ML14072A084
| ML14072A084 | |
| Person / Time | |
|---|---|
| Site: | South Texas  |
| Issue date: | 02/18/2014 |
| From: | Leavitt J South Texas |
| To: | Office of Nuclear Reactor Regulation |
| Shared Package | |
| ML14072A075 | List: |
| References | |
| NOC-AE-14003075 CHLE-012, Rev 4 | |
| Download: ML14072A084 (31) | |
Text
NOC-AE-14003075 CHLE-01 2:
Ti MBLOCA Test Report, Revision 4
PROJECT DOCUMENTATION COVER PAGE Document No: CHLE-012 Revision: 4 Page 1 of 30
Title:
T1 MBLOCA Test Report Project: Corrosion/Head Loss Experiment (CHLE) Program Date: 02/18/2014 Client: South Texas Project Nuclear Operating Company Summary/Purpose of Analysis or Calculation:
The Medium Break (MB) Loss of Coolant Accident (LOCA) 30-day tank test is one experiment within the Chemical Head Loss Experiment (CHLE) test program created to assess the generic safety issue (GSI) 191 chemical effects at the South Texas Project Nuclear Operating Company (STP) facility. The primary objectives of the MBLOCA tank test are to determine the quantity of aluminum corrosion reflective of median STP chemistry and exposed aluminum surface area and detect the presence of chemical products that may form under the given conditions.
The suite of measurements used to assess the test objectives determined a very small release of aluminum mass into solution, the possible presence of calcium solid in solution and a very dilute presence of other particles. The final experimental aluminum corrosion mass reflective of STP conditions was calculated to be 0.90 g which was very close to the WCAP predicted mass of 0.81 g. While there is a degree of uncertainty associated with the experimentally determined corroded mass of aluminum, both the experimental and WCAP predicted corroded mass of a very small quantity is in agreement. The presence particles were confirmed by EDX analysis of the in-line membrane filters and deposits on the fiber beds. These products were present in very dilute concentration as confirmed by low turbidity measurements, SEM images of the inline membrane filters and fiber bed deposits. While the debris bed were only "indicator" beds, the relatively stable head loss measurements and the presence of filtered particulate captured by the bed may indicate that chemical products generated under these conditions would not significantly impact head loss.
Role:
Name:
Signature:
Date:
Prepared:
Janet Leavitt
<signed electronically>
10/15/2012 Reviewed:
Kerry Howe
<signed electronically>
1/8/2013 Oversight:
Zahra Mohaghegh
<signed electronically>
2/12/2014 Approved:
Ernie Kee
<signed electronically>
2/18/2014 Revision Date Description 0
10/15/2012 Draft document for internal review 1
11/27/2012 Addresses internal review comments 2
12/19/2012 Add statistical analysis to soluble metal concentration section Addressed internal review comments; revised the predictions for the release 01/9/2013 of aluminum, calcium, and silicon from fiberglass to be consistent with the quantity of fiberglass used in the tests.
4 02/18/2014 Review for NRC submittal
Title:
T1 MBLOCA Test Report Contents List of Figures........................................................................................................................................
3 List of Tables.........................................................................................................................................
4 1.0 Introduction.........................................................................................................................................
5 2.0 Sum m ary of Results.............................................................................................................................
5 3.0 Continuous M easurem ents............................................................................................................
6 3.1 Head Loss M easurements through the Fiberglass Debris Beds.................................................
6 3.2 Approach Velocity through Fiberglass Debris Beds.................................................................
11 3.3 Tem perature..................................................................................................................................
11 4.0 Discrete M easurem ents....................................................................................................................
15 4.1 Solution pH....................................................................................................................................
15 4.2 Solution Turbidity..........................................................................................................................
16 4.3 Soluble M etal Concentrations...................................................................................................
17 5.0 Post Test Sam ple Analysis............................................................................................................
21 5.1 Corrosion.......................................................................................................................................
21 5.2 Fiber bed........................................................................................................................................
26 5.3 In-line mem brane filters and zeta potential.............................................................................
28 6.0 Conclusion.........................................................................................................................................
29 7.0 References.........................................................................................................................................
30 Document No: CHLE-012, Rev 4 Page 2 of 30
Title:
T1 MBLOCA Test Report List of Figures Figure 1: 18 gram NEl-processed (A) and blender-processed (b) debris beds used in this test.............. 6 Figure 2: Temperature-corrected head loss through NEI-processed fiberglass debris bed.................... 7 Figure 3: Temperature corrected head loss measurements produced by the blender-processed beds..... 8 Figure 4: Post-test blender-processed beds. A: Post-test (TO) blender processed bed from column 3 lying face-down. B: Post-test (Ti) blender processed bed from column 3 laying face-up.............................
9 Figure 5: SEM images of dense fiber nodule from blender-processed bed (50 ýtm bar) (A) and a general section of the NEI-processed bed (200.tm bar) (B)................................................................................
10 Figure 6: SEM images (200 pm bar) of soaked NEI-processed fiber in borated-buffered solution, not rinsed (left), borated-buffered solution, rinsed (middle), and DI water (right)....................................
11 Figure 7: Superficial filtration velocity through fiberglass debris beds used in this test.......................
11 Figure 8: Comparison of simulated and experimental temperature profile. Experimental temperature is m easured in the center of the corrosion tank.......................................................................................
12 Figure 9: Time periods used for the WCAP calculations for release predictions as compared the entire sim u late d p ro file o ve r.................................................................................................................................
13 Figure 10: Comparison of simulated and experimental temperature profile. Experimental temperature is m easured in the center of the corrosion tank.......................................................................................
13 Figure 11: Difference between column temperature and tank (pool) temperature over the test........ 14 Figure 12: Difference between pool and vapor space temperature over the test.................................
15 Figure 13: T1 solution pH m easurem ents..............................................................................................
16 Figure 14: T1 solution turbidity m easurem ents....................................................................................
17 Figure 15: T1 measured and predicted aluminum concentration overtime.........................................
18 Figure 16: T1 measured and predicted calcium concentration overtime............................................
19 Figure 17: T1 measured and predicted silicon concentration over time...............................................
20 Figure 18: STP alum inum scaffolding used in this test..........................................................................
