Return to the Main Document

(Section 4, continued)

4.2.4 Waste Package and Drip Shield Degradation

The roles of the waste package and drip shield are discussed in detail in Section 3. This section addresses the expected performance of these components in the potential repository and, along with Sections 4.2.1 through 4.2.6, [4.2.1, 4.2.2, 4.2.3, 4.2.4, 4.2.5, 4.2.6] provides an explanation of the relationship of the waste package and the geologic environment at Yucca Mountain.

The degradation process models and the abstracted models discussed in this section serve as feeds to the WAPDEG code, which integrates the various models to address the overall performance (degradation rates) of the waste package and the drip shield. WAPDEG results, in turn, are used as feeds to the overall TSPA-SR. Specifically, the integrated model included in the WAPDEG performance assessment code used repository environmental conditions as a function of time from other process models to estimate the performance of the waste package and drip shield in terms of time to failure.

As noted in Section 4.1.4, the DOE is evaluating operation of the repository at lower temperatures. The conceptual basis and model abstractions presented in this section reflect the effects of higher-temperature operating modes, specifically those implemented in Total System Performance Assessment for the Site Recommendation (CRWMS M&O 2000a, Section 3.4). Alternative thermal operating modes and/or conservatisms and conceptual uncertainties have been reevaluated since the TSPA-SR model and are reported or summarized in FY01 Supplemental Science and Performance Analyses (BSC 2001a, Section 9; BSC 2001b, Sections 3 and 4).

4.2.4.1 Conceptual Basis

Lifetimes of the drip shield and waste package depend on the environmental conditions to which they are exposed and the degradation processes that occur in that environment. Section 4.2.3 describes the conceptual understanding of the evolution of physical and chemical conditions in the repository emplacement drifts, the models used to represent those conditions, and the experimental data that support and contribute to the validation of the models. Environmental conditions within the drifts that influence the degradation of the waste package and drip shield are tightly coupled to the thermal-hydrologic and geochemical processes occuring in the rock surrounding the drifts. These processes involve the vaporization and condensation of water under changing thermal conditions, redistribution and precipitation of dissolved salts, and the effects of gaseous species on solution chemistry. Included in the conceptualization are the contributions of construction material degradation processes (i.e., rock structural support materials and cementitious grout) and the effects of microbial action.

Once the exposure environments have been established, the most important and relevant degradation processes can be identified, which in turn can be used for selecting engineered materials for the drip shield and the waste package. This section discusses the degradation modes of the waste package and drip shield materials under the changing environmental conditions. Corrosion is the degradation process most relevant and important to the selection of the materials for the waste package and drip shield. Mechanical deformation of the waste package and drip shield are estimated to be less significant to the waste package containment time than corrosion (CRWMS M&O 2000n, Section 1.5). A number of corrosion processes have been investigated in detail and the results used to support the selection of materials and the design of these components.

Waste Package and Drip Shield Materials—Degradation modes for the drip shield and waste package are dependent on the materials used in these components and as mentioned earlier, on the environment in which they function. Performance of these materials are reviewed in General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq) and General Corrosion and Localized Corrosion of the Drip Shield (CRWMS M&O 2000cr). Titanium alloys were selected for construction of the drip shield because of their high resistance to corrosion. This corrosion resistance is due to the formation of a passive oxide film, which is stable over a relatively wide range of environments. The rates of general corrosion and dry oxidation (or dry-air oxidation) of this material have been shown to be very low (CRWMS M&O 2000n, Sections 3.1.1.1, 3.1.5.1, and 3.1.5.4).

Alloy 22 (UNS N06022) was selected for construction of the waste package outer barrier. The main alloying elements of this material are nickel, chromium, molybdenum, iron, tungsten, and cobalt. Alloy 22 is less susceptible to localized corrosion in environments that contain chloride ions than Alloys 825 and 625, materials of choice in earlier waste package designs (CRWMS M&O 2000n, Section 3.1.1.2). This material is one of the most corrosion-resistant nickel alloys for the expected range of repository environments (Gdowski 1991, Section 1.2.5). Alloy 22 and its predecessor alloys have been in use for the past 50 years in a variety of environments and have performed extremely well. Figure 4-79 shows the appearance of a test coupon made from Alloy C, which is a predecessor of Alloy 22, after almost 60 years of exposure to a marine environment. Its shiny, mirror-like appearance was restored by rinsing the dirt and sand from the surface. In comparison to Alloy C and C-4, Alloy 22 has greater corrosion resistance. This is based on the fact that Alloy C-4 and Alloy C-276 have a comparable corrosion resistance (Gdowski 1991, Section 1.2.4), and resistance of Alloy 22 to crevice corrosion is greater than Alloy C-276 (Gdowski 1991, Tables 22 and 25).

Stainless Steel Type 316NG will be used for construction of the structural support container inside the waste package outer barrier to increase the overall strength of the waste package. This material is less susceptible to localized corrosion in environments that contain chloride ions than stainless steel 304, but it is more susceptible than other corrosion-resistant materials such as Alloys 22, 625 and 825, which were considered in various waste package designs (CRWMS M&O 2000n, Section 3.1.1.3). However, the stainless steel layer is used primarily for structural support for the outer barrier and not as a corrosion barrier to the ingress of water into the waste package. The key factor in placing the structural material on the inside is that its strength does not begin to degrade until the outer shell is breached by corrosion or other degradation modes. This is in contrast to the VA design in which the structural carbon steel was the outer shell, with degradation of strength beginning soon after repository closure.

Figures 4-80 and 4-81 provide a visual perspective to illustrate the physical arrangement of the waste packages and the drip shield within the drift. Figure 4-80 shows schematically the arrangement of different types of waste packages and drip shield. Figure 4-81 shows a schematic sketch of a typical waste package designed for 21 pressurized water reactor fuel assemblies, along with the materials used for the various components.

The dual-barrier design of the waste package and the number of options for the thermal design of the potential repository required a comprehensive testing program to evaluate how materials would perform under the wide range of possible conditions in the potential repository.

Degradation processes evaluated for the drip shield and the waste package include general and localized corrosion under humid air and aqueous conditions, stress corrosion cracking, and hydrogen-induced cracking. The effects of microbially influenced corrosion and aging of the waste package outer barrier were also included in the modeling. An integrated model was developed to evaluate the combined effects of the various degradation modes and was used to estimate the range of lifetimes of the drip shields and the waste packages, including an evaluation of uncertainties.

4.2.4.2 Summary State of Knowledge

Surface Environment—The starting waters present at Yucca Mountain are classified into two types: (1) bicarbonate-type water (e.g., J-13 water) and (2) unsaturated zone pore water (chloride-sulfate water). Chemical modeling and laboratory testing of these water compositions have shown that the bicarbonate-type water evolves by evaporative concentration to a high-pH brine, whereas the chloride-sulfate-type water evolves to a brine with nearly neutral pH (CRWMS M&O 2000ck, Section 6.1).

In the model used, hygroscopic salts on the drip shield or waste package formed due to evaporation of the dripping water will be limited to sodium or potassium salts. As a bounding condition, it is assumed that sodium nitrate—the salt with the lowest deliquescence point at elevated temperature—will determine the minimum relative humidity at which aqueous conditions can occur (CRWMS M&O 2000ck, Section 6.4.2).

The exact chemistry of the water that contacts the drip shield and waste package surfaces cannot be known precisely. However, the range of potential types of aqueous solutions has been estimated from the range of potential starting water compositions, from knowledge of near-field and in-drift processes that can alter these compositions, and from laboratory experiments and natural analogue observations. From these results, a range of water compositions was developed and is being used for corrosion testing; the range also includes the potentially important effects of processes at the engineered component surfaces.

Composition of Aqueous Solutions Used for Corrosion Testing—Test solutions developed for laboratory corrosion testing of titanium, Alloy 22, and other materials were selected to represent a range of dilute and concentrated conditions, pH, and temperatures that could result from evaporative concentration in the potential repository (CRWMS M&O 2000ck, Section 6.12). The test solutions include the following:

Corrosion tests were conducted at Lawrence Livermore National Laboratory on specimens of waste package and drip shield materials, including titanium and Alloy 22, exposed to the above environments. The tests were comprehensive and examined many forms of corrosion: pitting, stress corrosion cracking, galvanic corrosion, corrosion in crevices, and general corrosion. In addition, some long-term tests had counterpart shorter-term tests for stress corrosion cracking and galvanic corrosion. The results of the tests were used for developing various corrosion models to predict the long-term behavior of the materials in the repository.

Long-Term Corrosion Testing—Test environments for long-term corrosion testing were structured to simulate concentrated solution conditions. The samples were exposed in the aqueous phase, in the vapor phase above the solutions, and at the waterline (DOE 1998, Volume 2, Section 5.1.4.1). To date, some 13,000 specimens, including welded samples, have been tested in water for periods from 6 months to 2 years in duration. The first materials tested were arranged in three categories: corrosion-allowance materials, intermediate corrosion-resistance alloys, and corrosion-resistant alloys (DOE 1998, Volume 2, pp. 5-44 to 5-45). The specimen designs and test procedures were based on specifications developed by the American Society for Testing and Materials, including specifications ASTM G 1-90, Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens, ASTM G 30-94, Standard Practice for Making and Using U-Bend Stress-Corrosion Test Specimens, and ASTM G 46-94, Standard Guide for Examination and Evaluation of Pitting Corrosion. Figure 4-82 shows the arrangement of the test specimens in the racks, and Figure 4-83 shows the typical appearance of the specimens after 12 months of exposure in the corrosion test facility.

Complementary Short-Term Tests—The long-term corrosion tests were a key component for evaluating waste package materials. They were complemented by short-term tests designed to develop mechanistic understanding. For example, short-term tests measured the relative susceptibilities of the candidate materials to general, localized, and microbially influenced corrosion. The short-term tests were important for developing models of corrosion behavior.

Field Test Assessment—Field tests provide an independent confirmation of the performance of materials in the Yucca Mountain environment. Specimens included in the accelerated thermal field tests were characterized to determine how candidate materials degrade with exposure to actual field environments at Yucca Mountain under repository temperature conditions. The specimens are subjected to changing environmental conditions, including temperature, relative humidity, and possible intermittent water contact. In comparison, laboratory experiments are being run at fixed and constant environmental conditions (DOE 1998, Volume 2, Section 5.1.4.1), and the laboratory experiments also include bounding or conservative aggressive environments.

In each field study, test specimens were placed in drifts and boreholes that were well characterized with respect to temperature and relative humidity (DOE 1998, Volume 2, Section 5.1.4.1). The data on corrosion generated from the field tests were used in activities related to performance assessment, materials selection, model development, and potential repository design.

