ML080280072

Update on North Anna Zirlo Characterization and LOCA Embrittlement Testing

ML080280072
Person / Time
Site:	North Anna
Issue date:	01/08/2008
From:	Billone M, Burtseva T, Yan Y Argonne National Lab (ANL)
To:	Office of Nuclear Regulatory Research
References
Job Code N6282
Download: ML080280072 (15)

Text

1Update on North Anna ZIRLO Characterization and LOCA Embrittlement Testing M. C. Billone, Y. Yan and T. A. Burtseva January 8, 2008

1. Pre-Test Characterization Figure 1 shows the axial locations of North Anna ZIRLO Rod AM2-L17 from which Studsvik sectioned and defueled 80-mm-long cladding segments for ANL. Segment axial orientation relative to the bottom of the fuel rod was not maintained during defueling. Characterization results for LOCA pre-test planning have been generated for samples sectioned from segments labeled 5, 6 and 8 in Fig. 1.

Results for corrosion layer thickness and/or oxygen concentration were used to estimate the axial location of these samples.

Table 1 summarizes the characterization results for corrosion-layer th ickness, fuel-cladding-bond layer thickness, LECO-measured oxygen content, and LECO-measured hydrogen content. For LECO measurements, short cladding rings (2-mm-long) were sectioned and then snipped into four quarter rings. The hydrogen or oxygen content and mass were measured for each quarter ring. This mass included the masses of the corrosion layer, the cladding metal and the fuel-cladding bond. The average hyd rogen or oxygen content for a ring was determined from the total hydrogen or oxygen evolved from the four quarter-rings normalized to the combined masses of the rings. The four measurements per ring were also used to determine the one-sigma circumferential variation in hydrogen content.

As shown in Table 1, there is considerable circumferential variation in hydrogen concentratio n at any one axial location within the uniform burnup region. If the data point at 2803 mm (from the rod bottom) is ignored because the location appears to be close to a grid spacer, the axial average of the remaining 5 sets of data points is 620+/-50 wppm. The circumferential variation for Segment 5 and 6 samples is more significant ( +/-150 wppm) than the axial variation along the segments. Me tallographic results for a Segment-6 sample were consistent with LECO-hydrogen results: dense hydride rim near the cladding outer surf ace varied in equivalent thickness from about 40 to 70 µm around the circumference.

Figure 2 shows a high-magnification micrograph of the corrosion layer at 2909 mm from the rod bottom, as imaged from a polished met samp le. Indications of circumferential and radial cracks may be due to sample mounting in epoxy - drying and contraction of epoxy may cause or exaggerate circumferential cracks - and/or the effects of grinding and polishing. The corrosion layer thickness (43+/-2 µm) was measured for eight such locations at this position. Figure 3 shows a high-magnification and high-contrast image of the fuel-cladding bond layer in the as-polished condition. Based on eight images around the circ umference of the etched cross section, the bond-layer thickness was measured to be 7+/-2 µm. Because of the highly nonuniform bond-fuel interface, the reported bond thickness is that of an equivalent oxide bond with a smooth fuel interface. The cladding metal wall thickness was measured to be 544+/-2 µm. Figure 4 shows the composite of the eight images taken around the circumference of the etched cross section at the 2 Fig. 1. Sectioning diagram for North Anna ZIRLO defueled cladding segments sent to ANL from Studsvik in second shipment. Corresponding ANL ID numbers are 648E, 648F and 648H for Segments 5, 6, and 8, respectively.