22 Figure 19: XPS results for Al 2 p spectra of aluminum sample...............................................................
22 Figure 20: Two scale layers detected by XPS analysis using Al 2p spectra............................................
23 Figure 21: Pre-test SEM images of "submerged" (A) and "vapor space" (B) aluminum samples......
23 Figure 22: Post-test SEM images of "submerged" (A) and "vapor space" (B) aluminum samples........ 24 Figure 23: Bahn et al. [9] aluminum hydroxide precipitation map in the 'pH+p[AI]t' vs. temperature domain. The red line shows the CHLE MBLOCA test results (T1)........................................................
26 Figure 24: Fiber beds post-test. Series A are NEI-processed beds and Series B are blender-processed b e d s.............................................................................................................................................................
2 7 Figure 25: Particulate NEI-processed (A) and blender-processed (b) fiber beds with particulate......
27 Figure 26: SEM images of example white (A), black (B), grey (C) particles captured on the NEl-processed d e b ris b e d...................................................................................................................................................
2 8 Figure 27: SEM images of days 0, 13, and 26 in-line membrane filters with similar disperse particulate ca p tu re........................................................................................................................................................
2 8 Document No: CHLE-012, Rev 4 Page 3 of 30 Document No: CHLE-01 2, Rev 4 Page 3 of 30
Title:
T1 MBLOCA Test Report List of Tables Table 1: WCAP calculation of STP material release over the first 80 minutes of testing as a function of experim ental vs. sim ulated tem perature profile..................................................................................
12 Table 2: T-test results for total and filtered soluble aluminum sample results.....................................
18 Table 3: T-test results for total and filtered soluble aluminum sample results with outliers eliminated.. 19 Table 4: T-test results for total and filtered soluble calcium sample results..........................................
20 Table 5: T-test results for total and filtered soluble silicon sample results..........................................
20 Table 6: W eight change of alum inum sam ple......................................................................................
24 Table 7: Scale com position as predicted by XPS....................................................................................
24 Table 8: Change in distribution of scales between post-and pre-test sample. Case 1 was calculated assuming aluminum phosphate-aluminum hydroxide scale. Case 2 was calculated assuming an alum inum phosphate-alum inum oxide scale.........................................................................................
25 Table 9: EDX results associated with Figure 25 A and B SEM particle images........................................
28 Document No: CHLE-012, Rev 4 Page 4 of 30 Document No: CHLE-01 2, Rev 4 Page 4 of 30
Title:
T1 MBLOCA Test Report 1.0 Introduction The Medium Break (MB) Loss of Coolant Accident (LOCA) 30-day tank test is one experiment within the Chemical Head Loss Experiment (CHLE) test program created to assess the generic safety issue (GSI) 191 chemical effects at the South Texas Project Nuclear Operating Company (STP) facility. The primary objectives of the MBLOCA tank test are to determine the quantity of aluminum corrosion reflective of median STP chemistry and exposed aluminum surface area and detect the presence of chemical products that may form under the given conditions.
This 30-day tank test was conducted from August 22, 2012 to September 25, 2012 with the following characteristics [1]:
- 1.
Temperature profile of a 6" cold leg MBLOCA predicted by MELCOR and Relap-5.
- 2.
Approximately 1.5 times the fiberglass volume predicted by CASA for a 6" break to provide fiber beds equivalent to past testing [2].
- 3.
STP aluminum scaffolding as the source of aluminum corrosion material. The submerged aluminum surface area was scaled to maintain the STP fiberglass ratio of a MBLOCA while the total aluminum surface area ratio was conserved [2].
- 4. Material exposure to baseline chemicals of boric acid and lithium hydroxide from time zero with incremental additions of trisodium phosphate (TSP), hydrochloric acid and nitric acid.
- 5.
Column approach velocity of 0.01 ft/s.
- 6.
Two types of detector beds. The 30-day test was performed using the NEI-processed debris bed.
On day 30, columns were isolated and blender-processed beds were installed. Fresh TSP-buffered borated solution was allowed to circulate through the blender-process beds to establish baseline behavior for 2 days. The columns were then linked to the test and the test solution was allowed to circulate through the blender-processed beds for 2 days.
2.0 Summary of Results While the results of this test are detailed in the following sections, a summary of results is presented below.
- 1.
Chemical effects of a MBLOCA did not impact head loss measurements of the NEI-processed debris beds overtime.
- 2.
Blender-processed debris bed head loss measurements remained relatively similar to the 48 hour5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> base line measurements after exposure to the test solution for 2 days.
- 3.
The final aluminum concentration of 0.28 mg/L was approximately 30% of the predicted concentration determined by the WCAP-16530-NP calculations.
- 4. The final calcium concentration of 1.7 mg/L was approximately twice the concentration predicted using the WCAP-16530-NP calculations.
- 5. The final silicon concentration of 4.6 mg/L was approximately 80% of the predicted concentration determined by the WCAP-16530-NP calculations.
- 6. Turbidity measurements peaked at the beginning of the test at approximately 0.6 NTU and gradually decreased about 0.3 units until the end of the test.
Document No: CHLE-01 2, Rev 4 Page 5 of 30
Title:
T1 MBLOCA Test Report 3.0 Continuous Measurements Many parameters required to simulate the 6" MBLOCA under STP conditions were monitored continuously using a CompactRIO acquisition system and LabVIEW program. Head loss, temperature, and velocity measurements were continuously monitored and saved every minute to a spread sheet for analysis. Results and discussion associated with these continuously monitored parameters are discussed in the following sub-sections.
3.1 Head Loss Measurements through the Fiberglass Debris Beds In the early stages of the CHLE testing program, head loss measurements across debris bed were the primary measurement used to detect the presence of chemical products formed as a result of test conditions. For this test, head loss measurements across two types of indicator beds were used as a complimentary diagnostic tool from a suite of measurements that were used to assess the presence of chemical product formed as a consequence of the test scenario. This approach allowed for a sensitive test while avoiding operational issues related to the blender-processed debris beds [2] by operating in a two-step process: (1)First, the NEI-processed debris bed, which is thought to be more representative of what would occur during a LOCA, served as a nucleation site for or a filter of possible chemical products that may form during testing for 30 days, Figure 1A. (2) Then, test solution was filtered for 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> through a base-lined blender-processed debris bed which is highly sensitive but not as likely to form as a function of test conditions, Figure lB.
