A maximum of 1% of SNF discharged from reactors could be defective. Volume expansion associated with the oxidation/hydration of the SNF matrix or zirconium may crack/unzip defective cladding (Cunnane et al., 2003). Figure 7.15 shows a schematic of this unzipping process (DOE, 2002). Unzipping was observed in the Argonne National Laboratory 1.5-year long tests, caused by stress generated by corrosion product accumulation in the gap of cladding and the fuel matrix from uniform corrosion of Zircaloy cladding at 175°C (347°F) (Cunnane et al., 2003).

Oxidation/hydration may occur with either residual moisture inside the intact canister or container, or from moisture that has intruded into the failed canister or container. This cladding failure may affect the magnitude of the radionuclide release fraction and challenge the retrievability of the SNF materials, and lead to configuration changes in internal structure that impact nuclear criticality.

(b)

During reactor operations, the cladding metals, mainly Zircaloy, corrode in water. This introduces hydrogen into the Zircaloy. Hydrogen can degrade the strength of Zircaloy through overall embrittlement caused by a disper­sion of radially oriented hydrides (perpendicular to the hoop stress) (Chung, 2004). The hydrides formed during reactor operations are mostly circum­ferential hydrides (parallel to the hoop stress). Circumferential hydrides may not affect the strength significantly, depending on the magnitude of severity. However, circumferential hydrides are known to become radially reoriented in the presence of appropriate applied stress and temperature (Chung, 2004). Figure 7.12 compares hydrides oriented circumferential or perpendicular to the hoop stress and Fig. 7.13 shows ductility loss with radial hydrides (Yagnik et al., 2004).

Another hydrogen effect is delayed-hydride cracking (DHC). Small cracks that develop on the inner or outer surface of cladding may lead to crack propagation when assisted by hydrogen diffusion to the crack tip, thus forming radially oriented hydrides at the crack tip. The mechanism has not been proven to exist under dry storage conditions. Figure 7.14 shows a schematic for the mechanism of the DHC process. The crack density and size from hydrogen embrittlement of hydride reorientation and DHC can be conservatively assessed like the SCC of stainless steel described in Section 7.4.2.

This section presents the performance of cladding in aqueous disposal environments and dry storage environments. Hydrogen-induced cracking of cladding may be a major detrimental degradation mechanism for both disposal and storage conditions. Crack opening area allows radionuclide release under both conditions. Oxidation (or general corrosion) of cladding is very slow and localized corrosion is unlikely to occur in near-neutral pH disposal environments (Ahn, 1996b). Oxidation of cladding is only possible in the presence of residual water and/or oxygen in dry storage canisters. Initially defective cladding may be further cracked (unzipped) by the pres­sure imposed on it by corrosion products of the SNF matrix or zirconium
itself. Longer longitudinal cracks that develop from the initial cracking/ unzipping will increase radionuclide release under both conditions.

Table 7.1 summarizes the dissolution rates for oxidizing and reducing dis­posal environments (Ahn et al. , 2011a) used in a performance assessment model (Markley et al., 2011). A range of environmental conditions are con­sidered, mostly near-neutral pH and ambient temperature. The variation of pH and temperature can be adjusted in terms of dissolution rate as user — defined parameters. For this base case, radionuclide release is estimated combining the reducing and oxidizing environments, to simulated residual radiolysis of water by actinides in the reducing environment. Figure 7.8 shows the estimated dose from the radionuclide release for this combined case.

Considering all radionuclide release fractions from the UO2 matrix, an exercise was conducted to estimate the doses to workers or members of the public from airborne fragments of the SNF matrix caused by SNF oxidation and SNF drop/collision (after Kamas et al., 2006). The most significant dose contributor in the release fraction is aerosol SNF fines (i. e., small solid particles). Tritium, noble gases, iodine, crud, ruthenium, caesium, strontium and SNF fines were part of the source term considered. In Fig. 7.11. the

Table 7.1 Summary of SNF dissolution rates in oxidizing and reducing environments

Parameter name Value Description and basis

Spent MOX fuel is also included in the table and the reference numbers quoted are from the reference by Ahn et al. (2011a).

7.11 Example dose estimate for (a) oxidation and (b) collision (/drop) of SNF assemblies (after Kamas et al., 2006). Used with permission from American Nuclear Society (ANS).

radionuclide release fraction of the aerosol SNF fines, 2.0 x 10-6 for the drop/collision case and 1.2 x 10-3 for the SNF oxidation case, were used to estimate the dose to workers or members of the public (Ahn et al., 2011b; Kamas et al. , 2006). A site boundary was defined, for the dose to workers within the boundary and to members of the public outside the boundary.

The left figure is for SNF oxidation under normal operations. The wake effects are a modification of the radionuclide transport path right outside any storage building if any building shadow exists. Consequently, radionu­clide transport will stop. Within a short distance from the building, the radionuclide transport will not be reached. The right figure is for drop/col — lision cases. In both cases, arbitrary dose rate units are used for the log scale. The oxidation case gives a dose rate ten times higher than the collision case in the same log-scale unit.

