.987 cal mol-1K-1). This is close to the value of 62 kcal mol-1 obtained for precipitation in Alloy C-276. This value is also typical for diffusion of relevant elements in nickel. For example, the activation energy for diffusion of chromium in nickel is 272.6 kJ mol-1, that of iron is 253-270 kJ mol-1, and that of tungsten in nickel is 300-308 kJ mol-1.

3.1.4.3 Kinetics of Intermetallic Precipitation in Alloy 22 Welds

The HAZ of a weld is the region of the base metal near the weld that is subjected to a significant thermal pulse during the welding process. Intermetallic precipitation processes in the HAZ are expected to be similar to that in the base metal, but actual rates of precipitation (kinetics) may be different. The high temperatures, approaching the melting point, seen in the HAZ of welds might trigger nucleation of intermetallic carbide precipitates. If nuclei are already present, precipitation will proceed much faster than in the base metal where they are not present.

Very few precipitates have been observed in the HAZ of weld samples thus far, but only two weld samples have been examined: one in the as-welded condition and one after aging at 427°C for 40,000 hours. These precipitates may simply be carbides that were present in the mill-annealed (as-received) condition. Carbides are known to be present in nickel-based alloys similar to Alloy 22, but they are usually within the grains. These are generally called primary carbides to distinguish them from other secondary phases that often form, on the GB after an aging treatment.

Welding causes melting of the alloy and the development of an as-cast structure upon cooling. As an Alloy 22 weld solidifies, molybdenum and chromium are rejected from the solid phase causing their concentration to increase in the liquid. Therefore, the interdendritic regions, which are the last solid to form in a weld, tend to have high concentrations of these elements relative to typical values for Alloy 22. Because formation of the intermetallic phases, which are also enriched in molybdenum and/or chromium, are favored by higher molybdenum and chromium concentrations, these phases are present in the interdendritic regions of Alloy 22 welds.

Because precipitates are present in Alloy 22 welds from the beginning, kinetics of precipitation in welds is not an issue. Corrosion data available from the LTCTF shows no measurable difference between welded and base metal samples.

3.1.4.4 Kinetics of Reactions in Alloy 22

The LRO is treated in a manner similar to that discussed for intermetallic and carbide precipitation. However, very little kinetic data exists for LRO in Alloy 22. Thus far, LRO has been observed in five samples. A very fine dispersion of ordered domains was seen in Alloy 22 base metal after aging for 30,000 and 40,000 hours at 427°C, and was also seen in a weld similarly aged. The ordering in these cases was so fine that it would have been very difficult to measure the volume fraction of the ordered domains. LRO was also observed in Alloy 22 base metal aged at 593°C for 16,000 hours, and at 538 and 593°C for 1,000 hours. The volume fraction of ordered domains has not been measured in these samples. No LRO was observed with TEM in Alloy 22 base-metal samples aged for 40,000 hours at 260 and 343°C, or for 1,000 hours at 482°C.

A bounding argument may be made by using two facts: LRO is just beginning after aging for 30,000 hours at 427°C together; LRO domains are small after aging for 1,000 hours at 538°C. The corresponding points are graphed as an Arhenius plot in Figure 3-8. Samples aged for shorter times at 427 and 538°C have not yet been examined in TEM. The curve in Figure 3-8 may shift to shorter times (down) after more data are collected. Because the LRO domains are very small after aging at 427°C for 30,000 hours, the point corresponding to this aging condition is not likely to change much (Figure 3-8). In other words, LRO is not likely to occur in Alloy 22 base metal at 427°C in times significantly less than 30,000 hours. After more data are collected, the data point corresponding to aging at 538°C for 1,000 hours will most likely shift down more than that corresponding to aging at 427°C for 30,000 hours. Since this shift will cause the slope to increase, the curve in Figure 3-8 represents a bounding case. This graph indicates that LRO may occur in less than 10,000 years at 300°C. Equation 3-6 was obtained by curve fitting and is shown in Figure 3-8:

(Eq. 3-6)

Solving Equation 3-6 for temperature T yields:

(Eq. 3-7)

Based upon this analysis, it is concluded that no LRO will occur after 10,000 years (8.8107 hours), provided that the temperature remains below approximately 260°C (530 K). More samples are being tested to confirm this conclusion.