Figure 1: 18 gram NEI-processed (A) and blender-processed (b) debris beds used in this test.
The differential pressure measurements obtained from the CHLE tests were taken using a temperature corrected, wet/wet uni-directional pressure transducer mounted on each column connected by two ports. One port is connected by a short lead of stainless steel tubing which is partially insulated. The temperature of the solution in this short, partially insulated lead is similar to that inside the column. The other port is connected by approximately 32 inches of PTFE tubing which is positioned outside the insulated cover. The solution in the second port is in equilibrium with the room temperature and assumed to be equivalent to room temperature. This configuration results in a differential pressure measurement that must be corrected for the difference in temperatures related to the instrument mounting. The corrected differential pressure measurement was then corrected for temperature Document No: CHLE-012, Rev 4 Page 6 of 30 Document No: CHLE-01 2, Rev 4 Page 6 of 30
Title:
T1 MBLOCA Test Report related viscosity effects to produce a temperature correct head loss measurement. The temperature corrected head loss measurement equation is listed below:
HL,c = (DPraw + (Pt -Prt)
)("-')
jut
- Where, H1, is the corrected head loss DProw is the instrument differential pressure measurement pt and p, are the densities at test temperature(t) and at room temperature (rt) g is the gravitational constant h is the length of the PTFE tubing connecting the column to the DP cell AU and/,ud. are the viscosity at test temperature(t) and at standard temperature (std) of 20 'C The above correction was applied to all the raw data collected during testing. Note that the viscosity of water triples as it cools from 85 C to 20 °C so viscosity can have a significant impact on head loss, whereas the maximum discrepancy due to the density correction is about 0.75 inches, which is not significant unless the total head loss is very small.
As seen in Figure 2, the NEI-processed debris bed did not measure a significant increase in head loss during testing. During the first 30 days of testing, the head loss in column 3 was slightly higher than that in columns 1 and 2. However, all three head loss measurements remained below 0.5 inches and remained relatively unchanged over time.
2.0
-Column 1
1.5 Column 2
-Column 3 V1 1.0 o 0.5 0.0 0
5 10 15 20 25 30 Time (day)
Figure 2: Temperature-corrected head loss through NEI-processed fiberglass debris bed.
Document No: CHLE-012, Rev 4 Page 7 of 30 Document No: CHLE-01 2, Rev 4 Page 7 of 30
Title:
T1 MBLOCA Test Report After 30 days of testing, the columns were isolated and drained which resulted in a temporary spike of head loss measurements. The columns were then loaded with room temperature borated, TSP-buffered solution and blender-processed debris beds were formed under the same velocities as the NEI-processed debris beds. These columns were allowed to circulate solution at 0.01 ft/s while isolated from the corrosion tank for 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> to obtain a base-line head loss measurement for each bed.
When linking the columns to the corrosion tank, the system pressure was reduced to zero and the debris beds were monitored closely for disturbance. While there was no visual indication of bed movement, all three columns experienced a release of materials which quickly dissipated. Test solution circulated through the blender-processed debris beds for 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> after linking to monitor for the presence of chemical products.
The baseline head loss measurements of columns 1 and 3 experienced little to no change during the 2 days of isolation, Figure 3. In column 2 the base line measurements taken over the two days of isolation cycled slightly over time, but the overall trend in head loss measurements remained relatively constant.
After the columns were linked to the corrosion tank, the head loss measurements for columns 1 and 3 were relatively stable and did not increase significantly. Although the head loss measurements for column 2 were erratic, they cycled around the stable head loss measurements of column 1.
2.0
-Column 1
1.5
-Column 2
-Column 3
1.0
-41 o 0.5 0.0 30 31 32 33 34 35 lime (day)
Figure 3: Temperature corrected head loss measurements produced by the blender-processed beds.
The head loss measurements across the blender-processed bed observed during the baseline period of testing (day 30-32), was different than observations obtained in past testing (TO) with similar test conditions [2]. The blender-processed beds of test TO experienced a continually increasing head loss measurement at varying rates in all of the columns within 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of test initiation. The post-test blender-processed beds from this test (T1) were compared to the post-test blender-processed beds Document No: CHLE-012, Rev 4 Page 8 of 30
Title:
T1 MBLOCA Test Report obtained from TO to investigate the difference in head loss behavior observed under similar test conditions.
During the post-test examination of the Ti blender-processed beds, pronounced dense nodules in the pattern of the support screen were noted as shown in Figure 4A. This observation is similar to the observation taken during the post-test examination of the TO blender-processed beds with one distinct difference; The nodules observed on the TO blender-processed beds were brown in color, Figure 4B. It is assumed that the color difference between the two post-test beds nodules is silicon carbide which was trapped from previous single column preliminary tests and released during this longer term test.
During single column preliminary tests with silicon carbide [3], column 3 was used much more frequently during testing than column 2, and column 2 was used more than column 1. Although the test apparatus was certified cleaned per test plans [4, 5] the piping and valve components of the columns may have had trapped trace amounts of silicon carbide that was slowly released during the first long term test, TO. The slow release of trapped silicon carbide also provides explanation of the observed poor repeatability with head loss measurements of test TO.
A B
Figure 4: Post-test blender-processed beds. A: Post-test (TO) blender processed bed from column 3 lying face-down. B: Post-test (T1) blender processed bed from column 3 laying face-up.
In effort to better understand the fiber packing of the dense nodules in the blender-processed bed, a section with pronounced dense nodules was examined with SEM and compared to a general section of the NEI-processed bed, Figure 5. At first glance, the examined section of T1 blender-processed bed appears to have a collection of captured debris while the NEI-processed examined section appears clean. To determine if this observation is truly indicative of test occurrences, the SEM sample preparation method was reviewed.
Document No: CHLE-012, Rev 4 Page 9 of 30 Document No: CHLE-01 2, Rev 4 Page 9 of 30
Title:
T1 MBLOCA Test Report Figure 5: SEM images of dense fiber nodule from blender-processed bed (50 pim bar) (A) and a general section of the NEI-processed bed (200 lim bar) (B).