The materials selected for the drip shield (titanium) and the waste package outer barrier (Alloy 22) are highly corrosion-resistant. Based on literature and prior industrial experience, these materials are not expected under repository exposure conditions to be subject to degradation processes that could lead to failure in a short time period (Gdowski 1991, pp. 1 to 3; Gdowski 1997, pp. 38 to 39). Those degradation modes are localized corrosion (pitting and crevice corrosion), stress corrosion cracking, and hydrogen-induced cracking (applicable only to the drip shield). Both the drip shield and waste package degrade by general corrosion at very low rates. The current experimental data and detailed process-level analyses also indicate that the candidate materials would not be subject to rapidly penetrating corrosion modes under the expected repository conditions. The exception to this is the closure-lid welds of the waste package, where unmitigated residual stresses could potentially lead to stress corrosion cracking. To preclude this occurrence, weld residual stress mitigation processes will be implemented on each of the dual-lid Alloy 22 waste package closure welds. As a result, the estimated long lifetime of the waste packages in the current analyses is attributed mostly to two factors: (1) stress mitigation in the dual closure lid weld area and (2) the low general-corrosion rate, which will very slowly remove the beneficial compressive stress zones at each weld surface, thereby providing a long delay before stress corrosion cracking can potentially initiate and grow to penetrate the weld thickness.

Models were created to predict the performance of the various materials in the expected repository environment. The modeling effort served the following two major purposes:

  1. It supported the selection of materials for which a key selection criterion was the predictability of the component material performance.

  2. It furnished information about how the selected materials would perform in the repository environment. Consistent with ASTM C 1174-97, Standard Practice for Prediction of the Long-Term Behavior of Materials, Including Waste Forms, Used in Engineered Barrier Systems (EBS) for Geological Disposal of High-Level Radioactive Waste, relevant data from testing activities were interpreted with a mechanistic understanding of how materials behave.

4.2.4.3 Process Model Development and Integration

This section addresses waste package and drip shield material degradation models and their abstractions and alternative models considered for the waste package and drip shield lifetimes. Degradation modes for the components are discussed and the dominant modes determined. Available test data are summarized. The section concludes that the modes of waste package and drip shield degradation (i.e., corrosion) and their dependence on expected thermal-hydrologic and geochemical conditions are sufficiently understood and conservatively captured in the model abstractions to support evaluation of postclosure performance.

Waste Package Material Degradation Modes—The degradation processes were selected for modeling on the basis of an extensive review of available information on the candidate materials for the waste package and drip shield. These processes have been documented in a number of degradation mode surveys (Gdowski 1991; Gdowski 1997). In addition, a review and analysis of features, events, and processes relevant to the degradation of the waste package and drip shield was recently completed and is described in Section 4.3. The degradation models provided a quantitative analysis of early failure of waste packages. The degradation models also calculated the range of expected degradation histories of both waste packages and drip shields. These models consisted of several individual process models or analyses and associated abstraction models. Figure 4-84 shows the elements of each process model, associated abstraction models, and the interrelationships among the various process and abstraction models (CRWMS M&O 2000n, Figure 1-4). The process models for general and localized corrosion of the waste package outer barrier include dry oxidation, humid-air corrosion, and the expected environment on the surface. In addition, the figure also shows how the process models feed the integrated model for waste package and drip shield degradation (WAPDEG), which is used to predict the lifetimes of these two components. Details of the process models and abstraction models are presented in the following analysis model reports and calculations (which also document evaluation of the applicable modes of degradation of the waste package and the drip shield and their models):

Figure 4-85 shows the model confidence foundation, along with the inputs and outputs for the various degradation process models. This figure shows the critical input parameters for the degradation mechanisms and data inputs and sources, which provide lifetimes for the components. Confidence in the overall model is based on the comprehensive nature of the input parameters, data, and degradation mechanisms considered in the overall performance assessment. Highlights of the waste package degradation process models are addressed in the following subsections.

4.2.4.3.1 Mechanisms for Early Failures

There is a potential for early failures of the waste package due to material defects and waste package fabrication processes, including welding. The probability of waste package fabrication defects, the uncertainty and variability of those defects, and the consequences of the defects on waste package failure times (e.g., number of potential failure sites and flaw-size distribution) were assessed.

A literature review was performed to determine the rate of failure from manufacturing defects for various types of welded metallic containers. Types of components examined included boilers and pressure vessels, nuclear fuel rods, underground storage tanks, radioactive cesium capsules, dry-storage casks for spent nuclear fuel, and tin-plate cans. In addition to providing examples of the rate at which defective containers occur, this information provided insight into the various types of defects that can occur, and the mechanisms that cause defects to propagate to failure (CRWMS M&O 2000cs, Section 6.1).

The fraction of the total population that failed due to defect-related causes during the intended lifetime of the component was generally in the range of 10-3 to 10-6 per waste package (equivalent to 1 component in 1,000 to 1,000,000 components). In most cases, defects that led to failure of the component required an additional stimulus to cause failure (i.e., the component did not fail immediately when it was placed into service). In fact, there were several examples that indicated that even commercial standards of quality control could reduce the rate of initially failed components well below 10-4 per package (or 1 out of 10,000 components) (CRWMS M&O 2000cs, Section 7).

The literature review identified 11 generic types of defects that could cause early failures in the components examined (CRWMS M&O 2000cs, Section 6.1.7). These were:

  1. Weld flaws

  2. Base metal flaws

  3. Improper weld material

  4. Improper heat treatment

  5. Improper weld flux material

  6. Poor weld joint design

  7. Contaminants

  8. Mislocated welds

  9. Missing welds

  10. Handling and installation damage

  11. Administrative error resulting in an unanticipated environment.

Four of these defect types were not considered applicable to the waste package: improper weld flux material, poor joint design, missing welds, and mislocated welds. This conclusion was based on the processes to be used for welding and the process qualification and weld inspection programs that will be implemented (CRWMS M&O 2000cs, Section 7).

The analysis also estimated the probability that specific defect types would occur on a given waste package, considering the expected preventive quality controls. The analysis applies to those defects for which probabilities are estimated using event sequence trees, namely: drip shield emplacement error, waste package handling error, waste package 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 only human error probability uncertainties. This analysis is much more rigorous and mathematically defensible in comparison to the prior analyses conducted for the Viability Assessment.

Results of the analysis for the remaining seven types of defects showed that, with the exception of the administrative error category, all the defect probabilities were less than 10-4 per waste package. This is equivalent to less than one waste package failure out of 10,000 in the entire repository. The administrative error defect rate will be reduced through implementation of stringent administrative controls (CRWMS M&O 2000cs, Section 7).

Subsequently, in reevaluating the potential of early failure mechanisms and their potential consequences, a more conservative approach resulted in the inclusion of improper heat treatment and subsequent failure of a few waste packages in the supplemental TSPA analysis. To ensure that the potential consequence of early waste package failures is treated conservatively, it is included in the nominal scenario, not as a sensitivity analysis (BSC 2001a, Section 7.3.6).

4.2.4.3.2 Aging and Phase Stability of the Waste Package Outer Barrier

Alloy 22, a nickel-based alloy, has excellent corrosion resistance and good strength and ductility. Under certain conditions, changes in the internal structure of the material can degrade its corrosion resistance and/or ductility. Three general areas were addressed in the models: complex phases, ordering, and welds.

Complex phases are known to form in Alloy 22 at temperatures above approximately 600°C (1,100°F). They can presumably form at lower temperatures, but it would take much longer than is typically observed in the laboratory because these types of changes require thermal energy and occur more slowly at lower temperatures. The rate at which these phases form in Alloy 22 was measured at temperatures above 600°C (1,100°F) as a function of temperature. Extrapolation of this rate relation to 300°C (570°F), the expected peak temperature in the emplacement drift, indicates that the rate of complex phase formation would be very slow under potential repository conditions and would have an insignificant contribution to corrosion. To bound this effect, however, a worst case was defined for the fully aged material with complete coverage of the grain boundaries with precipitates (CRWMS M&O 2000n, Section 3.1.4.2). Figure 4-86 shows the effects of aging on precipitation in the grain boundaries.

Samples with this worst-case structure were created and provided for corrosion testing. Samples of Alloy 22 were aged at 700°C (1,292°F) 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 using cyclic polarization technique. The results of this testing showed that the fully aged material exhibited a slightly higher corrosion rate than the unaged material. The maximum increase in corrosion rate was by a factor of 2.5 over the unaged sample. The results are described in greater detail in General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq, Section 6).

The second general area is ordering. Commercially available Alloy 22 is a single-phase alloy. At temperatures below approximately 600°C (1,100°F), the atoms slowly transform from a somewhat random arrangement into an ordered pattern, with the highest rates of transformation at higher temperatures below the 600°C (1,100°F) temperature limit. This process results in the rearrangement and segregation of alloying elements into specific locations in the crystal structure. This is a slow process, occuring over a long period of time. Ordering may affect the mechanical properties of alloy systems. For example, ordering can increase the strength of the alloy and reduce its ductility, which also decreases its resistance to stress corrosion cracking and hydrogen embrittlement. Because ordering occurs only at lower temperatures and because these reactions are slow, the data for the rate of ordering in Alloy 22 are limited. However, based on the available data, ordering will not become a problem, provided the temperature does not go above about 260°C (500°F) for significant periods of time (CRWMS M&O 2000n, Section 3.1.4.4). Testing will continue into the performance confirmation period to more accurately predict the performance of this alloy in the potential repository environment.

The third general area is welds. Complex phases form during welding of Alloy 22 and are present in the as-welded condition. These phases and segregation in the weld cause the welds to have lower corrosion resistance than the base metal. Work to determine the effects that long-term exposures to repository-relevant temperatures have on the properties of Alloy 22 welds is ongoing (CRWMS M&O 2000n, Section 3.1.4.3).

The aging times for the various stages of intermetallic precipitation in Alloy 22 base metal as a function of temperature were obtained from the examination of aged samples after approximately 1, 10, 100, 1,000, and in some cases 16,000 hours (CRWMS M&O 2000n, Section 3.1.4.2). These measurements are only intended as an initial estimate of the precipitation kinetics. These observations were used to generate the isothermal time-temperature transformation diagram for Alloy 22 base metal shown in Figure 4-87.

The long-term aging of Alloy 22 at elevated temperatures can cause the precipitation of undesirable intermetallic phases if the temperature is sufficiently high. The data shown in Figure 4-87 do not indicate that the phase stability of Alloy 22 base metal will be a problem at less than about 300°C (572°F). It is expected that the waste package surface temperature will stay below 300°C (572°F) in the emplacement drift (CRWMS M&O 2000n, Section 3.1.4). While this estimate is bounding, it is based on limited data. The analysis will be further refined and improved as additional data and analysis become available.