3.1.4.5 Effects of Thermal Aging on Corrosion Potential and Rate

The long-term aging of Alloy 22 at elevated temperature can cause the precipitation of undesirable intermetallic phases. Based upon the analyses discussed in the preceding section, it is recommended that the WP surface temperature be limited to levels below 300°C. An extrapolation of the data shown in Figure 3-6 indicates that the phase stability of Alloy 22 base metal will not be a problem below this limit. The significance of the uncertainties in this data is discussed in Section 3.1.9. At temperatures above 350°C, there is unacceptable degradation of cladding on the SNF. With these two constraints, the impact of aging and phase instability on the corrosion of Alloy 22 should be minimal.

Samples of Alloy 22 were aged at 700°C for either 10 or 173 hours. The corrosion resistance of these aged samples is compared to that of base metal in several standardized test media. Figure 3-9 shows a comparison of CP curves for base metal and thermally aged material in SAW at 90°C. Both curves exhibit generic Type 1 behavior (see Section 3.1.6.3). Type 1 behavior is indicative of passive film stability between the corrosion potential and the thermodynamic limit of the electrolyte (oxygen evolution). In this case, aging shifts the corrosion potential to less noble values, from -176 to -239 mV (versus a standard Ag/AgCl reference electrode). The passive current density is increased slightly, which is interpreted as a slight increase in corrosion rate. The highest non-equilibrium passive current density observed for the base metal is approximately 4 µmA cm-2, compared to approximately 10 µmA cm-2 for fully aged material.

Figure 3-10 shows a comparison of CP curves for base metal and thermally aged material in SCW at 90°C. In this case, aging also appears to shift the corrosion potential to less noble values, from -237 to somewhere between -328 and -346 mV versus a standard Ag/AgCl reference electrode. In all three cases, the anodic oxidation peak that is characteristic of generic Type 2 behavior is observed. See Section 3.1.6.3 for more detailed discussion. In tests with BSW-13, aging also appears to shift the corrosion potential to less noble values.

In summary, a fully aged sample of Alloy 22 appears to exhibit a less noble corrosion potential. Typically, the corrosion potential of such a sample is shifted approximately -63 mV in SAW at 90°C; -109 mV in SCW at 90°C; and more than -100 mV in BSW at 100°C. Based on this data, it appears that Ecorr can be corrected to account for fully aged material by subtracting approximately 100 mV from values calculated for the base metal. The shift in Ecritical (Threshold Potential 1) also appears to be approximately 100 mV in most cases. Thus, the difference Ecritical-Ecorr is virtually unchanged. This implies that even though the corrosion potential is shifted, the susceptibility to LC remains unchanged.

The effect of thermal aging on the corrosion rate is accounted for in an enhancement factor, Gaged, and is based upon a ratio of the non-equilibrium passive current densities for base metal and aged material.

(Eq. 3-8)

where dp/dt is the penetration rate for LC. The value of Gaged for Alloy 22 base metal is approximately one (Gaged ~ 1), whereas the value of Gaged for fully aged material is larger (Gaged ~ 2.5) (CRWMS M&O 2000c). Material with less precipitation than the fully aged material would have an intermediate value of Gaged (1 Gaged 2.5). Therefore, a value of 2.5 for Gaged is conservatively used to bound the potential aging effect. Corrosion is discussed in greater detail in the sections that follow.

3.1.5 General Corrosion

The integrated model for general corrosion of the materials of interest includes sub-models for dry oxidation, humid air and aqueous phase corrosion. Details of these sub-models are provided in supporting AMRs (CRWMS M&O 2000c, 2000d, and 2000e). A schematic representation of the integrated model for WP and DS materials, as well as an augmentation of this model, are shown in Figures 3-11 and 3-12. Only a brief summary of the integrated model and component sub-models is provided here.

3.1.5.1 Dry Oxidation

Dry oxidation (DOX) occurs at any RH below the threshold for HAC:

(Eq. 3-9)

This process results in the formation of an adherent, protective oxide film of uniform thickness. The rate of DOX may be limited by the rate of mass transport through the growing metal oxide film. In such cases, the oxide thickness is expected to obey a parabolic growth law (film thickness proportional to the square root of time). This scenario has been adopted for Alloy 22 and 316NG due to the availability of data at elevated temperature to support such a model. Reasonable values of the parabolic rate constant are discussed below. It must be noted that a logarithmic law may be more applicable at lower temperatures. However, there is insufficient data to support such a model for Alloy 22 and 316NG. There is sufficient data to support the application of a logarithmic law to the DOX of titanium. It is assumed that DOX occurs uniformly over each WAPDEG patch, which is comparable in size to that of a LTCTF sample with generic weight-loss geometry. Backfill effects are not included in the model for DOX threshold and rate.

