Fuel Cladding and Spent Nuclear Fuel

Transcription

1 Fuel Cladding and Spent Nuclear Fuel

2

3 CLADDING EVALUATION IN THE YUCCA MOUNTAIN REPOSITORY PERFORMANCE ASSESSMENT Eric R. Siegmann*, J. Kevin McCoy**, Robert Howard** *Duke Engineering and Services, Framatome Technologies, *** TRW Environmental Safety Systems, (all) 1211 Town Center Drive, Las Vegas, NV ABSTRACT The Yucca Mountain Project (YMP) 1998 Total System Performance Assessment Viability Assessment (TSPA-VA) analyzed the degradation of Zircaloy clad commercial fuel rods and the resulting exposure of the fuel in the event of a waste package failure. The cladding degradation mechanisms considered were damage before emplacement, mechanical failure from drift collapse, localized corrosion, general corrosion, delayed hydride cracking (DHC), hydride reorientation, creep rupture, and stress corrosion cracking (SCC). The potential for further cladding degradation due to cladding rupture as a result of fuel oxidation was also considered in the modeling effort. These models have been improved for use in future TSPAs. The current cladding degradation model divides the analysis into two phases, cladding failure (perforation) and cladding unzipping (crack propagation caused by the expansion of UO2 fuel after reaction with water). Cladding failure occurs during reactor operation, from creep strain failure during high temperature periods in dry storage or in the early periods in the repository, or localized corrosion. After a Waste Package (WP) containing spent nuclear fuel in the repository fails, moisture is assumed to enter the waste package and the failed cladding starts to unzip (tear open) from the formation of secondary uranium phases. This slowly exposes the fuel. In addition, the inventory of fission products located in the gap between the cladding and fuel pellet is rapidly released. The cladding model limits the amount of fuel that is exposed to moisture and becomes available for dissolution. As a result, the doses to the affected population are reduced (factor of 20 to 50 in TSPA-VA) from the case where cladding is not considered. INTRODUCTION Earlier studies have evaluated cladding degradation under repository conditions U3 and dry storage 4 ' 5 conditions which are similar to early repository conditions. Experiments also measured the releases from damaged cladding. 6 " 7 The Yucca Mountain Project (YMP) 1998 Total System Performance Assessment Viability Assessment (TSPA-VA) included cladding degradation as part of the fuel degradation modeling. TSPA-1995, a previous analysis of repository performance, neglected the presence of cladding, as did most earlier PAs. When cladding was neglected, all the fuel in the waste package (WP) was considered available for dissolution at the speed of the intrinsic fuel dissolution rate. For some radionuclides, solubility limits were reached which controlled the rate of those radionuclides' leaving the WP. In the current TSPA model the cladding is considered an integral part of the waste form. CLADDING MODEL The cladding model summarizes numerous studies of cladding degradation and is incorporated into the TSPA computer model. The model describes two phases, cladding perforation and cladding unzipping. Cladding perforation is the formation of small cracks or holes in the cladding from various sources ranging from failures during reactor operation to cladding creep rupture during repository storage. Perforation permits the fuel inside the cladding Mat. Res. Soc. Symp. Proc. Vol Materials Research Society 3