For comparison, estimated temperature of the waste package surface as a function of time in the repository is shown in Figure 4-88 for repository designs with and without backfill. This figure shows that even for the hottest waste package containing design basis spent fuel waste the peak temperature does not exceed 250°C (482°F) for the design with no backfill. The sharp increase in surface temperature seen in both curves is due to the assumed (for this analysis only) termination of ventilation at 25 years after emplacement. Based on this comparison, the impact of aging and phase instability on the corrosion of Alloy 22 is not expected to be a problem for the repository design without backfill (CRWMS M&O 2000n, Section 3.1.4.2). The significance of the uncertainties in this data is discussed in Section 3.

Since the completion of the TSPA-SR model, aging and phase stability of Alloy 22 was reevaluated using new data and analyses. These analyses confirm the conclusion of the TSPA-SR model that aging of the Alloy 22 barrier is not a concern (BSC 2001a, Section 7.2.1).

4.2.4.3.3 General and Localized Corrosion

Three separate process models were developed to address general and localized (including microbial) corrosion of the drip shield, waste package outer barrier, and stainless steel structural material. The design described in this report uses Grade 7 titanium for the drip shield, Alloy 22 for the waste package outer barrier, and Stainless Steel Type 316NG as stainless steel structural material. While the stainless steel structural material is not specifically intended to be a corrosion barrier, it may affect the chemistry and the rate of water entering the waste package and retard the rate of radionuclide release from the breached waste package. Given the limited availability of corrosion data for Stainless Steel Type 316NG, data for Stainless Steel Type 316L were used as representative class of materials. This is appropriate since the compositions of the two materials are similar, and aqueous corrosion characteristics are expected to be similar (CRWMS M&O 2000n, Section 1.5.4).

The process model for general and localized corrosion includes submodels for dry oxidation, humid-air and vapor-phase corrosion, and aqueous phase corrosion.

Dry Oxidation—The dry oxidation submodel assumes that dry oxidation could be treated as a single mode of attack, that is, as general corrosion. Corrosion rates were estimated as a function of temperature (CRWMS M&O 2000n, Section 1.5.4.1).

Humid-Air and Vapor-Phase Corrosion—Humid-air and vapor-phase corrosion are treated as a single mode of attack, that is, as general corrosion. Corrosion rates were estimated as a function of temperature (CRWMS M&O 2000n, Section 1.5.4.2).

Aqueous-Phase Corrosion—The process model for aqueous-phase corrosion of the drip shield and waste package accounts for both general and localized corrosion. Two different modeling methods were used to model this process. The first method determined the threshold corrosion potential for localized attack of the material from short term experiments that used relevant test media. Test environments covered the range from the least aggressive environment of J-13 water to the most aggressive environment of concentrated J-13 water, and the temperature range was between 30° and 90°C (86° and 194°F). Using data from published literature and tests at elevated temperature and pressure, the second method determined the threshold temperature for localized attack. The aqueous-phase corrosion model was applied independently to a large number of small regions (patches) on each waste package (CRWMS M&O 2000n, Section 1.5.4.3).

Abstracted models were developed to account for general and localized corrosion of the drip shield and waste package materials. The abstracted models were input to the TSPA analysis using the WAPDEG code. The abstracted models included thresholds for initiation of various modes of corrosion, as well as the corresponding penetration rates. The relative humidity and temperature thresholds for initiation of humid-air and aqueous-phase corrosion were included, as well as the electrochemical potential for initiation of localized corrosion during aqueous-phase corrosion.

In the case of the drip shield and waste package outer barrier, distributions of general corrosion rates were based upon data from the Long-Term Corrosion Test Facility, while project and published data were used as the basis of estimating localized corrosion rates. For the Stainless Steel Type 316NG inner barrier, both general and localized corrosion rates were based upon data presented in Degradation of Stainless Steel Structural Material (CRWMS M&O 2000cw, Section 6). These general and localized corrosion rates included estimates of the uncertainty and variability. Sufficient information was also provided to determine expected failure mode characteristics (e.g., number of failure sites and opening size).

General and Localized Corrosion of Alloy 22—Based on a detailed analysis of the test data, the mean value of the general corrosion rate of Alloy 22 after 24 months of exposure was 10 nm/yr (0.0000004 in./yr). The mean is the average over the number of duplicate samples in the range of test temperatures and the chemical environments summarized earlier. The general corrosion rates determined thus far are so low that the depth of penetration is not deep enough to accurately determine the sensitivity to the temperature and chemical composition of the water. The low general corrosion rates at all test temperatures and chemical environments indicate that the sensitivity is not large. Extrapolation of the mean corrosion rate to 10,000 years implies an average penetration of the Alloy 22 of only 0.1 mm (0.004 in.) of the waste package outer barrier. Even at the highest corrosion rate measured in this data set, the maximum penetration would be less than 1 mm (0.04 in.) over a 10,000-year time period, far less than the package thickness. The TSPA-SR uses the entire range of measured values from the long-term tests using a stochastic approach.

At planned times, some of the samples were withdrawn from the tests for examination. No evidence of localized corrosion was observed on the surfaces of the exposed Alloy 22 specimens for the three exposure times—6, 12, and 24 months (CRWMS M&O 2000n, Section 3.1.5.4). The tests are continuing with the remaining samples. Because of the low general-corrosion rate and the absence of localized corrosion and stress corrosion crack initiation, Alloy 22 is expected to be extremely long-lived as a waste package shell.

General and Localized Corrosion of Titanium—General corrosion rates for the titanium drip shield material were based on Long-Term Corrosion Test Facility weight-loss samples. These rates appeared to be independent of temperatures between 60° and 90°C (140° and 194°F) and the chemistry of the test medium. The median rate was approximately zero, with most measurements for uncreviced samples lying between -200 and +200 nm/yr.

Crevice corrosion rates also appeared to be independent of temperature and test medium. As with the uncreviced samples, the median rate was approximately zero, with most of the corrosion rates between -350 and +350 nm/yr. The largest measured rate was less than +0.35 µm/yr and would not lead to failure of the drip shield during the first 10,000 years of its lifetime. Based upon these data, the life of the drip shield does not appear to be limited by crevice corrosion of titanium at temperatures less than those involved in the test (90°C [194°F]). Testing and model results indicate that the highest probable corrosion rate for titanium is approximately 25 nm/yr and the maximum rate is less than 350 nm/yr.

Microbially Influenced Corrosion of Alloy 22 and Stainless Steel—It has been observed that nickel-based materials such as Alloy 22 are relatively resistant to microbially influenced corrosion (CRWMS M&O 2000cq, Section 6.8). Corrosion rates of Alloy 22 are enhanced by microbially influenced corrosion by only a factor of approximately two (CRWMS M&O 2000cq, Table 25). The augmentation of corrosion rates due to microbially influenced corrosion is accounted for in the model documented in General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq, Section 6.8). Corrosion studies have shown that microbes can enhance corrosion rates of 304 stainless steel by a factor of approximately ten (CRWMS M&O 2000cq, Table 25). It is assumed that microbially influenced corrosion will have the same effect on Stainless Steel Type 316NG.

The principal nutrient-limiting factor to microbial growth in situ at Yucca Mountain has been determined to be low levels of phosphate. Yucca Mountain bacteria grown in the presence of Yucca Mountain tuff are apparently able to dissolve phosphate contained in the tuff to support growth to levels of 106 cells per milliliter of groundwater. When exogenous phosphate is added (10 mM), then levels of bacterial growth increase to 107 to 108 cells per milliliter. It may be noted, however, that the two-fold enhancement of corrosion included in the model was in the presence of sufficient phosphate to sustain higher levels of bacterial growth (in an effort to achieve accelerated Alloy 22 attack).

Other environmental factors that could affect levels of bacterial growth include temperature and radiation. However, these factors are closely coupled to relative humidity. As temperature and radiation decrease in the repository, relative humidity is predicted to increase. There are some types of microorganisms that can survive elevated temperatures (up to 120°C [248°F]) and high radiation doses; if there is no available water, then bacterial activity is completely prevented. Thus, because water availability is the primary limiting factor, and this factor is coupled to other less critical limiting factors, water availability (as expressed by relative humidity) was used as the primary gauge of microbial activity.

It has been assumed that a critical mass of bacteria exists for microbially influenced corrosion. Bacterial densities in Yucca Mountain rock have been determined to be on the order of 104 to 105 cells per gram of rock. In absolute terms, this is almost certainly above the threshold required to cause microbially influenced corrosion. Further, bacterial densities were shown above to increase 1 to 2 orders of magnitude when water was available. More germane concerns are the types of bacteria present, their abundance, and how their relative numbers are affected when water is available for growth. Corrosion rates will be affected (at least on some waste package materials), for example, if organic acid producers outcompete sulfate reducers or inorganic acid producers for available nutrients when water is sufficient to support growth. No data are currently available regarding the composition of the bacterial community over the changing environmental conditions anticipated during repository evolution. As described previously, this issue has been addressed for Alloy 22 in the current model with a corrosion enhancement factor that was determined from the ratio of measured corrosion current densities for inoculated and abiotic samples. The enhancement factor has an upper bound of 2 (for the inoculated condition) and a lower bound of 1 (for the sterile condition). The enhancement factor is applied to the entire surface of the waste package outer barrier when the relative humidity at the surface is greater than a threshold value (i.e., 90 percent relative humidity).

Effect of Gamma Radiolysis on Corrosion Potential—Anodic shifts in the open circuit corrosion potential of stainless steel in irradiated aqueous environments have been experimentally observed (CRWMS M&O 2000cq, Section 6.4.4). Experiments performed at ambient-temperature cyclic polarization of Stainless Steel Type 316L samples in 0.018 M NaCl solution during exposure to 3.5 Mrad/hr gamma radiation showed that the corrosion potential shifted in the anodic direction by approximately 200 mV. This shift in corrosion potential was shown to be due to the formation of hydrogen peroxide. This finding was subsequently confirmed by another cyclic polarization experiment at ambient-temperature with 316 stainless steel in acidic (pH~2) 1.5 M NaCl during exposure to 0.15 Mrad/hr gamma radiation which showed a 100 mV anodic shift in the corrosion potential, with very little effect on the corrosion current. Note that these experiments were performed on stainless steels, not Alloy 22.

To determine the maximum impact that gamma radiolysis could have on the corrosion potential, hydrogen peroxide was added to the test media used for testing Alloy 22. As the concentration of hydrogen peroxide in simulated acidic concentrated water approaches 72 ppm (calculated from number of added drops of hydrogen peroxide), the corrosion potential asymptotically approaches 150 mV, well below the potentials where localized attack would be expected. Similarly, as the concentration of hydrogen peroxide in simulated concentrated water approaches 72 ppm, the corrosion potential asymptotically approaches -25 mV, well below any threshold where localized corrosion would be expected. Therefore, gamma radiolysis is not expected to result in the localized corrosion of Alloy 22 since the maximum shift in corrosion potential (induced by hydrogen peroxide additions) is less than that required for initiation of localized corrosion.