3.1.5.1.1 Dry Oxidation of Alloy 22 and 316NG

DOX of Alloy 22 and 316NG stainless steel is expected to occur at any RH < RHcritical, thereby forming an adherent, protective oxide film of uniform thickness. It is assumed that the protective oxide film is primarily Cr2O3. The oxidation reaction is given as

(Eq. 3-10)

The rate of DOX is limited by mass transport through this growing metal oxide film with the film thickness being proportional to the square root of time. This is represented by Equation 3-11:

(Eq. 3-11)

where x0 is the initial oxide thickness, x is the oxide thickness at time t, and k is a temperature-dependent parabolic rate constant.

To facilitate an approximate calculation, published values of k can be used (CRWMS M&O 2000c, Sections 1.6 and 6.10). The highest WP temperature in the repository is expected to be approximately 350°C (623 K), which corresponds to the limit for the SNF cladding. Note that the cladding will be hotter than the WPOB. The value of k corresponding to this upper temperature limit is 2.73 10-24 m2 s-1 (8.6110-5 square µm per year). After one year, this corresponds to a growth of about 0.0093 µm. The estimated rate (9.3 nm y-1) is comparable to that expected for APC at lower temperatures, based upon data presented in this PMR. The parabolic law is used to represent the DOX Alloy 22 and 316NG, and is relatively conservative.

3.1.5.1.2 Dry Oxidation of Titanium Grade 7

As discussed in the AMRs for general and LC of the DS (CRWMS M&O 2000d, Section 6.1), the logarithmic growth law may be more appropriate at low temperature than the parabolic law. However, such a logarithmic expression predicts that the oxide thickness (penetration) asymptotically approaches a small maximum value. In contrast, the parabolic law predicts continuous growth of the oxide, which is much more conservative. Figure 3-13 shows a regression analysis of DOX rate data for titanium, where X is the oxide thickness at time t and Xmax is the maximum oxide thickness.

3.1.5.2 Humid Air Corrosion

HAC is assumed to occur above a threshold RH, provided that there are no impinging drips:

(Eq. 3-12)

The threshold RH for HAC (RHcritical) is assumed to obey Equation 3-2. Note that "threshold RH" and "critical RH" are synonymous terms. The existence of this threshold is due to the relationship between water adsorption and RH.

It can be conservatively assumed that the rate of HAC can be represented by the same corrosion rate distribution used for APC during the period where HAC is operable. It is further assumed that the corrosion rate is constant and does not decrease with time (at times greater than two years). Less conservative corrosion models assume that the rate decays with time. The rates for APC of stainless steel 316NG, Alloy 22, and Titanium Grade 16 (analog of Titanium Grade 7) are described in detail elsewhere in this section.

3.1.5.3 Aqueous Phase Corrosion

At a given temperature, the existence of liquid-phase water on the surface of the WP depends upon the presence of a salt deposit. In the presence of such a deposit, a thin-film liquid phase can be established at a higher temperature than otherwise possible. In the model discussed here, it is assumed that two conditions must be met for APC: RH above the deliquescence point of the deposit at the temperature of the WP surface and drips impinging on the WP surface. The threshold RH for APC is identical to that for HAC:

(Eq. 3-13)

This threshold RH for APC (RHcritical) is also assumed to obey Equation 3-2, which is based upon the AMR entitled Environment on the Surface of Drip Shield and Waste Package Outer Barrier (CRWMS M&O 2000a, Section 6.4.2). For the time being, the composition of the electrolyte formed on the WP surface is assumed to be that of SCW below 100°C, and that of SSW above 100°C. The distributions of GC rates for APC of stainless steel 316NG, Alloy 22, and Titanium Grade 16 (analog of Titanium Grade 7) are described in detail in this PMR. It is conservatively assumed that the corrosion rate is constant and that it does not decrease with time (at times greater than two years).

3.1.5.4 Rates of General and Localized Corrosion

Penetration rates based upon GC are used in the model if the threshold potential (Ecritical) is not exceeded. GC rates have been estimated based on the weight-loss data from the LTCTF. LC rates and failure mode characteristics (e.g., number of failure sites and opening size) must be estimated from other published data. Since pitting of Alloy 22 or Titanium Grade 16 has not been observed in the LTCTF, it is assumed that crevice corrosion is the primary mode of LC (if localized attack occurs at all). This accelerated mode of attack is assumed to occur uniformly over the entire affected WAPDEG patch. Uncertainty is accounted for in the WAPDEG stochastic simulation.