4 to begin to react with the moisture or air and leads to the cladding unzipping phase. In the unzipping phase, the cladding is torn open by the formation of secondary mineral phases on the fuel, and the radionuclides are available for release. The various components of the model are discussed below. Cladding Condition as Received The initial cladding condition analysis describes the condition of the commercial nuclear fuel as it is expected to be received at the YMP site. This analysis generates the initial boundary condition for the subsequent analysis of degradation of the cladding in the repository. It also evaluates the fraction of fuel rods that are perforated before emplacement and are immediately available for cladding unzipping when the WP fails. Earlier studies of cladding initial conditions have been performed 2 ' 4 ' 8. The TSPA-VA used a single value for these initiating conditions but statistical distributions have since been developed. The cladding degradation model is based on the Westinghouse 17x17 rod fuel design. This design represents over 30% of the PWR fuel discharged to date and also has the thinnest Zircaloy cladding. It is assumed that the BWR cladding degrades in a similar manner. This is conservative since BWR cladding is thicker and is discharged with lower burnups and stresses. In addition, most BWR assemblies are enclosed in flow channels (sheet metal boxes) which offer additional protection. Starting with a distribution of PWR fuel burnups that are anticipated for storage at YMP, this model develops distributions for various cladding properties. Table 1 summarizes these distributions and includes the mean and upper 5% values. Table 1 Model Results of Expected Fuel Stream into YMP Property Burnup Internal Pressure Oxide Thickness Peak Hydride Content Crack Size Stress (27 C) Stress Intensity Factor, Ki Mean Value 44.1 MWd/kgU 4.8 MPa 54fim 358 ppm 19 urn 38.4 MPa 0.47 MPa-m u * Upper 5% Value 63.3 MWd/kgU 7.3 MPa 112^m 738 ppm 57 im 61.8 MPa 1.08MPa-m U5 A distribution for the fraction of cladding within a WP that failed as a result of reactor operation was developed from the fraction of rods failed as a function of calendar years by assuming that the fuel assemblies are loaded into WPs in their order of discharge from the reactor. This loading sequence tends to place fuel with high failure rates (BWR fuel in 1970, also , and PWR fuel in 1972,1983, and 1989) into consecutive WPs and produces larger variations in rod failure fractions than would be expected with thermal blending. A factor of four uncertainty was applied to represent the uncertainty in rod failure data to address incipient failures of the surrounding four rods in the square array assembly. The creep failure analysis included rod failure from dry storage and transportation using temperature profiles starting at 350 C. This analysis shows that a small fraction of the fuel with high stresses would fail if exposed to design basis storage and shipping temperatures. Table 2 gives the calculated percentage of rods that have failed cladding at emplacement. These fuel rods will undergo cladding unzipping and fuel dissolution when the WP fails.

5 Table 2 Model Prediction of Percent and Cause of Rods failed in a WP Rod Failure Mode Reactor Operation incl. Incipient Failures Pool Storage Dry Storage Dry Storage & Transportation, Creep' Dry Storage & Transportation, DHC Transportation (Vibration, Impact) Fuel Handling Total Matsuo's creep correlation used. Percent of Rods Failed/WP (Range: 0.0 to 21.1) (Range 0.1 to 4.9) (Included in above) 0.62 (Range: 0.16 to 26) Localized Corrosion Corrosion of zirconium has been observed in fluoride-containing environments. Since fluoride is present in Yucca Mountain groundwater, fluoride corrosion may occur in waste packages. Two scenarios for fluoride corrosion have been considered. In the first (bathtub scenario), the waste package is full of water, and fluoride ions are transported to the cladding by aqueous diffusion. In the second (flow-through scenario), water enters the waste package through a breach on the top and drips out through a breach on the bottom. These two scenarios represent extremes of the rate of drainage. The flow-through scenario is the more severe of the two. In this scenario, fluoride can be rapidly transported through the waste package by advection, whereas in the bathtub scenario it is transported by diffusion, which is a comparatively slow mechanism. In the flow-through scenario, advective flow is directed downward by gravity, so fluoride attack can be localized on a relatively small area of cladding. In contrast, diffusion does not have a preferred direction, so the fluoride can be transported to a large volume of the waste package in the bathtub scenario. Spreading the fluoride over a larger area of cladding means that more fluoride will be consumed in breaching each fuel rod. Since the flow-through scenario is more severe, the bathtub scenario was not considered further. A bounding approach has been used to describe the flow-through scenario. It might be expected that the corrosion of zirconium is sufficiently slow and the flow of groundwater through the waste package is sufficiently fast that some fluoride will simply flow through the waste package without reacting. Credit has not been taken for this loss of fluoride. Instead, it is assumed that corrosion of the cladding is limited by the supply of fluoride. In determining the amount of fluoride that is necessary to breach a fuel rod, it is assumed that fluoride removes all the cladding from a 10-mm length of the fuel rod by reacting to form ZrF4. The as-manufactured thickness of the cladding may be used because, although some of the zirconium may be oxidized, the zirconium atoms remain in the products of corrosion. Fluoride attack is assumed to completely degrade one fuel rod before degradation begins on another rod. This assumption is conservative because rods breach as soon as enough fluoride is available; there is no delay in breaching one rod because fluoride is being diverted to start degrading another.