All SEM fiber samples were allowed to dry in a desiccator before analysis. The evaluated section of the NEI-processed fiber debris is not as tightly packed and dried very quickly for analysis. The blender-processed bed required a significantly longer time to dry before analysis which indicates the presence of larger volumes of trapped chemical solution as compared to the examined NEI-processed fiber debris.
Therefore, it was hypothesized that the collection of debris associated with the blender bed was a product that formed as the solution evaporated and is not representative of test occurrence. To test this theory, fiber debris was prepared per the NEI-preparation method. Three samples were taken from the prepared fiber debris and soaked in either DI water or in borated, TSP-buffered solution overnight. One of the fiber samples soaked in the borated, TSP-buffered solution was taken directly out of solution and dried for analysis. The other sample soaked in borated, TSP-buffered solution was rinsed with DI water and then dried for analysis. The third sample was soaked in DI water over night and dried. The resulting SEM images, Figure 6, indicate that the un-rinsed fiber soaked in the chemical solution had a collection of debris that was similar to that on the dense nodule taken from the test blender bed, Figure 5A. The fiber that had been rinsed of chemical solution before drying still had dispersed occurrences of the debris but much less than the un-rinsed sample, while the fiber soaked in DI water was clean.
Therefore the collection of debris noted in the SEM image of the dense nodule obtained from the blender bed, Figure 5A, is likely an artifact of sample preparation and not chemical products filtered from solution.
Document No: CHLE-01 2, Rev 4 Page 10 of 30
Title:
T1 MBLOCA Test Report Figure 6: SEM images (200 pm bar) of soaked NEI-processed fiber in borated-buffered solution, not rinsed (left), borated-buffered solution, rinsed (middle), and DI water (right).
3.2 Approach Velocity through Fiberglass Debris Beds The approach velocity was maintained near 0.01 ft/s in all three columns throughout testing, Figure 7.
Approach velocity for each column was adjusted by throttling a valve on the discharge side of the centrifugal pump followed by adjustment of the variable speed drive. This approach in velocity control prevents fiber buildup for occurring within the valve resulting in minimal disturbance of the velocity during testing.
4-0.
0.04 0.03 0.02 0.01 0.00
-Column 1
-Column 2
Column 3 0
5 10 15 20 Time (day) 25 30 35 Figure 7: Superficial filtration velocity through fiberglass debris beds used in this test 3.3 Temperature The test was designed to simulate the temperature profile of a 6-inch cold leg MBLOCA event as determined by MELCOR and Relap-5 simulations [1]. The simulation predicts large variations in temperature over the initial minutes of testing, Figure 8. The temperature is then expected to oscillate slightly until about 80 minutes into the event where it begins a uniform decrease. Since this behavior cannot be experimentally achieved with the current experimental configuration, it was decided to begin Document No: CHLE-012, Rev 4 Page 11 of 30
Title:
T1 MBLOCA Test Report the test at 85'C (185 0F). At test initiation, a controlled linear temperature decrease to the point where the simulated profile reaches a uniform temperature decrease began.
90 85
\\
-Simulated Profile
- _T1 Experimental Profile so 85 S70
&65 E
I-60 55 s5 45 0
400 800 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 Time (day)
Figure 8: Comparison of simulated and experimental temperature profile. Experimental temperature is measured in the center of the corrosion tank.
While the deviation from the simulated profile prevents exposure of materials to the predicted peak temperature of 89.60 C (193.2 0F) which excludes the opportunity of higher corrosion at that temperature, the overall corrosion expected to occur experimentally is calculated to be slightly higher than the WCAP-16530-NP predicted corrosion of material exposed to the simulated temperature profile as shown in Table 1. The corrosion expected to occur during the first 80 minutes of the experimental and simulated temperature profile were predicted using the time periods and temperatures shown by Figure 9. The higher calculated release of material under experimental conditions is due to the staring temperature of 85 0C (185 0F) and the linear controlled decrease in temperature during the first few hours of testing. This approach exposes the test materials to higher temperatures over a longer time which results in slightly more corrosion than the corrosion of materials exposed to higher peak temperatures over smaller periods of time.
Table 1: WCAP calculation of STP material release over the first 80 minutes of testing as a function of experimental vs.
simulated temperature profile.
Ca Release Si Release Al Release Case (kg)
(kg)
(kg)
Experimental temperature profile 0.09 0.10 0.14 Temperature profile simulated in MELCOR 0.09 0.08 0.11 Document No: CHLE-012, Rev 4 Page 12 of 30 Document No: CHLE-01 2, Rev 4 Page 12 of 30
Title:
T1 MBLOCA Test Report 90 85 80
-75 70 65 160 El t-55 45 Il Simulated profile points used in WCAP calculation
-Simulated Profile I Experimental profile points used in WCAP calculation 40 0
600 1200 1800 2400 Time (sec) 3000 3600 4200 4800 Figure 9: Time periods used for the WCAP calculations for release predictions as compared the entire simulated profile over.
After the first 80 minutes, the experimental temperature profile closely traced the simulated profile, Figure 10. During day 28 to 34, the experimental temperature oscillates around the simulated profile temperature as a result of non-optimized operation of heater controller.
85 75 Simulated Profile U 65 T1 Experimental Profile 55 E
, 45 35 25 0
5 10 15 20 25 30 35 lime (day)
Figure 10: Comparison of simulated and experimental temperature profile. Experimental temperature is measured in the center of the corrosion tank.
Document No: CHLE-012, Rev 4 Page 13 of 30 Document No: CHLE-01 2, Rev 4 Page 13 of 30
Title:
T1 MBLOCA Test Report As indicated by Figure 11, the pool temperature appears to be slightly higher than the column temperature. After the columns were linked to the tank, there is less than a 0.50C temperature differential between the two measurements. Given that the calibration of the column thermocouples resulted in an accuracy of +/- 0.7 oC and the tank thermocouple calibrated to an accuracy of +/-0.20C [6],
the temperature difference between the columns and the pool solution are not greater than the inaccuracy associated with the measurements. Therefore, the temperatures difference between the columns and the pool are within the noise of the instrument and may not exist.