3Table 1 Summary of North Anna ZIRLO Characterization Results for Rod AM2-L17 Sample ID # W (ANL) Location from Bottom mm Corrosion Layer from Eddy Cur. (EC), µm Corrosion Layer from Metallographyµm Fuel-Clad Bond Layer from Met. µm LECO Oxygen Content wt.% LECO Hydrogen Content wppm 5 (648E) 2800- 2880 38-42 ---

--- ---

--- 1.31+/-0.24

@2801 --- 464+/-71 @2803 mm 563+/-145 @2855 mm 6 (648F) 2880- 2960 43-48 41+/-3

@2900 mm ---

43+/-2 @2909 mm TBD @2900 mm ---

7+/-2 @2909 mm ---

---

1.72+/-0.01

@2906 mm 590+/-168 @2899 mm 605+/-152 @2901 mm 657+/-148 @2908 mm 8 (648H) 3320-3400 40-45 --- --- 1.62+/-0.23

@3999 mm 670+/-40 @3997 mm same axial location. The circumferential vari ation in hydride distribution and morphology is indicated. In Fig. 5, higher magnification images are shown at circumferential locations of thickest (70 µm) and thinnest (40 µm) equivalent hydride-rim thicknesses.

4 Fig. 2. Micrograph of polished met sample at 2910 mm from rod bottom (see Segment 6 in Fig. 1) showing high-magnification of the ZIRLO corrosion layer.

Fig. 3. High-magnification and high-contrast image of the fuel-cladding bond layer at 2910 mm from rod bottom (see Segment 6 in Fig. 1).

5 Fig. 4. Images at eight locations around the circum ference of the etched cl adding cross section at 2910 mm from the rod bottom showing radial and circumferen tial variation in hydride density. The cladding metal-wall thickness (544+/-2 µm), as well as the corrosion-layer (43+/-2 µm) and fuel-cladding-bond (7+/-2 µm), is relatively uniform.

6 (a) (b) Fig. 5. Local cladding regions at 2910 mm from rod bottom showing: (a) thickest (70 µm equivalent) hydride rim; and (b) thinnest (40 µm equivalent) hydride rim. The dashed red line in the images was used to estimate equivalent (i.e., dense) thickness. Variation in hydride-rim thickness correlates with variation in LECO-measured hydrogen content (500-840 wppm).

7LOCA post-quench ductility and embrittlement results are highly sensitive to hydrogen content in the metal. The upper-bound hydrogen c ontent can be calculated by assuming that all the LECO-measured hydrogen is in the metal. Me tallographic results can be used to determine the factor (f H) needed to correct for the reduced mass of the metal relative to the LECO mass.

Based on the corrosion (c), metal (h) and fuel-cladding bond (b) layer thicknesses, along with oxide density (5.63 g/cm

3) and metal density (6.54 g/cm 3), correction factors have been determined for corrosion layer thickness values in the range of 35-45 µm. A linear approximation for the correction-factor variation with corrosi on layer thickness (c in µm) is:

f H = 1.069 + 1.7 10-3 (c - 35) (1)

Table 2 shows the results for the upper-bound hydrogen content in the cladding metal based on the correction factor determined from Eq. 1. Also shown are metal hydrogen contents based on LECO-measured oxygen at three locations. Both approaches lead to similar results.

Table 2 Calculated Values fo r Hydrogen Content in Cladding Metal for North Anna ZIRLO Rod AM2-L17 Segments; calculations assume all measured H is in cladding metal f H Hydrogen Content in Metal, wppm Sample ID # W (ANL) Location from Bottom mm Based on EC* c Based on Met c LECO Hydrogen Content wppm Using Layer Thicknesses Using LECO Oxygen 5 (648E) 2803 2855 1.074 1.078 --- --- 464+/-71 563+/-145 498+/-76 607+/-156 482+/-74 --- 6 (648F) 2899 2901 2908 --- --- --- 1.079 1.079 1.083 590+/-168 605+/-152 657+/-148 637+/-180 653+/-163 712+/-160 --- --- 704+/-159 8 (648H) 3997 1.077 --- 670+/-40 722+/-43 712+/-42 Hydrogen pickup fraction (HPF) is a metric commonly used to estimate cladding hydrogen content from corrosion layer thic kness. HPF represents the ra tio of cladding hydrogen pickup and hydrogen generated in the formation of the corrosion layer. The corrosion layer thickness can be estimated from eddy-current measurement performed either poolside (most common) or in a hot cell. Metallographic re sults and/or LECO-oxygen results can be used to more precisely determine the hydrogen pickup fraction. The ANL method for performing this calculation is similar to the method used to determ ine the hydrogen co rrection factor (f H): cylindrical geometry; layer thicknesses and densities used to determine masses per unit length for corrosion layer, metal, and bond layer; and subtraction of as-fabricated concentrations of hydrogen and oxygen.