6 The resulting model is that the fraction of fuel rods failed by fluoride corrosion starts at zero when the waste package is breached. After breach, the fraction failed is proportional to the volume of water that has entered the package, reaching one when 2400 m 3 of water has entered the waste package. An alternative description is that the fraction of fuel rods that fail in a given year is the volume of water that enters the waste package during that year divided by 2400 m 3. Upper and lower limits are 10 times and 1/10 of the best estimate rate to represent the uncertainties in this model. This analysis makes the rod failure fraction linearly dependent on the water ingression rate (% failed = x m 3 water entering the WP). The water ingression into the WP increases with time as additional patches open. Rod failure rate also depends on the location of the WP group because of different drip rates apply in different repository regions. As an example, with 50 liters/year of J-13 water (2.2 ppm fluoride) entering the WP, 20% of the rods would fail by fluoride corrosion in 10,000 years. Creep Failure Repository design features such as backfill, drip shields, or thermal loading affect the fuel temperature. A statistical distribution, of rod properties has been developed so that creep failure is included in the model. In the creep analysis, the rods were exposed to a temperature history that includes 20 years of dry storage starting at 350 C, three weeks of transportation at 350 C, and then a temperature history for the repository. The temperature profile within a WP is handled by considering rods in six zones across the WP. Matsuo's creep correlation 10 was used although more recent analyses using Murty's correlation 3 gave similar results. The strain failure criteria was a distribution based on eighteen tests of irradiated cladding 11 with a mean failure strain of 3.3%, and a range from 0.4% to 11.7%. Figure 1 gives the fraction of rods failed by strain as a function of WP surface temperature. It shows a high activation energy for creep and suggests a basis for the cladding temperature limit of 350 C, since the peak cladding temperature is about 50 C to 60 C above the WP surface temperature. The plateau on the left side of Figure 1 gives the fraction of rods that have failed during dry storage and transportation. Using Murty's creep correlation would increase the mean value from 0.46% to 2%. Mechanical Damage Seismic failures (fuel failure within an intact WP from seismic motion) and mechanical loading from a rubble overburden was analyzed. The seismic analysis showed that the rods failed only from very severe earthquakes (once per million year events), but then most of the rods would fail. Therefore the seismic failures are treated as disruptive events, and when such an earthquake occurs, all cladding is failed and available for unzipping. The analysis of mechanical loading from a rubble bed (after drift collapse and WP degradation) showed that the rods would fail under these conditions. The current waste package design is predicted to offer structural protection for hundreds of thousands of years and therefore, failure from this mechanism is not included. Based on the results of the seismic analysis, damage from rock fall is also unlikely. Other Failure Mechanisms A review of the various hydride degradation mechanisms was performed. Delayed Hydride Cracking (DHC) of existing cracks was analyzed using the distribution of stresses and crack sizes summarized in Table 1. Stress intensity factors with a mean of 0.47 MPa-m 05 (range to 2.7 MPa-m 5 ) were calculated; these are below the threshold stress intensity factors,

7 which are in the range of 5 to 12 MPa-m 5. Therefore, crack propagation by DHC is not expected. These stress intensities are also below those needed to produce SCC in all but the highest stressed rods. SCC is addressed in a later analysis. Hydride reorientation has not been modeled. Cladding Unzipping and Fuel Dissolution In TSPA-VA, the fuel in rods that were failed before emplacement was assumed to be completely exposed for dissolution while the fuel in exposed patches was available for dissolution but the remaining ends of the rods were not. This model was not necessarily conservative and is being upgraded. Fuel rods with perforated cladding are expected to remain intact until the WP fails and permits air and moisture to enter. While the humidity is low, dry unzipping could occur. Since the WP is expected to last for at least 200 years, the fuel temperatures will be too low for dry unzipping (fuel conversion to U3O8) to occur. Wet unzipping is modeled to start at WP failure. The fuel matrix is dissolved at the intrinsic dissolution rate and precipitates locally as metaschoepite. This secondary phase isolates most of the fuel from the moisture but the fuel in the torn cladding region continues to react, increasing volume, and forcing the tear further along the cladding. This reaction region is cone shaped and propagates along the rod at approximately 40 times the intrinsic dissolution rate. It is assumed that the perforation is in the center of the rod. This maximizes the release rate. Figure 2 gives the time to unzip a rod as a function of temperature. The unzipping time is also a function of local chemistry and ph. In addition, the gap inventory is instantly released when the cladding is perforated. Figure 1. Creep Failures vs. WP Temperature Figure 2. Rod Unzip Times vs. Temperature r--} Peak WP Surface Temperature, C WP Temperature, C