2.0 1.5
-Pool
- Column 1
-Pool
- Column 2
-Pool
- Column 3 E
A-1 1.0 0.5 0.0 0
5 10 15 20 25 30 35 Time (day)
Figure 11: Difference between column temperature and tank (pool) temperature over the test.
A difference between the tank solution temperature and tank vapor temperature of approximately 3°C is shown in Figure 12. The cycling associated with this difference (~0.5 0C) is attributed to the change in room temperature from day to night since the cycling period is approximately 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. The large spikes in temperature differences at day 30 and 34 are due to the removal of the tank lid for experimental needs.
Document No; CHLE-012, Rev 4 Page 14 of 30 Document No: CHLE-01 2, Rev 4 Page 14 of 30
Title:
T1 MBLOCA Test Report 10 9
8 Pool-Vapor space 1
6 E
0- 4 0
5 10 15 20 25 30 35 lime (day)
Figure 12: Difference between pool and vapor space temperature over the test.
4.0 Discrete Measurements While many parameters were continually measured, some parameters such as soluble metal concentration, pH and turbidity were measured on discrete time frames. Samples for soluble metal concentration measurements were taken daily for the first ten days, followed by three times a week until the end of testing. The test solution pH was monitored both continuously with an in-line pH meter and at discrete times with a bench top pH probe to monitor for in-line pH meter drift. The discrete pH measurements were performed on the same sampling schedule as the soluble metal concentration measurements. Tank turbidity measurements were taken daily while turbidity measurements at the inlet and outlet of the heat exchanger were taken on the same sampling schedule as the soluble metal concentration measurements. Results and discussion associated with these parameters are discussed in the following sub-sections.
4.1 Solution pH The solution pH was measured by both in-line and bench top automatic temperature correction (ATC:)
pH meters. The in-line pH meter only calibrates with a two-point curve and was calibrated using the pH 7 and the pH 10 standards resulting in unreliable measurements below pH 7 for this test; Therefore the starting solution pH was obtained using the bench top meter. Above pH 7, the bench top and in-line pH results were within 0.1 pH units of each other. As determined by the bench top pH meter, the solution pH at test initiation was 4.5 and increased to 7.2 during the addition of TSP. The solution pH remained at 7.2 + 0.1 for the duration of the test, Figure 13. Post-test calibration of the in-line pH meter showed a 0.09 unit drift up.
Document No: CHLE-012, Rev 4 Page 15 of 30
Title:
T1 MBLOCA Test Report 10 9
T1 Experimental Profile Bench Top pH It 6
5 4
0 5
10 15 20 25 30 35 Time (day)
Figure 13: Ti solution pH measurements.
4.2 Solution Turbidity Turbidity measurements of the bulk test solution and solution from two locations in the heat exchanger loop (upstream and downstream of the heat exchanger) were collected. All turbidity measurements throughout the test were below the turbidity level that correlated well with 1 mg/L aluminum precipitate in previous testing with similar solution chemistry [2]. The turbidity was the highest on the first day of testing, -0.6 NTU, and gradually decreased over the thirty days of testing to -0.3 NTU, Figure
- 14. An increase in turbidity measurement occurred when the blender beds were linked to the tank and was likely due to loosely attached fiber and binder released from the new blender beds. During the thirty days of testing, the solution turbidity when cooled by the heat exchanger remained relatively similar to the test-temperature solution turbidity, Figure 14.
Document No: CHLE-012, Rev 4 Page 16 of 30
Title:
T1 MBLOCA Test Report 1.0 A Tank 0.8 Pre-Heat Exchanger e Post-Heat Exchanger 50.6 A 6
A A
02 AA 0.4 A
s A '
0.0 000 5
10 15(Ds 20 25 30 3
Figure 14: Ti solution turbidity measurements 4.3 Soluble Metal Concentrations Solution concentrations of aluminum, calcium, and silicon were measured daily. For both filtered and unfiltered samples, all three analytes were at detectable levels within the first day of testing and remained relatively constant throughout the testing. The measured concentrations of these analytes collected during testing were compared to the predicted concentrations over time as calculated using the WCAP-16530-NP equations [7]. Also, the filtered and total measured concentrations for the individual analytes were subjected to statistical analysis using a t-test for both a 1-tail and 2-tail test with a < 0.05 p-value to determine whether the measurements were statistically equivalent.
The final measured aluminum concentration was approximately a third of the predicted concentration and reached a steady state concentration immediately, Figure 15. Although the filtered and total aluminum concentrations appeared equivalent (Figure 15), they were statistically evaluated to determine whether a difference existed between the two types of measured concentrations. The resulting t-statistic value is greater than the t-critical value which determines that the measured total and filter sample concentrations are different, Table 2. Upon closer examination of the raw data, Figure 15, samples taken on day 7, 28, 30, and 33 had the largest difference in measured quantities for the total and filtered samples. When the day 28, 30 and 33 samples are removed from the analysis, the results (t-statistic < t critical) determine that the total and filtered sample concentrations are equivalent, Table 3. It is possible that a product formed in solution around day 28 accounting for the difference between the two types of measurements. However, given that the measured concentrations were low and a very small difference in the mean concentration (0.003 mg/L) changes the statistical result, it is likely that the two measurements, filtered and non-filtered, are equivalent.
Document No: CHLE-012, Rev 4 Page 17 of 30
Title:
Ti MVBLOCA Test Report 1.0 0.8
-0.6 0.0 Filtered Aluminum 0
Total Aluminum WCAP Prediction "a
q if 0
10 15 20 25 30 35 Time (day)
Figure 15: T1 measured and predicted aluminum concentration over time Table 2: T-test results for total and filtered soluble aluminum sample results Al total (mg/L)
Al Filtered (mg/L)
Mean Variance Observations Pearson Correlation Hypothesized Mean Difference df t Statistic P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail 0.2800 0.0002 20 0.8080 0
19 2.4629 0.0118 1.7291 0.0235 2.0930 0.2745 0.0003 20 Document No: CHLE-012, Rev 4 Page 18 of 30 Document No: CHLE-01 2, Rev 4 Page 18 of 30
Title:
T1 MBLOCA Test Report Table 3: T-test results for total and filtered soluble aluminum sample results with outliers eliminated Al total (mg/L)
Al Filtered (mg/L)
Mean 0.2806 0.2776 Variance 0.0003 0.0003 Observations 17 17 Pearson Correlation 0.8630 Hypothesized Mean Difference 0
df 16 t Statistic 1.4286 P(T<=t) one-tail 0.0862 t Critical one-tail 1.7459 P(T<=t) two-tail 0.1724 t Critical two-tail 2.1199 The measured calcium concentration was approximately twice the predicted concentration and reaches a steady state concentration within the first days of testing as predicted, Figure 16. Upon examination of the calcium total and filtered concentration results, the filtered results are consistently lower than the total results. Statistical analysis of the data, Table 4, produced a t-statistic value that is greater than the t-critical value; therefore the total and filtered concentrations are statistically different. This result supports the possibility that calcium solids, >0.45 lim, were present in the test solution.