3.1.5.4.1 Rates for 316L Stainless Steel – Published Data

The samples involved in the YMP corrosion testing program do not include stainless steel specimens. Therefore, data published in the literature must be used to determine corrosion rates. Rates of LC are used if the threshold potential is exceeded. Otherwise, GC rates are used. The distribution of general and LC rates were estimated from the published data for Types 304 and 316 stainless steels presented in the AMR on degradation of the stainless steel structural material (CRWMS M&O 2000e, Section 6.5). These data are shown in Figures 3-14 through 3-17. [3-14, 3-15, 3-16, 3-17] Curves are shown for GC in atmospheric and aqueous-phase environments, as well as for LC in aqueous-phase environments. The corresponding distributions can be represented by the following general correlation:

(Eq. 3-14)

where y is the cumulative probability or percentile, and x is the logarithm of the corrosion rate, which is expressed in microns per year. Parameters are given in Figure 3-15. These cumulative distribution functions are truncated for any nonsensical calculated values above one-hundred percent (100%). These distributions do not reflect any environmental dependence, since such a correlation could not be established based upon published data. It is assumed that these distributions are primarily due to variability. In lieu of rigorous estimates of uncertainty, the uncertainty is assumed to be comparable to the variability.

From a comparison of the data for Types 304 and 316 stainless steels, the advantages of molybdenum additions are evident. Since Type 316 stainless steel contains more molybdenum than Type 304 stainless steel, the corrosion rates of Type 304 stainless steel are higher than comparable rates of Type 316 stainless steel. From Figure 3-14, the localized corrosion rates of Type 316 stainless steel appear to lie between 103 and 104 µm y-1. The general corrosion rates for APC of Type 316 appear to lie between 10-1 and 102 µm y-1. The general corrosion rates for HAC (atmospheric corrosion) of Type 316 stainless steel appear to lie between 10-3 and 10-1 µm y-1. It is assumed that the published rates for Type 316 stainless steel are representative of those for Type 316NG stainless steel. The regression line shown in Figure 3-15 is assumed to represent the combined uncertainty and variability. Figures 3-16 and 3-17 show the probability distribution of the corrosion rates and comparison of the observed penetration rates for Types 304 and 316 stainless steels, respectively.

3.1.5.4.2 Rates for Nickel-Based Alloy 22 – Weight-Loss Measurements from Long Term Corrosion Test Facility

The Long-Term Corrosion Test Facility (LTCTF) provides a relatively complete source of corrosion data for Alloy 22 in environments relevant to the potential high-level radioactive waste (HLW) repository at Yucca Mountain. The results from that facility are described in detail in the AMR on GC of Alloy 22 (CRWMS M&O 2000c, Section 6.5.2). Testing is done in a wide range of plausible media, including SDW, SCW, SAW, and SCMW. SDW has ten times (10) the ionic content of J-13 well water, while SCW has 1000 the ionic content. The measured pH levels of the 10 and 1000 J-13 waters are 9.5 to 10. SAW is acidified water that has 4000 the ionic content of J-13 water and a pH of approximately 2.7. These concentrated test solutions are intended to mimic the evaporative concentration of electrolytes on the hot WP surface. Due to solubility limitations, not all salts in the water concentrate to the same nominal levels. However, the more soluble anions such as chloride, sulfate, and nitrate (which have the biggest effects on corrosion) do reach nominal levels.

Specimens are tested at two temperatures (60 and 90°C) in each water chemistry. Half of the specimens are fully immersed, while the remaining half are exposed to the wet vapor above the water line. A few specimens are placed right at the water line. Half of the specimens contain welds, while the remaining half are unwelded.

Crevice specimens are mounted to support racks with Teflon-coated fasteners and washers. Teflon crevice-forming washers are spring loaded to ensure that contact is maintained between the washers and specimens (crevice effects are more severe in tight crevices).

At least 144 test specimens are measured during each exposure period (6-, 12-, and 24-month, thus far). The general corrosion rates are determined from weight-loss measurements made at the end of predetermined exposure periods. These measurements are based upon ASTM G 1-81 Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens or the more recent ASTM G 1-90, Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens. All specimens are cleaned in accordance with applicable American Society for Testing and Materials (ASTM) procedures prior to making weight and dimensional measurements. Uncertainty in the measured rates decreases with increasing exposure time. Details of the facility, sample configurations, and the procedure used for handling the samples are prov