8However, it was found that a simple layer a pproach with no correction factors gave the same HPF results (within +/-0.5% absolute) as the detailed method within the relevant range of parameters. The correlation derived from the layer approach (based on mass per unit area) is given in Eq. 2.

HPF = (2% µm/wppm) L H/c , (2) where L H is the LECO-measured hydrogen content in wppm and c is the corrosion layer thickness in µm. Results from detailed ANL calcu lations are presented in Table 3 for 5 axial locations at which LECO-hydrogen was measured. For HPF va lues of 24%, 27%, 29%, 30%

and 33%, Eq. 2 predicts 24%, 27%, 28%, 31% and 34%. Given the uncertainty in the density of the corrosion layer (5.01 g/cm 3 for 90%-dense layer, 5.63 g/cm 3 for the Pilling-Bedworth ratio of 1.56 and 5.8 g/cm 3 for 100%-dense stoichiometric monoc linic oxide), Eq. 2 gives a reasonable approximation for hydrogen pickup fracti on for corrosion-layer thicknesses 50 µm. It has not been tested at higher corrosion-layer thicknesses for which the dens ity of the layer decreases and curvature effects may become more significant.

Table 3 Calculated Hydrogen Pi ckup Fraction vs. Axial Location for North Anna ZIRLO Rod AM2-L17 Samples; fuel-cladding bond thickness assumed to be 7 µm and cladding metal thickness calculated from h = 544 µm + (43 µm - c)/1.56 c, µm Sample ID # W (ANL) Location from Bottom mm Based on EC Based on Met LECO Hydrogen Content wppm Hydrogen Pickup Fraction % 5 (648E) 2803 2855 38 41 --- --- 464+/-71 563+/-145 24 27 6 (648F) 2900 2908 --- 41 43 600+/-150 657+/-148 29 30 8 (648H) 3997 40 --- 670+/-40 33

2. LOCA Apparatus Thermal and Me tallurgical Benchmark Tests Thermal benchmark tests were conducted with two thermocouples welded onto a fresh (as-fabricated) ZIRLO sample, in addition to the three TCs permanently welded to the Inconel holder just above the sample. Te sts were repeated with fresh samples until the furnace control parameters were established to give the desired temperature ramp and hold temperature. The output from one of the holder TCs (control TC) is used to control furnace power during the temperature ramp and hold periods. For te sts to be conducted using corroded high-burnup cladding, the thermal history for a pre-oxidized sample was determined. Pre-oxidation was 9conducted at 1000ºC to grow inner- and outer-surface oxide layers that would result in about the same reduced heat of oxidation during the heating ramp to 1200ºC as would be characteristic of high-burnup cladding with an outer-surface corrosion layer and an inner-surface oxide bond layer. Following thermal benchmarking, additio nal oxidation tests were conducted with no TCs welded to the samples. For these tests, weight gain and surface oxide layer thicknesses were measured and compared to Cathcart-Pawel (CP) predicted values. The ANL criterion for a successful metallurgical benchmark test is that the measured wei ght gain and average oxide layer thickness be within 10% of the CP-predicted values. Agreement between measured inner-surface and outer-surface oxide layers is also used to validate adequate steam flow to the cladding inner surface. A small (e.g. < +/-3 µm) variation in measured oxide layer thickness is used to validate temperature uniformity.

Thermal and/or metallurgical benchmark tests were conducted in three laboratory locations:

cold-lab EL-208, just out-side IM L Cell#4 and inside IML Cell #4.

Table 4 summarizes the test locations, test conditions and results of the seven benchmar k tests conducted. Prior to conducting these tests, the old radiant-heating furn ace, which has been used for over 8 years, was replaced with a new furnace. The old furnace was highly sensitive to relocation because of the design of the filaments within the vertical heating bulbs. The new furnace is more robust in design and is relatively insensitive to movement from lab-to-lab.