3.0
Filtered Caldum 2.0 1.0 0.5 0.0 0
5 10 15 20 25 30 35 Time (day)
Figure 16: T1 measured and predicted calcium concentration over time Document No: CHLE-012, Rev 4 Page 19 of 30
Title:
T1 MBLOCA Test Report Table 4: T-test results for total and filtered soluble calcium sample results Ca total (mg/L)
Ca Filtered (mg/L)
Mean Variance Observations Pearson Correlation Hypothesized Mean Difference df t Statistic P(T<=t) one-tail t Critical one-tail P(T<=t) two-tail t Critical two-tail 1.7313 0.0463 16 0.7901 0.0000 15 2.5733 0.0106 1.7531 0.0212 2.1314 1.6438 0.0413 16 The measured silicon concentration was approximately 20% less than the predicted value, but the appearance of the analyte in solution was different than the predicted trend, Figure 17. While a few filtered and total concentrations measurements appear slightly different, statistical analysis of the overall test data (Table 5) supports the conclusion that the total and filtered results are equivalent since the t-statistic value is less than the t-critical value.
6.0 5.0 4.091 j3.0 2.0 1.0 0.0 0
- .ANNE111 o a.
a 0
E E
- Filtered Silicon Total S icon
_WCAP Prediction 5
10 15 20 25 30 35 Time (day)
Figure 17: T1 measured and predicted silicon concentration over time Table 5: T-test results for total and filtered soluble silicon sample results Document No: CHLE-012, Rev 4 Page 20 of 30 Document No: CHLE-01 2, Rev 4 Page 20 of 30
Title:
T1 MBLOCA Test Report Si total (mg/L)
Si Filtered (mg/L)
Mean 4.6300 4.6050 Variance 0.0306 0.0384 Observations 20 20 Pearson Correlation 0.8395 Hypothesized Mean Difference 0
df 19 t Statistic 1.0450 P(T<=t) one-tail 0.1546 t Critical one-tail 1.7291 P(T<=t) two-tail 0.3092 t Critical two-tail 2.0930 5.0 Post Test Sample Analysis 5.1 Corrosion When aluminum metal corrodes, the aluminum can be released into solution or it can form a scale layer on the material itself. Once in solution, the corroded aluminum can remain in solution, precipitate and be separated from solution by sedimentation or filtration, or form scale on other surfaces in the system.
The concentration of material remaining in solution is easily obtained by inductively coupled plasma optical emission spectrometry (ICP-OES) measurements. The concentration of aluminum remaining in solution after 30 days is approximately 0.28 mg/L, Figure 15. The mass of aluminum released into solution that may have formed scale on other surfaces or precipitated and then been separated from solution by sedimentation or filtration is cannot be measured, but may be estimated as a result of a total aluminum mass balance.
To determine the aluminum corrosion that formed a scale layer on the material itself requires the knowledge of the original and final scale composition. In controlled laboratory testing, it is common to use clean materials of known alloys with pre-test scales of aluminum oxide. The aluminum material used during testing was scaffolding provided by STP. The sample was cleaned with mild laboratory soap to remove particulate, allowed to dry, and cut to size for testing, Figure 18. It is a non-homogenous sample with unknown constituents from years of use which remained after cleaning and the alloy of aluminum is not known. The samples which were to be submerged were from the side of the scaffolding and had a different texture and appearance than the samples cut for the vapor space. Therefore, unused samples taken from locations similar to that shown in Figure 18 (vapor space and submerged samples) were analyzed using X-ray photoelectron spectroscopy (XPS) and scanning electron microscope (SEM) with energy dispersive X-ray spectroscopy (EDX) to evaluate the original scale composition.
Document No: CHLE-012, Rev 4 Page 21 of 30 Document No: CHLE-01 2, Rev 4 Page 21 of 30
Title:
T1 MBLOCA Test Report Figure 18: STP aluminum scaffolding used in this test XPS analysis of the pre-test samples detected the presence of multiple elements, Figure 19.
Interpretation of the analysis determined that two scale types, aluminum phosphate and aluminum oxide/aluminum hydroxide scales, were present on both the vapor space and submerged pre-test samples, Figure 20. SEM analysis of these samples resulted in images of disperse deposits for both sample types, Figure 21. EDX analysis taken of the surface during SEM are in agreement with the XPS analysis results and detected different trace elements with large composition of aluminum and oxygen or aluminum, oxygen and phosphorus.
12.
10.
0?,
Ch, 001.
7 2
P C.:V P 2P Si 2.
A4.
529 0100 2929100 3r 9100 1099 9100 10le 9100 3449100 130 *100
-l 91 00 13004100 3 0113 29M99 309-I 2 '022 0 2194 5 2232 2 B-I5 2 0959 2 5'52 3 '599 A-29629 230 51624"4 903 109 20129"D 039 M9 2223 025 202' ;41 0-2 046 1090132 1392021 AU%
I,-
103 030 036 104 133 3005
'90 000 6.
6.
4.
C'.
.LAA 4,
4,
-ASi.
3
.n z".
C 1000 800 600 400 200 0
Binding Enegy (eV)
Figure 19: XPS results for Al 2p spectra of aluminum sample Document No: CHLE-012, Rev 4 Page 22 of 30
Title:
T1 MBLOCA Test Report Binding Energy (eV)
Figure 20: Two scale layers detected by XPS analysis using Al 2p spectra Figure 21: Pre-test SEM images of "submerged" (A) and "vapor space" (B) aluminum samples.