Pitting and Crevice Corrosion of Alloy 22 and Titanium—Short-term tests evaluating the susceptibility of Alloy 22 and titanium to pitting and crevice corrosion were conducted in accordance with ASTM G 61-86, Standard Test Method for Conducting Cyclic Potentiodynamic Polarization Measurements for Localized Corrosion Susceptibility of Iron-, Nickel-, or Cobalt-Based Alloys, to determine relative susceptibility to these types of localized corrosion. The results of these tests do not indicate a susceptibility to localized corrosion in plausible repository environments.

Since the completion of Total System Performance Assessment for the Site Recommendation (CRWMS M&O 2000a), the DOE has performed further research and analysis utilizing available project data to develop a temperature-dependent general corrosion rate and calculated the change in performance of the waste package over a range of operating modes. The inclusion of a temperature-dependent corrosion rate was included as a supplemental sensitivity analysis (BSC 2001b, Section 3.2.5.3). In the revised supplemental TSPA model discussed in Total System Performance Assessment—Analyses for Disposal of Commercial and DOE Waste Inventories at Yucca Mountain—Input to Final Environmental Impact Statement and Site Suitability Evaluation (Williams 2001a, Section 5.2.4.1), the general corrosion model was independent of temperature. Results of both supplemental analyses are consistent with the conclusions drawn from the results of the TSPA-SR model analyses and also provide additional insights into the behavior of the disposal system.

4.2.4.3.4 Stress Corrosion Cracking Models

Stress corrosion cracking is a mechanism of cracking in some materials caused by tensile stresses in an aggressive environment (CRWMS M&O 2000cu). This mechanism requires a combination of three elements: material susceptibility, tensile stresses, and an appropriate environment. Unique aspects of the stress corrosion cracking mechanism include brittle-type cracking in ductile materials, crack initiation and propagation under constant load without a need for cyclic load, and occurrence of cracking at lower stresses than would be expected without the presence of an aggressive environment. This process model accounted for the possibility of stress corrosion cracking of the drip shield, the waste package outer barrier, and the stainless steel structural material. The model for stress corrosion cracking evaluated two alternative methods. The first method was based on the stress-intensity threshold criterion; the second method was based on a mechanistic film-rupture model yielding a finite rate of stress corrosion crack propagation. The modeled rate of stress corrosion crack propagation was dependent upon both the local environment and the stress intensity factor at the crack tip. The stresses for initiation and propagation of stress corrosion cracking in the models were due to unannealed closure welds, deformation caused by rockfall, and the weight of the waste package.

Stress Corrosion Cracking of the Stainless Steel Waste Package Inner Shell—The stainless steel structural material may or may not be susceptible to stress corrosion cracking. However, because the corrosion rate of stainless steel is expected to be much higher than that of Alloy 22, the TSPA is using a simplified model in which the stainless steel is assumed to fail immediately after the Alloy 22 outer barrier. Potential performance contribution of the stainless steel structural material will be evaluated as additional data and analyses become available.

Stress Corrosion Cracking of the Titanium Drip Shield—Stress corrosion cracking of the titanium drip shield is not expected in a design that uses backfill, because drip shield stresses that are relevant to stress corrosion cracking are insignificant in this case. The major source of drip shield stress is loading due to earthquakes.

For designs without backfill, rockfall directly on the titanium drip shields could result in localized areas of high residual stresses which could lead to stress corrosion cracking and through-wall penetration of the drip shields. While this is possible, it is expected that, because of their size, these penetrations will not prevent the drip shields from performing their intended function of diverting seepage water away from the waste package surface.

Stress corrosion cracks in passive alloys, such as Titanium Grade 7 and Alloy 22, tend to be very tight (small crack opening displacement) by nature (CRWMS M&O 2000cu). Stress analyses that consider rockfall have estimated the local residual titanium stresses following a large (i.e., greater than 4 metric tons) rock impact. The subsequent stress corrosion crack is expected to be approximately 100 µm (0.1 mm, or 0.004 in.) wide. The crack faces are expected to corrode slowly and eventually fill the crack space with corrosion products. While the crack faces are corroding slowly, there may be a small amount of water transported by surface diffusion into the crack and through the drip shield. However, the small temperature gradient present across the drip shield wall will result in evaporation of the slowly flowing water, and a resultant scale deposit will form over and within the crack on the upper drip shield surface. This formation of scale deposits is well documented in seawater environments and in heat exchangers. Such deposits form rapidly under flowing conditions and must be regularly removed to avoid loss of heat exchanger efficiency. In the case of J-13 water concentrated by evaporation, calcium carbonate precipitation is the first stage of the concentration process. Consequently, evaporation of J-13 water slowly flowing through an approximately 100-m crack opening would lead to rapid scale deposition. The rate of plugging would be inversely proportional to the volume flow rate through the crack. For a 100-µm wide crack through a 15-mm (15,000-µm) thick drip shield, scale deposition would be an efficient process involving the need for only a small volume flow through the crack. In addition, because of the expected high density of the deposits and lack of a pressure gradient to drive water through the crack, the probability of solution flow through the plugged crack would approach zero. It is therefore concluded that the function of the drip shield will not be compromised even in the event of stress corrosion cracking (CRWMS M&O 2000cu).

Stress Corrosion Cracking of the Alloy 22 Waste Package Outer Shell—For Alloy 22, the stress corrosion cracking film rupture model (the second model) assumes crack growth could begin at any surface defect that could generate a stress intensity, regardless of size and surface tensile stress level. However, examination of the relevant literature indicated that there is a threshold stress below which stress corrosion cracking would not initiate on a "smooth" surface. In the case of the waste package closure weld, the range of threshold stress was conservatively estimated at 10 to 40 percent of the material yield stress (CRWMS M&O 2000cu, Section 6.5.2). This threshold stress range was based on literature reporting stress corrosion cracking-initiation test results for stainless steels and nickel alloys (with known susceptibility to stress corrosion cracking) with comparable surfaces exposed to very aggressive environments.

Mitigation of the Alloy 22 Closure Weld Residual Stress—Examination of the weld residual stress profiles for the initial waste package outer barrier closure design revealed that tensile stresses exceeding 20 percent of Alloy 22 yield stress existed in the vicinity of the closure weld surface in both the radial and circumferential directions. This indicated stress corrosion cracking initiation in unacceptably short times could not be precluded for that design. Further, the stress intensity plot indicated that, at least for radially oriented stress corrosion cracks, through-wall crack propagation was possible once a crack initiated.

To reduce weld residual tensile stresses below the stress corrosion cracking-initiation threshold, improved waste package closure designs were evaluated. These included low residual-stress welding techniques and postweld residual tensile-stress-reduction techniques such as induction annealing (heating of the weld) and laser peening (subjecting the weld to laser-generated compressive stresses). Based on these evaluations and a review of the relevant literature (CRWMS M&O 2000cu, Section 6), the Alloy 22 closure design was improved to add a second flat closure lid on the top end of the waste package. This lid will be placed between the extended Alloy 22 closure lid and the stainless steel inner lid (see Figure 3-2 in Section 3). In the improved design, the inner lid closure weld is relieved of its tensile stress using laser peening. Then the outer lid closure weld is completed and its residual tensile stresses are relieved using postweld induction annealing.

To optimize the new Alloy 22 closure design, experimental measurements were made to quantify the expected stress reduction benefit resulting from laser peening. In addition, extensive ANSYS finite element model calculations were performed to determine the expected stress and stress intensity reduction benefits resulting from the application of both processes. The results of these calculations show that both processes reduced the surface residual tensile stresses below 20 percent of the yield stress (the threshold value for initiation of stress corrosion cracking). These stress reductions persist to depths of about 2 to 3 mm (0.08 to 0.12 in.) for laser peening and at least 6.5 mm (0.25 in.) for postweld induction annealing.

As the waste package outer surface corrodes away, the beneficial low-tensile-stress surface layers resulting from either process are removed. Use of the dual Alloy 22 closure lids in the design improved the longevity of the overall closure, since the flat closure lid does not begin to corrode until the extended closure lid is breached. Figure 4-89 shows the schematic of the dual closure lid design, and Figure 4-90 shows conceptual details of remote welding, annealing, and laser peening for the closure weld area of the waste package.

The closure lid design, as mentioned earlier, helps extend the life of the waste package. The closure weld area of the extended lid is subjected to localized induction annealing to mitigate the weld residual stresses. Stress corrosion cracking of this lid is delayed until the stress-mitigated layer is removed by corrosion. The closure weld area of the flat closure lid is subjected to residual stress mitigation by laser peening. This lid is exposed to the repository environment only after the extended lid is breached, and the stress corrosion cracking of this lid is delayed until after the mitigation layer corrodes away.

Figure 4-90 shows how welding, inspection, laser peening, and induction annealing can be performed at a single work station in the transfer cell facility. This conceptual arrangement will continue to evolve during the next few years.

Additional analyses and quantification of uncertainties performed since the completion of the TSPA-SR model provide additional confidence that the stress corrosion cracking model used in the TSPA-SR model is conservative and that the likelihood of stress corrosion cracking of the waste package in the repository environment will be less than predicted by the TSPA-SR model (BSC 2001a, Sections 7.2.5 and 7.3.3).

4.2.4.3.5 Hydrogen-Induced Cracking of the Titanium Drip Shield

Another potential failure mechanism for titanium and its alloys under waste disposal conditions is hydrogen absorption leading to hydrogen-induced cracking. Hydrogen-induced cracking is also called "hydrogen embrittlement," which is a process resulting in a decrease of fracture toughness or ductility of a metal due to the presence of atomic hydrogen. The usual failure mode for a ductile material is the ductile tearing observed during slow crack growth. In this case, the material will fail as the stress intensity factor reaches a threshold value.

The decrease of fracture toughness can also cause fast crack growth (brittle fracture) of a normally ductile material under a sustained load. During fast crack growth, the same material will fail as the stress intensity factor reaches another threshold value, which is less than threshold value for slow crack growth. The process model developed for this degradation mode established the conditions under which the drip shield would experience hydrogen uptake, potentially leading to hydrogen embrittlement and hydrogen-induced cracking.

Generally, the passive oxide film on titanium acts as an excellent barrier to the transport of hydrogen, and hydrogen absorption under natural corrosion conditions would not be expected and is generally not observed over normal operating periods (up to tens of years). In a repository situation, even very slow hydrogen absorption may be significant over thousands of years, leading to a significant accumulation of hydrogen and the danger of hydrogen-induced cracking. The model addressed these considerations, which are summarized below:

  1. The passive oxide is assumed to be permeable to atomic hydrogen.

  2. Atomic hydrogen is generated at the surface of the titanium alloy. This is described by a hydrogen generation rate, which is taken to be proportional to the general passive corrosion rate.