Tests ZLU#93A and #93B were conducted in sequen ce. Two TCs were welded to the bare ZIRLO sample, which was heated to 985ºC for a long enough hold time to grow 14-µm oxide layers on the cladding inner and outer surfaces. The sample was cooled to 300ºC and reheated to 1200ºC to determine the effects of the pre-oxide layers on the heating ramp and hold temperature.

There was an overshoot of 9ºC in the hold temperature. Post-test calibration of the sample TCs to the NIST-calibrated TC at ANL resulted in an additional temperature corre ction of 3ºC. As a result, the furnace control parameters were modified to reduce the control TC temperature by 10ºC and the sample temperature by 12ºC. The modified thermal history for the pre-oxidized ZIRLO sample is shown in Fig. 6.

The LOCA apparatus was moved into the IML, just outside IML Ce ll#4, and electrical, hydraulic and gas lines were re-connected. Three benchmark tests were conducted. ZLU#94 was run with a bare ZIRLO sample oxidized at 985ºC to grow CP-predict ed oxide layers of 14.5 µm. The measured weight gain was within 3%

of the CP-predicted weight gain and the measured outer-surface oxide layer (14+/-1 µm) was also in excellent agreement with the CP-predicted value. An additional weight-gain benchmark (ZLU#97) was conducted at the target hold temperature of 1200ºC. The measured wei ght gain was within 8% of the CP-predicted value. Test ZLU#98 was a thermal benchmark te st conducted with bare ZIRLO (see Fig. 6).

10Table 4 Data Summary for Out-of cell and In-cell Benchmark Tests Conducted with the New LOCA Furnace in Laboratory EL-208 (Out-of-cell), in the Irradiated Materials Labor atory (IML, Out-of cell) and in IML Cell #4 Parameter ZLU#93A (Bare) Test ZLU#93B(Pre-oxidized)

ZLU#94 (Bare) ZLU#97 (Bare) ZLU#98 (Bare) ZLU#100 (Bare) ZLU#101 (Bare) Target Temperature, °C 1000 1200 1000 1200 1200 1200 1200 Hold Temperature, °C 985 5 1209 2 985 a 1197 b 1197 2 b 1200 10 b 1200 10 b Test Time, s (from 300°C) 190 160 200 380 220 100 220 CP Wg, mg/cm 2 2.44 --- 2.45 11.6 8.41 4.57 8.41 Measured Wg, mg/cm 2 --- 7.85 2.52 12.5 9.10 4.82 9.08 CP-Oxide Layer, µm 14 --- 14.5 58.0 42.3 --- 42 Measured OD-Oxide Layer, µm

--- 40 2 m 13.5 0.9 --- --- ---

47 1 Measured ID-Oxide Layer, µm

--- 39 1 m --- --- --- ---

45 1 Lab Location EL-208 Out-of-cell EL-208 Out-of-cell IML Out-of-cell IML Out-of-cell IML Out-of-cell IML Cell #4 IML Cell #4 Comments 2 TCs weldedon sample 2 TCs welded on sample Weight GainBenchmark Weight GainBenchmark 2 TCs welded on sample Weight GainBenchmark Weight GainBenchmark aFurnace control parameters established for ZLU#93A used for this test bFurnace control parameters for ZLU#93B modified to reduce steady temperature of holder TC by 10°C 11 300 400 500 600 700 800 900 1000 1100 1200 1300500550600650700750800Time (s)Temperature (°C)Bare ZIRLOPre-oxidized ZIRLO Fig. 6. Comparison of thermal benchmark temperature histories for pre-oxidized (14-µm OD and ID) ZIRLO (Test ZLU#93B) and bare ZIRLO (Test ZLU#98). ZLU#93B temperatures have be en decreased by 12ºC based on the decrease in holder temperature from the #93B test to the

1. 98 test and the TC reading correction based on comparison to NIST-calibrated TC.