After obtaining background information of the pre-test samples taken from similar locations, the post-test aluminum samples were analyzed by both XPS and SEM with EDX. The aluminum phosphate and aluminum oxide/aluminum hydroxide scales that were detected in the pre-test aluminum samples were also detected on the post-test samples, but in slightly different ratios. Also, SEM analysis of the post-test submerged sample, Figure 22A, resulted in a surface layer that was visually similar to the pre-test submerged sample, Figure 21A. However, SEM analysis of the post-test vapor sample, Figure 22B, Document No: CHLE-012, Rev 4 Page 23 of 30
Title:
T1 MBLOCA Test Report produced a visually different surface layer when compared to the pre-test sample, Figure 21B, which is in agreement with the weight gain of the sample, Table 6.
Figure 22: Post-test SEM images of "submerged" (A) and "vapor space" (B) aluminum samples.
Using the pre-test and post-test information for masses of the aluminum samples (Table 2), XPS results (Table 3), and removing the scale layer per ASTM standards [81, an experimental aluminum corrosion mass under STP conditions was calculated. This calculation also incorporates a theoretical scale molecular weight, Table 7. Since aluminum oxide and aluminum hydroxide have similar binding energies for aluminum using the 2p spectral line (74.9 ev vs. 74.6 ev), the total corrosion mass was calculated with two types of scale mixtures. The first was an aluminum phosphate-aluminum hydroxide scale and the second was an aluminum phosphate-aluminum oxide scale. This analysis provides a mass balance for the total aluminum corroded in the system.
Table 6: Weight change of aluminum sample Sample Pre-test (g)
Post-test (g)
Mass change (g)
Submerged #1(g) 63.7445 63.6753
-0.0692 Submerged #2 (g) 64.1142 64.0490
-0.0652 vapor space (g) 840.10 840.46
+0.36 Table 7: Scale composition as predicted by XPS Scale (%)
AIPO 4-AL(OH) 3 Scale AIPO 4-AI20 3 Scale Molecular Mol AI/mol Molecular Mol AI/mol Sample AIPO 4 A120 3/AI(OH) 3 Weight scale Weight scale Pre-Test Submerged 58 42 104 1.00 114 1.4 Post-Test Submerged 51 49 100 1.00 112 1.5 Pre-Test Vapor 83 17 115 1.00 119 1.2 Post-Test Vapor 87 13 116 1.00 119 1.1 Document No: CHLE-01 2, Rev 4 Page 24 of 30
Title:
T1 MBLOCA Test Report The mass of aluminum associated with the metallic aluminum was tabulated (Table 8) assuming the following: (1) the total mass of the aluminum sample consists of only aluminum and elements bound to the aluminum scale and (2) the scale layer on the sample is a uniform thickness and composition as determined by XPS analysis. Given that the post-test scale layer consisted of more aluminum than the pre-test scale layer, (Table 8), the scale layer on the aluminum samples increased during testing on all test samples and consisted of 0.26 g of corroded aluminum. The final calculated aluminum mass release from the total exposed surface area in the test was calculated to be 0.64 g.
Table 8: Change in distribution of scales between post-and pre-test sample. Case 1 was calculated assuming aluminum phosphate-aluminum hydroxide scale. Case 2 was calculated assuming an aluminum phosphate-aluminum oxide scale Aluminum Mass Aluminum released to converted to Pre-Test Aluminum in Sample Post-Test Aluminum in Sample solution scale AIn Al in Al in Al in scale scale Al scale scale Aluminum Al metal Case 1 Case 2 metal Case 1 Case 2 Case 1 Case 2 Case 1 Case 2 Sample (g)
(g)
(g)
(g)
(g)
(g)
(g)
(g)
(g)
(g)
Vapor 834.11 1.41 1.59 833.52 1.61 1.78 0.38 0.40 0.20 0.19 Submerged 1 63.36 0.10 0.13 63.21 0.13 0.17 0.13 0.12 0.03 0.04 Submerged 2 63.72 0.10 0.13 63.56 0.13 0.18 0.13 0.12 0.03 0.04 Total Mass 961.19 1.62 1.86 960.29 1.87 2.12 0.64 0.63 0.26 0.27 Our water analysis detected approximately 0.28 mg/L which corresponds to 0.32 g of aluminum. A mass balance using the amount measured in solution and the amount released to solution predicts that half of the aluminum mass released may have formed scale on other surfaces or precipitated and then been separated from solution by sedimentation or filtration. When the mass of total aluminum released into solution for our test, 0.64 g, is converted to "pH+p[AIIT" for comparison to previous aluminum precipitation tests under similar conditions[9], Figure 23, the corresponding value is 11.9. At this value, the solution is not saturated with aluminum unless the temperature is lower than 80°F which is below the lowest measured test temperature of 940F. Therefore, it is likely the quantity of aluminum predicted to come out of solution from our mass balance is reflective of uncertainty within the calculated value for total aluminum released into solution and may not be associated with aluminum mass coming out of solution during the experiment.
This uncertainty in the total aluminum mass measurement may be largely associated with the vapor space aluminum sample. While the weight gain or aluminum scale may be reliable, the quantity of aluminum released into solution may have a larger degree of uncertainty. This sample was very large, so only a fragment (~1/10) of it was taken through the scale removal processes. The results of this fragment were obtained on a mass of scale to surface area basis and used to estimate the final scale quantity of the larger surface area. As shown by the SEM images of Figure 21 and Figure 22, the scale layer on this vapor space sample was not uniform. Therefore applying the scale results taken from sub Document No: CHLE-01 2, Rev 4 Page 25 of 30
Title:
T1 MBLOCA Test Report section of the sample to the entire test sample likely produced uncertainty in the measurement. The scale related to the submerged sample also has some uncertainty associated with it which is likely due to the assumption associated with the scale composition as opposed to the small uncertainty associated with the measured scale mass. The submerged sample was small enough for the entire sample to be taken through the scale removal process, providing a more accurate mass measurement of the scale.
Temperature (OF) 40 80 100 120 140 160 180 200 220 125 12 115
=. 11
- 0.