Dry-air oxidation or humid air hydration of SNF in air or in the presence of limited amounts of groundwater may play an important role in radionu­clide releases (Ahn and Mohanty, 2008; Ahn, 1996b). The UO2 matrix will fracture (or crack) upon oxidation or hydration by volume change. Lower oxidized oxides such as UO24 will contract, whereas higher oxides such as U3O8 or UO3 hydrates will expand. Lower and higher oxides are defined here as oxides with a (O/U) ratio smaller and larger than 2.4, respectively. Fractions (e. g., 10-6-10-3) of oxidized or hydrated phases are likely to be respirable aerosol less than 10 micrometer (3.9 microinch) in size. The aerosol will increase the radionuclide release in air. The oxidized or hydro­lysed phases also increase the area of SNF surface exposed to groundwater. This increase of the exposed surface area is in turn expected to increase radionuclide release in groundwater. Similarly, mechanical impacts such as those caused by seismic events can also fragment the SNF into particles.

Figure 7.10 shows the fraction of respirable particles, depending on the impact energy absorbed. The fine-grained and porous rim structure near cladding of high burnup (above about 60 GWd/MTU) UO2 may also affect the magnitude of the radionuclide release fraction (NRC, 2007 ).

Actinides such Pu-239 or Np-237 have low solubility limits, and they are released at a concentration below or equal to their solubility limits (or colloid concentrations), which in turn are determined by the SNF matrix dissolution rate, groundwater flow rate and solubility limit.

Regarding colloid formation, Ahn (1996a) summarized the processes involved. During the dissolution of the SNF matrix, suspended solid parti­cles containing mainly actinides of low solubility may form. The colloids can carry a large amount of actinides compared with dissolved species. The traditional processes of colloid formation (especially in actinide colloids) have been investigated under near-equilibrium conditions. Most studies in this regard pertain to chemical bonding among ions. Extending the chemical bonding process in equilibrium or non-equilibrium states, colloid formation may be described in macroscopic ways by three different processes: (a) condensation, (b) dispersion, and (c) sorption (pseudo-colloid formation). Colloids may form by precipitation in small particles because of supersatu­ration of actinides or the SNF matrix (i. e., condensation). The layer of the precipitated phases on the SNF matrix can be mechanically detached into small suspended particles (i. e., dispersion). Finally, the dissolved pure acti­nide species can be sorbed on the surface of non-radioactive inert ground­water colloids (i. e., sorption). Figure 7.9 shows schematics of these three processes (CRWMS M&O, 2001).

The dissolution rates by the electrochemical and chemical processes are (Ahn, 1996a):

S

Rdis = ±kef (E) [7.2]

and

where S is surface area of the dissolving phase, V is leachate volume, ke is rate constant for electrochemical dissolution, f(E) is dissolution rate as a function of electrochemical potential E, k. is rate constant for SF dissolu­tion, Cs is effective solubility limit of dissolving phase, C. is elemental con­centration under consideration, k+ is rate constant for growth of the reprecipitated phase, F is flow rate of ground water, and Npar is formation or growth rate of colloids per unit leachate concentration.

Equation [7.2] is for electrochemical process, Eq. [7.3] is for chemical process, and Eq. [7.4] is for release rate from the dissolution processes of the first two equations.

The fractional mobilization rate is the dissolution rate multiplied by the specific surface area of the SNF matrix. Conservatively, the fractional mobi­lization rates can be assumed constant within uncertainty ranges at a given temperature. The environmental conditions are important in determining the dissolution rates, including near field water chemistry, temperature, pH, or reducing or oxidizing conditions. Important water chemistry includes carbonates, and cations such as calcium or silica species (Ahn and Mohanty, 2008).

In connecting the dissolution rate to the fractional mobilization rate, the specific surface area is determined by the average fragment size (radius) and density of the waste form. Typically, the fragment size of commercial SNF is 0.1 cm (0.04 inch) (Ahn and Mohanty, 2008).

If the temperature exceeds 100°C (212°F), solid-state oxidation or hydra­tion will occur, depending on the RH. Higher uranium oxides (UO[.4 or U3O8) that form by oxidization of the UO2 matrix dissolve at a rate similar to the unoxidized UO[ matrix. Hydrated UO[-xH2O (x = 0.8, 2) dissolves 10-20 times faster than unhydrated oxides. However, the rate of hydration (i. e., the formation rate of UO3-xH2O) is slower than the aqueous dissolu­tion rate. Ahn and Mohanty (2008) summarized the effects of oxidation and hydration on the dissolution.