8 RESULTS AND CONCLUSIONS Earlier TSPAs modeled the waste form as bare UO 2 which was available for dissolution at the intrinsic dissolution rate. Water in the WP quickly became saturated with many of the radionuclides, limiting their release rate. In the current TSPA, cladding is modeled as part of the waste form and limited the amount of fuel available at any time to dissolve. The major components of cladding perforation were failure in reactor operation, creep failure, and localized corrosion. The cladding then unzips from the production of secondary uranium phases. The cladding model limits the amount of fuel that is exposed to moisture and becomes available for dissolution. As a result, the doses to the affected population are reduced (factor of 20 to 50 in TSPA-VA) from the case where cladding is not considered. ACKNOWLEDGEMENTS The authors are grateful to Te-Lin Yau and Michael G. Bale for their localized corrosion studies, Hee M. Chung for the hydride analysis, William J. O'Connell for his unzipping analysis, and Steven A. Steward for the intrinsic dissolution modeling. REFERENCES 1. T.M. Ahn, G.A. Cragnolino, K.S. Chan, and N. Sridhar, Scientific Bases for Cladding Credit in the High-Level Waste Management at the Proposed Yucca Mountain Repository, in Scientific Basis for Nuclear Waste Management XXII, edited by J. Lee and D. Wronkiewicz, (Mater. Res. Soc. Proc. 556, Warrendale, PA ,1999). 2. S. Cohen & Associates, Effectiveness of Fuel Rod Cladding as an Engineered Barrier in the Yucca Mountain Repository, S. Cohen & Associates, McLean, Virginia, P.J. Henningson, Cladding Integrity Under Long Term Disposal, Doc. ID: , Framatome Technologies, Lynchburg, VA, M.E. Cunningham, E.P. Simonen, R.T. Allemann, I.S. Levy, and R.F. Hazelton, Control of Degradation of Spent LWR Fuel During Dry Storage in an Inert Atmosphere, PNL-6364, Pacific Northwest Laboratory, Richland, Washington M. Peehs, Assessment of Dry Storage Performance of Spent LWR Fuel Assemblies with Increasing Burn-Up. Erlangen, Germany: Siemens KWU-NBT. Co-ordinated Research Program (CRP) on Spent Fuel Performance Assessment and Research (SPAR), First RCM held in Washington DC-USA, April 20-24, C.N. Wilson, Results from Cycles 1 and 2 of NNWSI Series 2 Spent Fuel Dissolution Tests, HEDL-TME-85-22, Richland, Washington: Westinghouse Hanford Company, C.N. Wilson, Results from NNWSI Series 3 Spent Fuel Dissolution Tests, PNL-7170, Richland, Washington: Pacific Northwest Laboratory, 1990.

9 8. T.L. Sanders, K.D. Seager, Y.R. Rashid, P.R. Barrett, A.P. Malinauskas, R.E. Einziger, H. Jordan, T.A. Duffey, S.H. Sutherland, and P.C. Reardon, A Methodfor Determining the Spent- Fuel Contribution to Transport Cask Containment Requirements, SAND , Albuquerque, New Mexico: Sandia National Laboratories, E. Hillner; D.G. Franklin,; and J.D. Smee,. The Corrosion of Zircaloy-Clad Fuel Assemblies in a Geologic Repository Environment. WAPD-T-3173, West Mifflin, Pennsylvania: Bettis Atomic Power Laboratory Y. Matsuo, "Thermal Creep of Zircaloy-4 Cladding Under Internal Pressure." Journal of Nuclear Science and Technology, 24 (1\ Tokyo, Japan: Atomic Energy Society of Japan, H.M. Chung, F.L. Yaggee, and T.F. Kassner, "Fracture Behavior and Microstructural Characteristics of Irradiated Zircaloy Cladding." Special Technical Publication, 0 (939) 775. Philadelphia, Pennsylvania: American Society for Testing and Materials, 1987.

10