105 10 9
20 30 40 50 60 70 8o 90 100 Temperaure C0C)
Figure 23: Bahn et al. [9] aluminum hydroxide precipitation map in the 'pH+p[AI]t' vs. temperature domain. The red line shows the CHLE MBLOCA test results (T1).
Even with the uncertainty associated with the value, the calculated total aluminum mass loss due to corrosion of 0.90 g is relatively close to the WCAP prediction of 0.81 g. However the experimental value is representative of corroded aluminum mass converted to scale and that released into solution; while the WCAP calculation assumes the entire corroded mass is release to solution.
5.2 Fiber bed After testing, both bed types from all columns appeared relatively clean, Figure 24. Both had a disperse collection of white and greyish/black particulates, Figure 25.
Document No: CHLE-012, Rev 4 Page 26 of 30 Document No: CHLE-01 2, Rev 4 Page 26 of 30
Title:
T1 MBLOCA Test Report B1 B2 B3 Figure 24: Fiber beds post-test. Series A are NEI-processed beds and Series B are blender-processed beds.
Figure 25: Particulate NEI-processed (A) and blender-processed (b) fiber beds with particulate.
Representative sample areas of fiber with grey, white and black particles taken from the top of the debris beds were evaluated with SEM. Some particles had unexpected constituents of titanium or fluorine and others had constituents of probable chemical products derived from the solution chemistry.
Figure 26 and Table 9 present examples of these results. Pipe joint compound, Teflon tape and high temperature plastics were used to build the experimental apparatus. It is suspected that particulates with titanium in them were derived from pipe joint compounds and particulates with fluorine in them are from the Teflon tape used within the piping or from plastic components of valves or pumps that may have eroded during use. Therefore, the visible captured particulates in the debris beds are a mixture of equipment debris (Figure 26, A and C) and possible chemical products (Figure 26, B).
Document No: CHLE-012, Rev 4 Page 27 of 30 Document No: CHLE-01 2, Rev 4 Page 27 of 30
Title:
T1 MBLOCA Test Report Figure 26: SEM images of example white (A), black (B), grey (C) particles captured on the NEI-processed debris bed.
Table 9: EDX results associated with Figure 25 A and B SEM particle images White Particle Black Particle Grey Particle Element (Atomic %)
Element (Atomic %)
Element (Atomic %)
0 74.0 0
81.0 F
90.6 Al 0.5 Na 2.2 Al 0.6 Si 0.7 Al 2.7 Si 1.0 Ca 22.1 Si 11.9 Ca 4.0 Ti 2.8 Ca 2.2 Ti 3.8 5.3 In-line membrane filters and zeta potential Test solution was run through a 0.1 jim in-line membrane filter upstream and downstream of heat exchanger to determine whether or not predicted temperature decreases caused precipitation to occur.
Filtering of upstream solution occurred first to establish a baseline of particles existing before the simulated temperature drop. Once the upstream membranes were obtained, solution was then filtered downstream of the heat exchangers to capture precipitation, if it occurred. During SEM evaluation of these in-line membrane filters, a visual difference between the upstream and downstream membranes was not noticed. This could be a result of the heat exchanger design which allowed for a majority of heat loss at high temperatures to occur in the lines before the heat exchanger or it could be due to the fact nothing precipitated as a function of temperature.
Regardless of location, the debris captured on the inline membrane filters had constituents indicative of both possible chemical products and equipment debris. While debris was captured on the membranes, the membranes used throughout the tests remained relatively clean, Figure 27, indicating that the particulates were present in dilute quantities.
Figure 27: SEM images of days 0, 13, and 26 in-line membrane filters with similar disperse particulate capture.
Document No: CHLE-01 2, Rev 4 Page 28 of 30
Title:
T1 MBLOCA Test Report Zeta potential and size distribution was another method used to evaluate particles in solution. Since turbidity measurements and aluminum concentrations supported the unlikely presence of precipitation products, zeta potential measurements were not taken until the end of testing. The results obtained from this analysis were inconclusive because the solution was too dilute to obtain an accurate measurement.
6.0 Conclusion The suite of measurements used to assess the test objectives determined a very small release of aluminum mass into solution, the possible presence of calcium solid in solution and a very dilute presence of other particles. The final experimental aluminum corrosion mass reflective of STP conditions was calculated to be 0.90 g which was very close to the WCAP predicted mass of 0.81 g.
While there is a degree of uncertainty associated with the experimentally determined corroded mass of aluminum, both the experimental and WCAP predicted corroded mass of a very small quantity is in agreement. The presence particles were confirmed by EDX analysis of the in-line membrane filters and deposits on the fiber beds. These products were present in very dilute concentration as confirmed by low turbidity measurements, SEM images of the inline membrane filters and fiber bed deposits. While the debris bed were only "indicator" beds, the relatively stable head loss measurements and the presence of filtered particulate captured by the bed may indicate that chemical products generated under these conditions would not significantly impact head loss.
Document No: CHLE-012, Rev 4 Page 29 of 30
Title:
T1 MBLOCA Test Report 7.0 References
- 1.
UNM, Test 1: Medium Break LOCA Tank Test Parameter Summary, 2012, University of New Mexico.
- 2.
U N M, CHLE Tank Test Results for Blended and NEI Fiber Beds with Aluminum Addition, 2012.
- 3.
UNM, Debris Bed Preparation and Formation Test Results, CHLE-008, 2012.
- 4.
UNM, Tank Clean Instruction, Rev 3, 2012.
- 5.
UNM, Column Instructions, Rev 3, 2012.
- 6.
UNM, Instrument Calibration, 2012, University of New Mexico.
- 7.
Lane, A.E., et al., Evaluation of Post-Accident Chemical Effects in Containment Sump Fluids to Support GSI-191, 2006, Westinghouse Electric Company: Pittsburge, PA.
- 8.
Standard, A., G1 - 03: Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens, 2011.
- 9.
Bahn, C.B., Kasza, K.E., Shack, W.J., Natesan, K and Klein, P., Evaluation of precipitates used in strainer head loss testing: Part Ill Long-term aluminum hydroxide precipitation tests in borated water. Nuclear Engineering and Design, 2011. 239: p. 1974-1925.
Document No: CHLE-012, Rev 4 Page 30 of 30 Document No: CHLE-01 2, Rev 4 Page 30 of 30