  3. A fraction of the hydrogen is absorbed into the oxide and assumed to directly enter the alloy. The remainder combines to yield hydrogen gas, which is lost to the surroundings. The rate of absorption is taken to be directly proportional to the hydrogen generation rate multiplied by an absorption efficiency coefficient.

  4. Once in the alloy, the hydrogen is distributed uniformly throughout the entire thickness of the material. That is, transport processes within the alloy are rapid compared to the rate of absorption.

  5. The hydrogen content of the alloy is allowed to increase until a critical level is reached. The material then is assumed to fail immediately. The model allows for the calculation of hydrogen content and for comparison with the critical concentration.

The hydrogen-induced cracking model was presented in Hydrogen Induced Cracking of Drip Shield (CRWMS M&O 2000cv, Section 6). In that report, extensive evidence supported a qualitative assessment of titanium alloys as an excellent choice of material for the drip shield with regard to degradation caused by hydrogen-induced cracking, where the hydrogen source is from general corrosion of the titanium. Quantitative evaluation based on this model indicated that the drip shield material was able to sustain the effects of hydrogen-induced cracking. Using available general corrosion test data, the model calculated a hydrogen concentration below 120 µg per gram, which is less than the critical hydrogen concentration of 400 µg per gram for the titanium grade that was tested.

The source of the hydrogen for embrittlement can also be through galvanic coupling of the titanium drip shield with less corrosion-resistant materials, such as carbon steel. At the bottom of the drip shield, this galvanic coupling is precluded by the use of a small Alloy 22 "foot" separating the drip shield and the carbon steel invert structure (CRWMS M&O 2000cv).

In a design without backfill, steel components such as rock bolts, wire mesh, and steel liners used in the drift may fall on the drip shield and undergo active corrosion. This could potentially lead to generation of hydrogen on the surface of the drip shield, leading in turn to hydrogen pickup levels higher than the critical levels needed for cracking. Preliminary evaluation shows that these embrittled titanium regions are likely to be self-limiting "hotspots" from the point of view of hydrogen absorption and embrittlement of the drip shield. The buildup of carbon steel corrosion products at the contact site would be expected to eventually break the contact. The drip shield is expected to be unaffected by contact with iron corrosion products, since the galvanic connection would be poor (CRWMS M&O 2000cv).

4.2.4.3.6 Limitations and Uncertainties

Uncertainties in each of the process models were identified in the discussion of the models in the previous sections and in the individual analysis model reports. The approach generally used in dealing with these uncertainties is to be conservative and bound the uncertainties. In several cases combinations of approaches were used. For example, in the case of Alloy 22 thermal aging, accelerated testing (i.e., higher-temperature aging) was used to predict long-term behavior based on short-term data, in accordance with ASTM C 1174-97. The data were then used with conservative bounding estimates. In the case of general corrosion, aggressive test media were used to represent potential concentrating effects in the repository. Features, events, and processes were evaluated to screen out degradation processes that have very low probability of occurrence (less than 10-4/yr) or that have very low consequences. Examples of screened-out processes include radiolysis-enhanced corrosion and inside-out corrosion of waste package barrier materials.

As noted in Section 4.1.1.2, the DOE has performed several activities to improve the treatment of uncertainty in current models. Additionally, as noted in Section 4.1.4, the DOE has evaluated the effect that operating the repository at lower temperatures would have on repository performance models. Uncertainty about the waste package corrosion rate may be reduced by avoiding the conservatively defined window of corrosion susceptibility for Alloy 22, which can be accomplished by keeping waste package temperatures at or below 85°C (185°F) or maintaining the relative humidity in emplacement drifts below 50 percent. The results of the analyses of lower-temperature operating modes are described in FY01 Supplemental Science and Performance Analyses (BSC 2001a; BSC 2001b).

A review of the uncertainties in the various models is presented below.

Thermal Aging—A graphical approach to bounding the uncertainty in the aging model is used. In Figure 4-91, the line representing a "best fit" to the data for "complete grain boundary coverage" predicts that more than 10,000 years at 300°C (572°F) will be required to completely cover the grain boundaries of Alloy 22 with inter-metallic precipitates. However, the line with the "minimum slope possible within the error bars" shows that complete grain boundary coverage might occur in as few as 100 years (very unlikely bounding case). The "best fit" line is the most likely scenario. In the case of bulk precipitation, none is predicted with the line representing the minimum possible slope. Thus, it can be concluded with reasonable certainty that no bulk precipitation will occur before 10,000 years at 300°C (572°F).

From the corrosion tests conducted, it appears that a fully aged sample of Alloy 22 could change the observed corrosion potential. For example, corrosion potential was shifted in a less noble (negative) direction by a small value (less than 100 mV) in simulated acidic concentrated water and simulated concentrated water at 90°C (194°F). The shift was not considered significant, and it was concluded that full aging of Alloy 22 (complete coverage of the grain boundaries) does not significantly alter passive film stability and result in significantly enhanced corrosion.

Thermal aging of Titanium Grade 7 at 300°C (572°F) is expected to have little impact on the corrosion resistance of this material. Since no credit is claimed for the corrosion resistance of Stainless Steel Type 316NG, all TSPA calculations are insensitive to the uncertainty associated with the corrosion of Stainless Steel Type 316NG.

Dry Oxidation—In the case of Alloy 22 and Stainless Steel Type 316NG, the rates of dry oxidation are negligible even over hundreds of years and the expected repository temperature. Therefore, uncertainty in the dry oxidation rate is not expected to have any significant impact on the performance of these materials. The current model is based upon published data and does not include estimates of uncertainty.

Humid-Air and Aqueous-Phase Corrosion—Uncertainty in the threshold relative humidity for these corrosion modes is primarily due to the composition of the salt film. The waste package and drip shield would always experience some combination of humid-air and aqueous-phase corrosion. The uncertainty in this parameter is discussed in more detail in Degradation of Stainless Steel Structural Material (CRWMS M&O 2000cw, Section 6), General Corrosion and Localized Corrosion of the Drip Shield (CRWMS M&O 2000cr, Section 6), and General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq, Section 6).

The distribution of general corrosion rates for either humid air corrosion or aqueous phase corrosion is represented by the curves given in Section 3.1.5.4 of the Waste Package Degradation Process Model Report (CRWMS M&O 2000n, Figures 3-14 through 3-26). Distributions for Stainless Steel Type 316NG are represented by distributions formed from published data. The distribution of rates for Alloy 22 and the distribution for Titanium Grade 7 rates are based upon data from the Long-Term Corrosion Test Facility. The dispersion in these curves is assumed to be entirely due to uncertainty. A detailed analysis of the error in general corrosion rate was given in each supporting report (CRWMS M&O 2000cw, Section 6; CRWMS M&O 2000cr, Section 6; (CRWMS M&O 2000cq, Section 6).

Determinations of corrosion and threshold potential are based upon three replicate cyclic polarization measurements at each combination of environment and temperature. The results are tabulated in General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq, Section 6). The uncertainty in the corrosion potential due to gamma radiolysis (i.e., a maximum positive shift in error of about 250 mV) is also addressed in General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq, Section 6.4.4) and Waste Package Degradation Process Model Report (CRWMS M&O 2000n, Section 3.1.6). Estimates of uncertainty in the selection of corrosion and threshold potential have been made and are embedded in WAPDEG analyses.

The rates of localized corrosion have been bounded with the range of values found in the published literature.

Stress Corrosion Cracking—Two alternative stress corrosion cracking models have been considered, one based upon a threshold stress intensity factor (Method A) and another based upon a threshold stress for a smooth surface (Method B) (CRWMS M&O 2000cu). In the second approach, cracks are assumed to propagate by the slip-dissolution mechanism after initiation. Method B is used as the basis of the performance assessment. The slip-dissolution model, Method B, predicts that crack propagation is a function of the local stress intensity at the crack tip. Thus, the uncertainty in this driving force must be estimated.

Local stress intensity is calculated from the local stress and the crack penetration. The uncertainties in the stress distribution (stress versus depth) are based upon analyses of measured residual stresses in welds, before and after mitigation, as well as finite element modeling with the ANSYS code. These uncertainties are abstracted for WAPDEG. Aside from the stress intensity, the parameters in the slip-dissolution model for stress corrosion cracking propagation are based upon measurements for stainless steel from the boiling water reactor industry. Since stainless steels are much more prone to stress corrosion cracking than Alloy 22, these parameter estimates are conservative for Alloy 22.

The threshold stress for initiation of stress corrosion cracking on a smooth surface is conservatively estimated to be approximately 10 to 40 percent of the yield stress, based upon the determination of such thresholds for related but more susceptible alloy systems exposed to very aggressive environments, such as boiling magnesium chloride.

Since completion of the TSPA-SR model, the DOE has continued to explore uncertainties associated with the model. Volume 1, Section 7 of FY01 Supplemental Science and Performance Analyses (BSC 2001a) describes in detail the additional work performed to further defend, model, and understand uncertainties associated with the waste package degradation models. It further discusses uncertainties related to mechanisms for early failures, aging and phase stability of the waste package outer barrier, general and localized corrosion, and stress corrosion cracking. In order to assess the performance consequences of early waste package failures, the supplemental TSPA model described in Volume 2 of FY01 Supplemental Science and Performance Analyses (BSC 2001b) included a model that assumed that improper heat treatment of welds could lead to cracks forming in a small fraction of waste packages. The formation of cracks in a few waste packages (up to three) resulted in the calculation of small doses (approximately 2 math symbol, multiply 10-4 mrem/yr) in the supplemental TSPA model (compared to zero dose in the TSPA-SR model). The inclusion of a thermally-dependent corrosion rate in the waste package performance model resulted in significantly improved long-term performance of the waste packages. In summary, the results of the additional analysis provide added confidence that the TSPA-SR model is conservative and that the behavior of the waste package in the repository environment is likely to be better than predicted by TSPA-SR model.

4.2.4.3.7 Alternative Approaches or Models

Alternative models have been considered for oxidation, localized corrosion thresholds, stress corrosion thresholds, stress corrosion cracking, stress mitigation, and hydrogen-induced cracking. These alternatives are summarized below.

As with limitations and uncertainties, alternative conceptual models for corrosion, early waste package failure, and stress corrosion cracking have been updated in supplemental analyses as summarized in FY01 Supplemental Science and Performance Analyses (BSC 2001a, Section 7; BSC 2001b, Sections 3.2.5 and 4.2.5). These analyses generally tend to confirm the results of the TSPA-SR. These analyses are also summarized in Section 4.4.5.5 of this report.

Dry Oxidation—This process can be represented by two different methods: the parabolic growth law and the logarithmic growth law. In the parabolic growth law, the oxide film is assumed to grow continuously with the rate of film growth controlled by the diffusion of oxygen through the film. The thickness of the oxide is proportional to the square root of the exposure time. In the case of the logarithmic growth law, the oxide film asymptotically approaches a small maximum thickness. In the case of dry oxidation, parabolic growth law is used for Alloy 22 and Stainless Steel Type 316NG, while logarithmic growth law is used for Titanium Grade 7. These model selections were based upon the published literature for these types of corrosion-resistant materials (CRWMS M&O 2000n, Section 3.1.5.1).