12The measured corrosion and bond layer thicknesses for high-burnup ZIRLO are 41-43 µm and 7 µm. Equivalent inner- and outer-surface pre-oxide layers ZIRLO are 12 µm. However, because of the low peak temperature (1000ºC) at the end of the rapid ramp and the slow heating from 1000ºC to 1200ºC, these differences are no t significant: the lower self-heating in pre-oxidized samples does not decrease the ramp temperatures from those recorded for bare ZIRLO.

It is clear from the thermal benchmark results that the performance of the new LOCA furnace is much less sensitive to physical movement than the old LOCA furnace. Also, it appears that the circumferential temperature variation is less for the new furnace as compared to the old furnace. However, achieving such circumferentially uniform temperatures requires some adjustments and rather precise positioning of the sample within the furnace. In the first thermal benchmark test (ZLU#96) conducted in the IML outside Cell#4, the temperat ure variation in the circumferential direction was 18ºC (i.e., +/-9ºC deviation from the average of the two TC readings).

Although it was measured for an initially bare ZIRLO sample , Fig. 6 suggests that the temperature history for ZLU#98 is adequate for planning test times and interpreting data for high-burnup ZIRLO cladding samples. Given the temperature uncertainty associated with positioning the sample in the furnace, the test temperature is better represented by 1200+/-10ºC than the benchmark value of 1197+/-2ºC. For in-cell tests, the technician mu st suit up in personal protective equipment (PPE) prior to entering the cell to position the quartz tube (containing the test train) within the furnace. The in-cell time will be limited with a high-burnup sample in the test train. Positioning within the furnace may not be as precise for in-cell tests as compared to out-of-cell tests.

After the LOCA apparatus was installed in IML Cell#4, two thermal benchmarks were conducted: ZLU#100 with a heating-phase test time of 100 s and ZLU#101 with a heating-phase test time of 220 s. The meas ured weight gain (4.82 g/cm

2) for the ZLU#100 sample was only 5% higher than the CP-predicted value. The agreement is excellent considering that most of the oxidation occurred during the ramp. The 100-s h eating-phase test corr esponds to 7.4% CP-ECR for high-burnup ZIRLO and is within the anticipa ted testing range of 3-8% CP-ECR. For the 220-s heating-phase test (ZLU#101), both the measured weight gain (9.08 g/cm

2) and the measured oxide-layer thicknesses (47+/-1 µm for OD and 45+/-1 µm for ID) were in good agreement with CP-predicted values. The agreement between ID and OD oxide layer thicknesses indicates adequate steam flow inside the sample. The small circumferential variation in oxide layer thickness indicates adequate temperature uniformity in the circumferential

direction

Average cooling rates were determined from Fig. 6 because of the possible sensitivity of post-quench ductility to cooling rate. The overall average rate from 1200ºC to the planned quench temperature of 800ºC is 13ºC/s with rates of 29.4ºC/s from 1200ºC 1000ºC, 11.2ºC/s from 1000ºC 900ºC, and 6.5ºC/s from 900ºC 800ºC.

133. Test Matrix for High-Burnup ZIRLO Cladding Test planning consists of targeting CP-ECR values and determining test times corresponding the target CP-ECR values. Test times refer to the time interval (t) between initiation of the ramp from 300ºC to the end of the heating phase. These times are determined by integration of the weight-gain rate equation over th e relevant part of the thermal history shown in Fig. 6 and by the relationship between CP-ECR (in %) and weight gain (Wg in mg/cm 2). The cladding wall thickness was measured to be 544 µm at a location for which the corrosion layer thickness was measured to be 43 µm. Based on uncertainties in the measurements, the cladding metal wall thickness used to calculate CP-ECR values is 540 µm (0.54 mm). This gives the following relationship:

CP-ECR = 1.623 Wg (3)

Table 5 lists test times, maximum temperatures achieved, CP-ECR values after the heating phase (CP-ECR h), and CP-ECR values after the total transient (CP-ECR t) including the cooling phase. The results are used to determine test times corresponding to target CP-ECR values.

Figure 7 shows the variation of CP-ECR t with test time. The near-linear variation is due to two competing phenomena: decrease in weight gain rate with increasing oxide layer thickness; and increase in weight gain rate with increasing temperature.