This section presents the degradation behaviour of SNF in mild and near­neutral environments under (i) oxidizing or reducing aqueous disposal conditions, and (ii) in dry storage environments. During the aqueous dis­solution of SNF, highly soluble fission products such as Tc-99 or I-129 are released congruently with (i. e., in proportion to) the SNF matrix (UO2 ) dissolution. On the other hand, actinides such Pu-239 or Np-237 are released at a concentration below or equal to their solubility limits (or colloid con­centration), which are in turn determined by the SNF matrix dissolution rate, groundwater flow rate and solubility limit. Colloids are suspended solid particles of less than 1 micrometer in size that can contain actinides. An oxidizing aqueous environment promotes electrochemical dissolution of the SNF matrix in soluble species with the aid of oxidants such as dis­solved oxygen and hydrogen peroxide (Shoesmith, 2000) . In a reducing environment, the UO) matrix will dissolve chemically in soluble species (Sunder and Shoesmith, 1991). Generally, the electrochemical dissolution rate is faster than the chemical dissolution rate. In the presence of radiolysis effects, the SNF matrix may dissolve in either an electrochemical or a chemical process, depending on the magnitude of the radiolysis (Ahn et al, 2011a). In conjunction with container failure and sorption and/or flow behaviour of backfill, the SNF matrix dissolution serves as the source term of radionuclide release in the PA. In a dry storage environment, mechanical degradation of the SNF matrix could occur by air oxidation/humid air hydration or impact fragmentation upon the canister failure under normal conditions (e. g., SCC failure) or external hazard conditions (e. g., aircraft or seismic impact). In the canister, if incomplete drying of SNF assemblies occurs, the residual water may increase RH sufficiently to oxidize (by oxygen from the radiolysis of water molecules) or hydrate the SNF matrix. With severe external hazards, high temperatures or impact stress may frag­ment the SNF matrix by oxidation or mechanical disintegration. The respir­able SNF particles (i. e., suspended aerosol, less than 10 pm [3.9 microinch] in size) produced by the fragmentation serve as the primary source term for radionuclide release in air.

Figure 7.7 shows an example of the radionuclide release fraction to the environment from a cask (Sprung et al. , 2000). Casks include canister and other overpacks. Strictly speaking, this figure was constructed for a trans­portation cask. Nevertheless, the radionuclide release behaviour would be similar in the storage cask. The radionuclide release fraction is expressed by (1.0 — Retention). In this range of the surface opening area, the surface

7.8 Example в-SOAR (Markley et al., 2011) dose results for only commercial SNF using combined degradation rate in a stylized reducing geological disposal system (Ahn et al., 2011a). Used with permission from American Nuclear Society (ANS).

opening is already wide enough to result in the release fraction approaching to 1. This retention mechanism is in addition to the low radionuclide release fraction from the degraded UO2 matrix and the failed cladding. In reality, the surface area opening by SCC may be smaller because the model of Eq. [7.1] is conservative.

Figure 7.8 (calculated using the в-SOAR model of Markley et al., 2011) shows an example of a calculation of radionuclide release from the seismic — induced SCC of various disposal containers (Gwo et al., 2011). There is an additional factor lowering the magnitude of radionuclide release due to the restricted perforation made by SCC.

Salt deposits on the canister surface open to the environment may be sig­nificant in coastal areas. SCC of the stainless steel canister needs to be considered when the relative humidity (RH) in air is appropriately high, the amount of salt deposits is sufficient to form aggressive and sufficient aqueous conditions at welds, and when a sufficient tensile stress is present. If the RH is too low, the aqueous condition would not exist. On the other hand, if RH is too high, the chloride concentration would not be high enough to initiate SCC. The weld area could have residual tensile stress and sensitized microstructure which is prone to SCC.

The RH of the environment surrounding the canister surface and salt deposits depends on the canister surface temperature. Over a long time, the surface temperature will decrease as the radioactivity inside the canister gradually decays. This will result in increasing RH of the environment immediately adjacent to the canister surface. Also, temperature, RH and the amount of salt deposits will not be homogeneous on the canister surface because of the SNF storage configuration and air flow surrounding the canister. In addition, primarily the weld areas will be susceptible to SCC. Considering these environmental and materials factors, the probability associated with SCC could be low enough for it to be screened out from performance assessment (PA), especially with appropriate remediation.

If SCC were to occur, radionuclide releases may be primarily caused by the release of aerosol radioactive materials, which may in turn be driven by the pressure of inert fill gas and fission gas inside the canister (from prior release from failed cladding). The release rates are also affected by the opening area of the canister surface caused by SCC. The SCC area density per weld area of the canister may be estimated conservatively (the estimate was originally under seismic events) by the following equation (Gwo et al., 2011):

5 = C a/E [7.1]

where 8 is crack areal density (m2/m2), a is applied stress (MPa), E is Young’s modulus (MPa) and C is geometric constant.

For example assuming no inspections and remediation, a calculation for stainless steel using Eq. [7.1] suggests that the crack areal density per unit weld area is approximately 1.2 x 10-3 at 170-310 MPa (25-45 ksi) of applied stress, (193-207) x 103MPa [(28-30) x 103ksi] of Young’s modulus (Gwo et al., 2011). The weld area fraction is about 10-2-10-1 of total surface area (ASM International, 1993). In a canister surface area of about 30 m2 (4.6 x 104 inch2), the surface opening area will become 3.6 x (102-103) mm2 (0.56­5.6 inch2). The model in Eq. [7.1] is conservative, assuming a distribution of uniform crack size. In reality, the number and size of cracks are likely to be smaller. This calculated area is obviously larger than that allowed for leak tightness (Institute for Nuclear Materials Management, 1997).