Localized Corrosion Threshold—Localized corrosion process can be expressed in terms of the threshold electrochemical potential or threshold temperature for the initiation of the localized corrosion. For the highly corrosion resistant materials under consideration, the threshold potential model is used because it is more solidly rooted in the theoretical concepts underlying passive film stability. Furthermore, it is expected that a good correlation of threshold and corrosion potential can be used to deduce a threshold temperature. The threshold temperature would be the temperature at which the corrosion and threshold potentials are equivalent.

Stress Corrosion Cracking—The stress corrosion cracking model includes two separate models, one based on the threshold stress intensity factor at the tip of a preexisting flaw and the second based on the threshold stress for propagation by slip-dissolution mechanism. The second method is used in the abstraction for TSPA because it is the more conservative model and because it has been used for predicting the performance of boiling water reactors. The method is considered to be conservative because it is based upon data for stainless steel, a material more prone to stress corrosion cracking than Alloy 22 (CRWMS M&O 2000cu). Predictions based upon a correlation for stainless steel would yield a conservative prediction for Alloy 22.

Weld Stress Mitigation—Two methods will be used for mitigation of the weld residual stresses: localized induction annealing and laser peening. Induction annealing will be used for the outer lid, and laser peening will be used for the inner lid on the Alloy 22 outer barrier of the waste package. This selection is based upon the ability of the induction annealing process to place compressive stress deeper into the weld. Laser peening is more compatible with the design of the inner lid weld, due to the occluded nature of such a weld.

Hydrogen-Induced Cracking of Titanium—Two methods are available to address hydrogen induced cracking: threshold electrochemical potential and threshold hydrogen concentration. The first method is applicable primarily for conditions that would lead to galvanic coupling of titanium. As this will be avoided by design, this method has not been used in the model. Therefore, hydrogen-induced cracking evaluation is based upon a threshold hydrogen concentration. Also, since such concentrations are possible to be measured with secondary iron mass spectrometry, this method is preferred.

The threshold concentration model is a simple and conservative model. The basic premise of the model is that failure will occur once the hydrogen content exceeds a certain limit or critical value. This model is conservative because it assumes that, once the environmental and material conditions can support that particular corrosion process, failure will be effectively instantaneous. Quantitative evaluation based on the hydrogen-induced cracking model described in Hydrogen Induced Cracking of Drip Shield (CRWMS M&O 2000cv, Section 6) indicates that the drip shield material (Titanium Grade 7) is able to sustain the effects of hydrogen-induced cracking. Available test data show that the hydrogen concentration is below 180 µg per gram, which is less than the critical hydrogen concentration of 400 µg per gram for Titanium Grade 7.

With the removal of backfill from the repository design, the potential for that galvanic interaction between carbon steel ground support and the drip shield is increased. The impact of this on the potential hydrogen pickup has been evaluated using the threshold concentration model. Preliminary results show that hydrogen-induced cracking is not a problem (CRWMS M&O 2000cv).

4.2.4.3.8 Model Calibration and Validation

According to ASTM C 1174-97, Standard Practice for Prediction of the Long-Term Behavior of Materials, Including Waste Forms, Used in Engineered Barrier Systems (EBS) for Geological Disposal of High-Level Radioactive Waste, model validation is the process through which independent measurements are used to ensure that a model accurately predicts an alteration behavior of waste package materials under a given set of environmental conditions (e.g., in the repository environment over 10,000 years). Obviously, models cannot be tested for 10,000 years. Therefore, validation relies on accelerated testing. According to the same standard, an accelerated test is a test that results in an increase in the rate of an alteration mode, when compared with the rates for service condition. Changes in alteration mechanism, if any, must be accounted for in the use of the accelerated test data.

The thermal aging model assumes Arrhenius-type kinetics (CRWMS M&O 2000ct, Sections 6.2 and 6.5). Precipitation and long-range ordering can be accelerated by increasing the temperature above those levels expected in the repository. If the model can accurately predict the kinetics of these phenomena at combinations of time and elevated temperature, it will be considered valid for making predictions at lower temperature and longer time.

Since all available data has been used to establish these correlations, the correlations are considered valid for their intended use. Additional data that are being collected will help to reduce uncertainties and improve the level of confidence in the model.

The effects of precipitation and long-range ordering on corrosion are determined with electrochemical techniques. Through application of electrochemical potentials more anodic than the open circuit corrosion potential, corrosion phenomena can be accelerated (CRWMS M&O 2000cq, Section 6.7). Variations in corrosion and threshold potential can be correlated with the extent of thermal aging. Similarly, variations in rates of dissolution through the stable passive film can also be correlated with the extent of thermal aging. These rates of dissolution are accelerated by application of a potential between the corrosion and threshold potentials and are proportional to the passive current density (CRWMS M&O 2000cq, Section 6.7.1). The corrosion rate enhancement factor is determined by calculating the ratio of measured passive current densities for aged and unaged samples (CRWMS M&O 2000cq, Section 6.7.3). Since all available electrochemical data has been used to establish the corrosion model for thermally aged samples, this model is considered valid for its intended use.

The models for dry oxidation of Alloy 22 (CRWMS M&O 2000cq, Section 6.1), Titanium Grade 7 (CRWMS M&O 2000cr, Section 6.1), and Stainless Steel Type 316NG (CRWMS M&O 2000cw, Section 6.1) are based upon published data found in the scientific literature. More specifically, the model for dry oxidation of Alloy 22 is based upon the parabolic growth of the oxide film at elevated temperature (CRWMS M&O 2000cq, Section 6.1). However, in the absence of any such low-temperature data, the parabolic rate constant for high temperature is applied at low temperature. Given the extremely small magnitudes of these rates, dry oxidation is expected to have no significant impact on waste package performance (CRWMS M&O 2000cq, Section 6.1).

The threshold relative humidity for humid-air corrosion is based on the deliquescence point of sodium nitrate (CRWMS M&O 2000ck, Section 7; (CRWMS M&O 2000cq, Section 4.1.2). The threshold for salt films deposited in the repository may be slightly different. However, salt deposits produced by evaporating simulated J-13 water to dryness support this basis.

Rates of humid air corrosion are expected to follow general distributions based upon weight-loss data from the Long-Term Corrosion Test Facility (CRWMS M&O 2000cq, Section 6.5.2; CRWMS M&O 2000cr, Section 6.5.2). The distributions are for Alloy 22 data for 6, 12, and 24 months of exposure to a variety of test media. Corroborative measurements made with the atomic force microscope and other surface analytical techniques have also been used as further means of model validation (CRWMS M&O 2000cq, Section 6.5.5). The test program will continue, ultimately providing data for 60 months of exposure. Future data will be considered independent and corroborative, and will be used to reduce uncertainties and conservatism in the model.

The threshold relative humidity for aqueous phase corrosion is the same as that used for humid air corrosion (CRWMS M&O 2000ck, Section 6.3; (CRWMS M&O 2000cq, Section 6.3). The same approach has been used for validation. Rates of general corrosion in the aqueous phase also obey the general distributions based upon weight-loss data from the Long-Term Corrosion Test Facility. The same approach described for validation of the rate model for humid air corrosion has been employed for validation of the rate model for general corrosion in the aqueous phase.

Comparisons of corrosion and threshold potentials are used to determine whether rates for general or localized corrosion are applicable. The initial correlations given in the Waste Package Degradation Process Model Report (CRWMS M&O 2000n, Sections 3.1.5 and 3.1.6) are based upon standard cyclic polarization measurements in simulated dilute water, simulated concentrated water, simulated acidic concentrated water, and simulated saturated water, covering a broad range of temperature.

The stress corrosion model is primarily based on published data. Limited data have been obtained under repository-relevant conditions. The data obtained under the Yucca Mountain project include precracked specimens tested under very aggressive environments (CRWMS M&O 2000cu, Sections 4.1.3 and 6.3.2). Thus, the model uses a conservative approach. Future data will serve to reduce the level of conservatism and improve the confidence in the model.

4.2.4.4 Total System Performance Assessment Abstraction

An integrated model was developed from the process and abstraction models for the various degradation modes. The integrated model included in the WAPDEG performance assessment code used repository environmental conditions as a function of time from other process models to estimate the performance of the waste package and drip shield. The WAPDEG model uses a stochastic approach to sample model parameters over the ranges including uncertainties. The following sections discuss model abstractions for the TSPA-SR. Details of the abstraction of individual process models and the development of the integrated model parameters are provided in Section 3.2 of the Waste Package Degradation Process Model Report (CRWMS M&O 2000n).

Abstraction of General Corrosion Models—The model abstractions are to develop two cumulative distribution functions to represent the general corrosion rate distribution for Alloy 22 outer barrier and the titanium drip shield. For each material, the weight loss and crevice sample penetration rate data were combined to yield one general corrosion rate data set. For Alloy 22, the general corrosion rate data with 6-month, 1-year, and 2-year exposure were considered. Since the variance in the corrosion rate data is reduced with the exposure time, and the median rate also decreases with the exposure time, it was concluded that the 2-year data are sufficiently conservative to represent the long-term general corrosion rate. Therefore, only the 2-year data were used in the model abstraction. For the drip shield, only the 12-month data were used in the model abstraction. The cleaning method employed with the 6-month titanium samples caused significant metal loss, thereby yielding inaccurate corrosion rates. Assumptions shown below were employed in the model abstraction.

As discussed in Section 6.5.5 of General Corrosion and Localized Corrosion of Waste Package Outer Barrier (CRWMS M&O 2000cq), the formation of silica scale deposit on the surface of the Alloy 22 sample coupons could bias the estimated general corrosion rate. The potential measurement bias for the weight loss sample coupons was estimated to be 0.063 µm/yr, and the Alloy 22 general corrosion rate was corrected for the maximum bias by adding a constant value of 0.063 µm/yr to the estimated value of the general corrosion rate. The same data treatment was used for the Titanium Grade 7 drip shield general corrosion rate data. As a result, the corrosion rate correction increased the median rate (50th percentile value) by about 50 percent.

Abstraction of Localized Corrosion Models—This section discusses the approaches and assumptions used in the abstraction of localized corrosion models for the waste package outer barrier and drip shield, as well as the abstraction results.

The model abstractions are to develop two localized corrosion initiation criteria: one representing the localized corrosion initiation criterion for the waste package outer barrier (Alloy 22) and the other for the localized corrosion initiation criterion for the drip shield (Titanium Grade 7). Cyclic polarization measurements were made in several synthetic concentrated J-13 waters. For each curve obtained, the critical potential for localized corrosion initiation and the corrosion potential were determined. The potential difference between these two was then fit to a function of relevant exposure parameters. According to the model, localized corrosion should initiate if the corrosion potential exceeds the critical potential. The abstraction results showed that localized corrosion of Alloy 22 and titanium do not initiate under repository conditions, based on extrapolation of the repository-relevant experimental data used in the analysis.