Three LOCA test samples, each 25-mm-long, were sectioned from Segment 8 in Fig. 1:

648H1, 648H2 and 648H3. The hydrogen content in these samples is expected to be relatively uniform from sample-to-sample and in the circumferential direction of each sample. Based on the results in Table 2, th e anticipated hydrogen content for these samples is 700 wppm. As shown in Table 6, these first three samples will be cooled without quench to establish the upper-bound on ductility and ductile-to-brittle transition CP-ECR. For 700-wppm-H prehydrided Zry-4 oxidized at 1200ºC and cooled with out quench, the ductile-to-brittl e CP-ECR is in the range of 6-7%. For high-burnup Zry-4 with a LECO-measured hydrogen content of 550+/-100 wppm and an estimated 580+/-100 wppm of hydrogen in the meta l, the ductile-to-brittle transition CP-ECR was 8%. As a first approximation, it is assumed that hydrogen has the same embrittling effects on ZIRLO as it does on Zry-4. Thus, the coolin g-without-quench tests are planned for CP-ECR values in the range of 4-8%.

Test 4 will be conducted with quench at 800ºC. The target CP-ECR value will depend on the post-oxidation ductility of the first three test samples. In order to have some assurance that the post-quench sample will have ductility at 135ºC, it is desirable to target the CP-ECR such that the ductility of the cooling-without-quench sample is 10%. The well-characterized Segment 6 sample (648F2), with 620+/-140 wppm hydrogen, will be used for Test 4. The average hydrogen content is lower, but th e circumferential variation is much larger than the hydrogen content of the first three samples. Based on test results for prehydrided Zry-4, a sample with 610+/-100 wppm hydrogen was brittle following 4.5% heating phase CP-ECR (4.7% total transient CP-ECR) and quench at 800ºC. Therefore, it is anticipated that this sample will be oxidized to 4% CP-ECR t to determine if there is any post-quen ch ductility. It is likely that a 5 th test will be needed to better determine ductile-to-brittle CP-ECR

t. If the embrittlement threshold is <4% CP-ECR, then the peak oxidation temperature will be <1130ºC.

14Table 5 Test Times to Achieve Target CP-ECR Values for High-Burnup ZIRLO Cladding Samples; test time is interval between initiation of ramp from 300ºC to end of heating phase; oxidation is two-sided; Tmax is maximum temperature achieved during heating phase; CP-ECR h is calculated for heating phase; CP-ECR t is calculated for total transient including cooling phase; and CP-ECR = 1.623 Wg for 0.54-mm-thick cladding wall, where CP-ECR is in % and weight gain Wg is in mg/cm 2 Test Time s Tmax ºC CP-ECR h % CP-ECR t % 15 1004 0.8 1.3 25 1037 1.6 2.0 35 1064 2.2 2.6 45 1098 2.9 3.2 55 1132 3.6 4.0 60 1142 4.0 4.4 70 1161 4.8 5.1 75 1166 5.2 5.6 85 1176 6.0 6.3 95 1184 6.8 7.1 100 1188 7.1 7.4 110 1191 7.9 8.1 0 2 4 6 8 10020406080100120Test Time (s)CP-ECR t (%) Fig. 7. Total transient CP-ECR vs. test time for high-burnup ZIRLO with 0.54-mm wall thickness. The test time is the interval from ramp initiation to the end of the heating phase.

15Table 6 LOCA Embrittlement Test Matrix for High-Burnup North Anna ZIRLO Cladding Test ID Sample ID H Content wppm Test Time s Tmax ºC CP-ECR h % CP-ECR t % Quench at 800ºC ZLI#1 648H3 670+/-40 110 1191 7.9 8.1 No Should be brittle ZLI#2 648H2 670+/-40 55 1132 3.6 4.0 No Should be ductile ZLI#3 648H1 670+/-40 85 1182 6.0 6.3 No Should be near transition ZLI#4 648F2 620+/-140 55 1104 3.6 4.0 Yes ECR depends on previous results ZLI#5 648F6 620+/-140 TBD TBD TBD TBD Yes ECR depends on previous results