Abstraction of Stress Corrosion Cracking Model—In the current waste package degradation analysis, two alternative stress corrosion cracking models, the slip dissolution (or film rupture) model and the threshold stress intensity factor model, are considered (CRWMS M&O 2000cu, Section 6). In the threshold stress intensity factor model, this factor is used to determine when stress corrosion cracking will occur. Provided that an initial flaw and corrosive environment are present, a stress corrosion cracking failure will occur when the applied stress intensity factor is greater than or equal to the threshold stress intensity factor. The slip dissolution model also assumes that incipient cracks or defects grow continuously when the oxidation reaction that occurs at the crack tip ruptures the protective film via an applied strain in the underlying matrix. The rate at which the crack grows is a function of the crack tip strain rate and environmental and material chemistries.

In the waste package degradation analysis (WAPDEG), the slip dissolution model is employed to calculate the growth rate of cracks initiated by stress corrosion cracking. The waste package degradation analysis employs a stochastic approach to model the initiation and propagation of cracks from stress corrosion. The major efforts in the abstraction are to develop an approach to represent the uncertainty and variability associated with the stress corrosion cracking initiation and crack propagation processes. The associated parameters in the model include two model parameters (A and n), stress intensity factor, threshold stress, and incipient crack density and size. The analysis also includes preexisting manufacturing defects in the closure lid welds. The manufacturing defect sizes are sampled for the closure lid weld patches, and the sampled flaws are included in the analysis. Because manufacturing defects are much larger than the incipient cracks, the closure lid weld patches with manufacturing defects are likely to fail initially because of stress corrosion cracking.

Abstraction of Stress and Stress Intensity Factor Profile—The Alloy 22 barrier has dual closure lids, referred to as outer and inner lids. The process model analyses calculated the stress and stress intensity factor profiles along the circumference of the welds for each of the closure lids (CRWMS M&O 2000cu, Section 6), and the results were analyzed to develop abstracted models to represent the uncertainty and variability of the profiles in the closure lid welds.

Assumptions employed in the abstraction are: Three cases have been evaluated: "optimum," "realistic," and "most conservative." The uncertainty range in the hoop stress (and corresponding stress intensity factor profiles based on the hoop stress) is bounded between ±5, ±10, and ±30 percent, respectively, of the yield strength and centered around the mean hoop stress profile. The technical basis for the three uncertainty ranges is discussed in Section 6.2.2.5 of Stress Corrosion Cracking of the Drip Shield, the Waste Package Outer Barrier and the Stainless Steel Structural Material (CRWMS M&O 2000cu).

Abstraction for Manufacturing Defects in Waste Package Closure Welds—Abstracted models were developed for the probability and size of manufacturing defects in the waste package closure lid welds. The flaw density is used as the parameter to represent the frequency of occurrence of flaws in a given length of closure weld. The flaw sizes are given as a probability density function on each closure lid weld.

Further details of the assumptions used in the abstraction analyses are discussed in Section 3.2 of Waste Package Degradation Process Model Report (CRWMS M&O 2000n). Major assumptions employed in the abstraction are:

The number of flaws that appear on a waste package patch is sampled stochastically as a Poisson random variable. For each flaw that occurs, a flaw size is randomly assigned to it by sampling from the calculated flaw size cumulative distribution function. This flaw's location and size are then used in the stress corrosion cracking analysis.

Because embedded defects can become surface breaking defects as general corrosion proceeds, the consideration only of preexisting surface breaking defects may not be appropriate. As an alternative conservative abstraction, both surface breaking defects and embedded defects within the outer quarter region of the weld surface are considered. Three observations for the sum of the fraction of surface breaking flaws and the fraction of flaws embedded within the outer quarter region of the surface are used in the alternative conservative abstraction. This, however, may be overly conservative because most embedded defects would be oriented in the radial direction that would not lead to stress corrosion cracking. The stress corrosion cracking analysis considers that hoop stress is the dominant stress in the closure lid welds and drives radial crack propagation.

Integrated Analyses for Drip Shield and Waste Package Degradation—This section reports WAPDEG analysis results for drip shield and waste package degradation. The section includes the discussion of results for the three cases (optimum, realistic, and most conservative) that are likely to represent the potential range of major corrosion model parameter values that could affect the long-term performance of the waste package and the drip shield in the repository. The optimum case represents the parameters achievable through stringent control of such processes as stress mitigation, material variability, welding, and other fabrication steps. The realistic case represents what is achievable through appropriate levels of process controls. The most conservative case represents the combination of the worst case parameters that might result from inadequate control of the processes. In the WAPDEG analyses for the three cases, the potential performance credit of the stainless steel structural material is not considered.

The WAPDEG model, an integrated model used for waste package and drip shield degradation analysis, is based on a stochastic simulation approach. It describes waste package degradation, which occurs as a function of time and repository location for specific design and thermal hydrologic modeling assumptions. The corrosion modes that were included in the analyses are:

The following corrosion parameters were abstracted and included in the analyses:

For the stress corrosion cracking analysis of the waste package closure lid welds in the WAPDEG analysis, the slip dissolution model has been adopted over the threshold stress intensity factor model. For the analysis with the slip dissolution model, the following should be met before initiating a stress corrosion cracking crack propagation in a patch: (1) the stress intensity factor (KI) should be positive, and (2) the stress state must be greater than or equal to the threshold stress. In the WAPDEG analysis, for those patches with a compressive stress zone (or layer) in the outer surface, the compressive stress zone is removed by general corrosion, and this delays the application of the slip dissolution model for the crack propagation rate. The delay time depends on the compressive zone thickness and the general corrosion rate sampled for the patch.

In addition, preexisting manufacturing defects in a patch are all assumed to be surface breaking for the optimum case. For the most conservative and the realistic cases, all flaws within the outer quarter thickness were included. All of the surface breaking flaws grow at the same general corrosion rate as each sampled patch. Growth of the defects at the general corrosion rate of the patch is a conservative assumption. Therefore, patches with preexisting defects would be subject to stress corrosion cracking earlier than patches without defects.

Corrosion enhancement factors for microbially influenced corrosion and aging and for phase instability are applied to the general corrosion rate of the waste package outer barrier. No microbially influenced corrosion and aging or phase instability factor is applied to the localized corrosion rate because no localized corrosion occurs.

Because temperature and relative humidity do not significantly affect waste package and drip shield degradation, except in the case of relative humidity threshold for corrosion initiation, a representative set of histories for these parameters were used in the current analysis. In addition, the threshold for localized corrosion initiation for the drip shield and waste package outer barrier, which requires the presence of drips, is much higher than the conditions expected in the repository. The stainless steel inner layer of the waste package was not considered in the analysis.

Additional waste package degradation model improvements were implemented after completion of the TSPA-SR analyses. After incorporation of new analyses, models, and data, the predicted performance against general and localized corrosion of the waste package was shown to improve. The TSPA-SR model shows no waste package failures before 10,000 years (CRWMS M&O 2000a, Section 5.3). The supplemental TSPA model conservatively included a small number of waste package failures (up to three) due to improper heat treatment of welds. These failures lead to small calculated doses before 10,000 years. Excluding those early failures, the supplemental TSPA models show significant improvement in waste package lifetime (BSC 2001a, Section 7.5; Williams 2001a, Figure 6-1). The results of the additional analyses provide added confidence that the TSPA-SR model is conservative and that the behavior of the waste package in the repository environment is likely to be better than that predicted by the TSPA-SR model.

The various cases analyzed for waste package and drip shield degradation constitute 100 realizations of WAPDEG simulation (or 100 WAPDEG runs) that used 100 inputs for uncertain corrosion model parameters sampled from their respective range. The parameters used in the analysis are:

The DOE codisposal waste package and the naval spent nuclear fuel waste package have a thicker outer barrier than commercial spent nuclear fuel packages. Therefore, the calculation of patch penetration times by general corrosion of those waste packages is conservative (i.e., the model calculates faster corrosion than is expected). However, radionuclides that contribute most to the peak dose (technetium-99, iodine-129, neptunium-237, and plutonium-239; see Section 4.4.2.2) from DOE spent nuclear fuel and high-level radioactive waste in the codisposal waste packages represent only about 10 percent of the total inventory of radionuclides in the repository (CRWMS M&O 2000bm, Table 3.1-4). Therefore, the delayed patch penetrations of the codisposal waste packages by general corrosion would not affect the peak dose significantly. In addition, because stress corrosion cracks, once initiated, propagate very rapidly, the crack penetration times by stress corrosion cracking in the closure lid welds would not be affected significantly for the codisposal and naval spent nuclear fuel waste packages with the thicker outer barrier. The WAPDEG analysis results (i.e., waste package and drip shield failure time and number of crack, pit, and patch penetrations) are reported as a group of degradation profile curves that represent the potential range of the output parameters.

The optimum case (the upper bound profile, which is the upper extreme of the probable range of the failure time) indicates that the earliest possible waste package failure time is about 51,000 years. An extremely low probability is associated with the estimated earliest possible failure time. The failure time of the median profile is about 80,000 years. The time to fail 10 percent of waste packages for the two profiles is about 80,000 and 97,000 years, respectively.

Because conditions for localized corrosion do not develop, degradation of the drip shield occurs primarily because of general corrosion. Stress corrosion cracking of the drip shield due to rockfall is not expected to affect drip shield functions (CRWMS M&O 2000cu, Section 6.5.5). Both the top and under sides of the drip shield are exposed to emplacement drift conditions and subject to corrosion. In addition, both sides are assumed to experience the same exposure conditions, regardless of whether the drip shields are dripped on or not. Results of the WAPDEG analysis show that for the optimum case (i.e., the upper bound profile), the drip shield failure starts at about 24,000 years, and 50 percent of the drip shields fail within several thousand years after the initial failure.

The most conservative case analysis was to evaluate the effects of the alternative conservative model abstractions of several key corrosion model parameters. Those parameters are stress corrosion cracking-related parameters and general corrosion parameters, along with corrosion rate bias to account for silicate deposits. This case represents the worst case combination of those parameters from the perspective of first waste package failure time. As shown in Figure 4-92, the results of this case indicate that in the TSPA-SR model the earliest possible failure time of a waste package for the upper bound profile is about 12,000 years, much earlier than the realistic case (about 50,000 years) (see Section 4.4.2.2 for the analysis result summary). This estimated earliest possible failure time has a very low probability. The results also show that the initial failure comes from a stress corrosion crack penetration. The failure time of the median profile is about 50,000 years. The time to fail 10 percent of waste packages for the upper bound profile is about 22,000 years. These results do not include the possibility of small releases resulting from early failures described in FY01 Supplemental Science and Performance Analysis (BSC 2001a).

For the most conservative case, the failure profiles of drip shields are not significantly affected because the failures occur only by general corrosion. The difference between the conservative and optimum cases consists only of bias correction for the silicate deposits. As with the optimum case analysis, both the outer and inner sides of the drip shield are exposed to the same conditions in the emplacement drift and subject to general corrosion. The results show that for the upper bound profile, drip shield failure starts at about 20,000 years, and 50 percent of the drip shields fail within a thousand years after the initial failure.

The realistic case represents what is achievable for fabrication and material parameters under present-day processes. The results for this case fall between those of the optimum and most conservative cases.

Analysis Summary—The candidate materials for the drip shield (Titanium Grade 7) and the waste package outer barrier (Alloy 22) are highly corrosion resistant. Under the expected repository exposure conditions, these materials are not expected to be subject to degradation modes that, if initiated, could lead to failure in a short time. Those degradation modes are localized corrosion (pitting and crevice corrosion), stress corrosion cracking, and hydrogen-induced cracking (applicable to drip shield only). Both drip shield and waste package degrade by general corrosion at a very low passive dissolution rate. The current experimental data and detailed process-level analyses, upon which the model abstractions that have been incorporated in the WAPDEG analysis are based, indicate that, except for the closure lid welds of the waste package, the candidate materials would not be subject to those rapidly penetrating corrosion modes under the expected repository conditions. The waste package and drip shield degradation analyses for the two "end-member" cases (i.e., the optimum case and the most conservative case) have shown that, based on the TSPA-SR corrosion model abstractions and assumptions, both the drip shields and waste packages do not fail within 10,000 years. From the perspective of first waste package failure, the analysis results are encouraging because the most conservative case represents the worst case combination of key corrosion model parameters that significantly affect the long-term performance of waste packages in the repository. The most conservative case accounts for potential uncertainties in the fabrication and process control steps, such as stress mitigation. However, the estimated long lifetime of the waste packages in the TSPA-SR model analysis is attributed primarily to two factors: (1) stress mitigation to substantial depths in the dual closure lid welds and (2) the very low general corrosion rate applied to the closure lid welds to corrode the compressive stress zones. Complete stress mitigation as proposed may not be possible for the closure lid welds, and because of the potential residual stresses, the closure lid welds may be subject to stress corrosion cracking. Once a stress corrosion crack begins, it penetrates the closure lid thickness in a very short time. Thus, stress mitigation in the closure lid welds is a key design element to avoid premature waste package failures from stress corrosion cracking.

Movement of Water into Breached Waste Packages to Contact Waste—The temperature conditions in the waste package over time (before and after waste package breach) are used to calculate the number of spent nuclear fuel rods that develop pinholes due to creep rupture, in addition to those that have pinholes prior to being delivered to the repository. If waste package breaches occur while temperatures are high, oxygen can enter the waste package, and thermally-driven oxidation of the spent fuel pellets exposed through the pinholes can cause large length cracks (unzipping) in a significant number of fuel rods. The amount of cracking is very small if the waste package breaches occur after temperatures fall, which is the more likely situation. In the TSPA-SR, the amount of water that is calculated to have access to the waste package interior is based on the degraded areas of both the drip shield and the waste package. The distribution of water within the waste package is modeled using a conservative mixing cell approach. Resident water within the cell is based on the water fraction in saturated corrosion products of the spent nuclear fuel of the high-level radioactive waste glass. No credit is taken in this model for the potential for water to not fully wet the large surface area available in the waste package interior or the possibility that residual heat in the waste package could evaporate the water before it contacts the waste. The mobilization of radionuclides by the water from the exposed waste (through the cladding pinholes and cracks) is described in Section 4.2.6.

Once radionuclides are mobilized, they are transported to the exterior of the waste package by a simple one-dimensional diffusion model. No credit is taken in this model for the time it could take for the radionuclides to reach the inner surface of the waste package breach location or for sorption by corrosion products within the waste package. The diffusion model uses the crack width to determine the cross-sectional area, the thickness of the closure weld for the diffusion distance, the mixing-cell concentration as the source concentration, and a zero sink concentration (by assuming advection of the contaminated effluent from the outer surface of the crack). The output of this diffusion model is the source term for the sequential engineered barrier system, unsaturated zone, and saturated zone transport calculations described in Sections 4.2.7, 4.2.8, and 4.2.9.

4.2.5 Water Diversion Performance of the Engineered Barriers

The water diversion functions of the engineered barrier system are to limit or delay the amount of water contacting waste packages for at least 10,000 years (CRWMS M&O 2000ab, Section 1.2.1.8) and to increase the predictability of environmental conditions at the waste package surface. The engineered barrier system components that will perform these functions include the drip shield, the invert (consisting of a steel support structure with crushed rock ballast), the waste package pallet, and the steel ground support. Engineered barriers are described in Section 2.4.

This section describes how the engineered barriers in the potential repository will function together with the natural barriers to minimize water contact with waste packages. Over time, the drip shields and waste packages will become susceptible to corrosion, as discussed in Section 4.2.4. Once these processes result in breaches in the drip shield and the waste package, water can contact the waste form and mobilize radionuclides. The water diversion model described in Total System Performance Assessment for the Site Recommendation (CRWMS M&O 2000a, Sections 3.6.1 and 3.6.2) identifies flow pathways and calculates, among other things, leakage through a breached drip shield as well as leakage from a breached waste package. Another aspect of engineered barrier system performance is to moderate the transport of released radionuclides from breached waste packages to the host rock at the drift wall. This will be accomplished for the case of a breached waste package under an intact drip shield using a diffusion barrier concept (see Section 4.2.7). For the case of a breached drip shield, transport will be moderated through partial performance of the drip shield by reducing the amount of water that contacts the waste package.

Finally, free drainage from the drifts is important for engineered barrier system performance. Sufficient drainage capacity will prevent partial inundation of waste packages or their supports, promote diffusion-barrier performance of the invert, and prevent saturated flow conditions that could lead to faster transport for released radionuclides through the engineered barrier system and the host rock. Complete saturation of the invert is possible if seepage inflow exceeds the drainage capacity of the floor.

As noted in Section 4.1.4, the DOE has conducted sensitivity studies of the operation of the repository at lower temperatures. The conceptual basis and model abstractions presented in this section reflect the effects of higher-temperature operating conditions, specifically those implemented in Total System Performance Assessment for the Site Recommendation (CRWMS M&O 2000a, Section 3.6). Alternative thermal operating modes and/or conservatisms and conceptual uncertainties in the model have been reevaluated since the TSPA-SR model and are reported or summarized in FY01 Supplemental Science and Performance Analyses (BSC 2001a, Sections 8 and 15; BSC 2001b, Sections 3.2.6 and 4.2.6).

The issues reevaluated since the TSPA-SR model include evaporative reduction of seepage, condensation under the drip shield, geometric constraints on advective flux into the waste package, and the effect of allowing accumulation of water within the waste package (i.e., the "bathtub effect").

4.2.5.1 Conceptual Basis

Initially, preclosure ventilation will remove heat and moisture from the emplacement drifts. Ventilation at ambient temperature in the Exploratory Studies Facility has been observed to dry the surrounding rock (Section 4.2.2), and even more drying of the rock will occur during preclosure operations because of the duration and the elevated temperature. After permanent closure, the emplacement drifts and the surrounding rock will heat up, and depending on the thermal loading, the dryout zone will extend around the drifts. Eventually, within approximately 1,000 years (for the mean and upper infiltration distributions, see Section 4.2.2), moisture will return to the surrounding rock. At a few locations in the potential repository, seepage of liquid water into the drifts may occur. The drip shield will prevent such seepage from contacting waste packages.

Seepage in the future may increase because of altered climate conditions or episodic fluctuations in infiltration at the ground surface. In addition, at a limited number of locations in the potential repository, faults or fractures may focus percolation into an emplacement drift. Drainage capacity will ensure water diversion performance for extreme seepage conditions.

Conceptual Basis for Types of Flow on the Drip Shield and Waste Package—Water that enters the emplacement drifts as seepage can flow along three types of pathways: (1) water flow directly to the invert; (2) water flow that contacts the drip shield, but is diverted to the invert; and (3) water flow through breaches in the drip shield or waste package. Breaches in the drip shield or waste package can take different forms, ranging from fine cracks caused by stress corrosion cracking of the waste package to patches where general corrosion has penetrated a larger area of the drip shield or waste package (Section 4.2.4). Conceptual features related to film flow and droplet formation are discussed in this section, some aspects of which are not directly modeled (CRWMS M&O 2001c, Section 6.1). As outlined in the process model and abstraction sections 4.2.5.3 and 4.2.5.4, the TSPA-SR conceptual model for fluid flux through the drip shield and waste package bounds many of the discussed conceptual processes using simplified and conservative approaches (CRWMS M&O 2000a, Sections 3.6.1 and 3.6.2).

Thin films can be predicted to form on any wetting or partially wetting surface that is exposed to humidity (Middleman 1995, Chapter 9). Films may then flow because of gravity or possibly because of other differences in water potential on the surface. Film flow capacity is limited by viscosity, the extent of the flow area, surface roughness, and other factors. Film flow on the engineered barriers can deliver liquid water at significant flow rates and could be significant to waste isolation performance. However, film flow through breaches in the drip shield is likely to continue as film flow on the underside of the drip shield and thus will tend to be diverted from the waste package. Also, film flow through fine cracks into the waste package will be limited by heat generation within the waste package, which will cause water to be rejected as vapor (CRWMS M&O 2000as, Section 3.1.1).

Depending on the aperture of a breach, capillary flow and droplet flow modes can occur. For fine cracks, flow will be dominated by capillarity. In capillary flow, water will be either strongly imbibed or excluded in a crack, depending on whether the surfaces are wetting or nonwetting, respectively. If the local environment has low relative humidity, which would occur mainly during the thermal pulse, a fine crack may dry out. Molecular films of water could then form on the crack surfaces, but the flow capacity would be negligible because the films would be very thin. As the relative humidity increases during cooldown, water films will thicken, and fine cracks will become increasingly water-saturated (CRWMS M&O 2001c, Section 6.1).

As saturation increases and water bridges a fine crack or pore, it will be held there at a negative potential by capillarity. It will not flow unless the adjacent downstream flow pathway is at a more negative potential, or unless sufficient water pressure is applied upstream to overcome the negative potential. For the case of water flowing through the drip shield or through the waste package under the impetus of gravity, neither of these conditions is likely unless the water evaporates at the downstream end (CRWMS M&O 2001c, Section 6.1). Thus, the movement of water through fine cracks is limited to very thin water films that can form in the crack and, when humidity conditions permit, to evaporation of wate