ML090400742
| ML090400742 | |
| Person / Time | |
|---|---|
| Site: | North Anna  |
| Issue date: | 02/09/2009 |
| From: | Allen Hiser NRC/NRR/DCI/CSGB |
| To: | Melanie Wong Plant Licensing Branch II |
| References | |
| GL-04-002, TAC MC4696, TAC MC4697 | |
| Download: ML090400742 (14) | |
Text
ENCLOSURE 1 North Anna Power Station Audit Report Corrective Actions for Generic Letter 2004-02 Chemical Effects 1.0 Introduction The U.S. Nuclear Regulatory Commission (NRC) staff has performed sample audits of nine licensees corrective actions for Generic Letter (GL) 2004-02, "Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Pressurized-Water Reactors," dated September 13, 2004. The purpose of these audits was to help verify that licensees have resolved the concerns in the GL. Audit candidates were selected based on a sampling basis related to reactor type, containment type, strainer vendor, NRC regional office, and sump replacement analytical contractor. North Anna Power Station (NAPS) was included in these nine audits, and the NRC staff evaluated the new sump design, associated analyses, and testing for NAPS Unit 2 in July 2007. The NAPS audit report is available in ADAMS under Accession No. ML072740400 [1]. Since licensees chemical effects evaluations were in progress during the nine earlier audits, the NRC staff was not able to reach a conclusion about the adequacy of chemical effects evaluations for the 69 U.S. operating Pressurized Water Reactors (PWRs). Therefore, the staff determined that it would be appropriate to perform additional limited scope audits focusing on chemical effects.
In general, these chemical effects audits will consider the chemical effects evaluation guidance document process flow sheet (see Figure 1, ADAMS Accession Number ML080380214) as a useful guide for the audit scope. The NRC staff is interested in the licensees overall strategy for evaluation and accommodation of chemical effects, including why the licensee thinks chemical effects have been addressed in a representative or conservative manner. Specific topics that are of interest to the staff include:
o Plant-specific debris mix (non-chemical) o Plant-specific debris bed formation (non-chemical) o Plant-specific sump fluid conditions (pH, buffer chemicals, temperature profile) o Method used to calculate the plant-specific chemical precipitate load o Supplemental testing (e.g., bench top tests) used as part of the chemical effects evaluation.
o Any assumptions used to reduce the predicted plant-specific precipitate load o Integrated (with chemical effects) head loss test protocol and any open generic issues related to the vendors test protocol o Precipitate generation method for integrated head loss testing o Settlement of chemical debris during head loss testing o Integrated head loss test plot(s) o Test termination and head loss extrapolation, if applicable o Data analysis NAPS was selected as one of the plants for a chemical effects audit since it is a representative plant for the chemical effects evaluation approach performed by Atomic Energy of Canada Limited (AECL). The NRC staff and an NRC contractor visited Dominions Innsbrook facility from November 12-14, 2008, to perform the chemical effects audit. Prior to the on-site portion of the audit, the staff had the benefit of reviewing relevant documents related to chemical effects bench testing and integrated head loss test results for NAPS.
The NRC staff and an NRC contractor also had the benefit of a previous visit to AECLs Chalk River facility during May 5-9, 2008, to observe integrated chemical effects head loss testing for the Dominion plants, including NAPS. A trip report summarizing observations from the staffs visit to Chalk River is provided in Appendix I to this audit report.
Table 1 lists key NRC staff, licensee personnel, and contractors, identifying attendance during the November 2008 chemical effects audit meetings at Dominions Innsbrook facility.
Table 1: North Anna GSI-191 Chemical Effects Audit Participation Name Organization 11-12-08 Entrance Meeting 11-14-08 Exit Meeting Michael Henig Dominion X
X Christopher Burks Dominion X
X David Rhodes AECL X
X David Guzonas AECL X
X Richard Redmond Dominion X
X Mike Sekulic Dominion X
X Addison Hall Dominion X
X Robert Litman NRC consultant X
X Allen Hiser NRC X
Paul Klein NRC X
X John Lehning NRC X
X Matthew Yoder NRC X
X Harry Blake Dominion X
Bob MacMeccan Dominion X
X Bill Corbin Dominion X
X Thomas Shaub Dominion X
X Eric Hendrixson Dominion By phone Mike Whalen Dominion By phone Megan Sharrow Dominion By phone Thomas Jones Dominion X
Martin Legg Dominion X
Mike Rezendes Dominion X
Gary Nayler Dominion X
Allen Price Dominion X
Mark Sartain Dominion X
Delbert Horn Dominion X
Donnie Harrison NRC By phone 2.0 Overall Chemical Effects Approach The licensee evaluated potential plant-specific chemical effects by considering possible interactions between the materials in containment and the projected post-LOCA environment.
NAPS Units 1 and 2 both control the post-LOCA pH by adding sodium hydroxide (NaOH) to the recirculation spray. A range of post-LOCA pool pH values were calculated using a Monte Carlo analysis methodology (95% confidence value) considering the volumes of fluids, concentrations of NaOH, etc. Post-LOCA temperature profiles were determined using the GOTHIC Code.
Given the postulated NAPS plant-specific conditions, the post-LOCA chemical source term was determined using a combination of analysis and experiments. Based on the available test data for aluminum corrosion in alkaline, borated waters, AECL developed an aluminum corrosion relationship as a function of pH and temperature. The total dissolved aluminum in a post-LOCA sump pool at NAPS was calculated using the AECL model. Bench testing was performed in simulated plant-specific post-LOCA sump pool environments at the AECL Chalk River facility.
One of the bench testing objectives was to determine the point where aluminum hydroxide would precipitate over a range of parameters of interest for the Dominion plants, including NAPS. Initially, aluminum was added to the bench tests at a high pH value, and the solution was titrated to lower pH until the onset of precipitation. After determining the pH for the onset of precipitation with this technique, longer-term (30 day) follow-on tests were performed with test solutions one pH unit higher to evaluate aluminum solubility for a time period that is more representative of an emergency core cooling system mission time.
The initial goal of the bench test program for NAPS was to show that no precipitation would occur in the projected plant-specific post-LOCA environment. Since the bench test results indicated precipitation could occur, the licensee concluded that additional chemical effects testing was needed. Therefore, Dominion performed integrated chemical effects head loss testing for NAPS at the AECL Chalk River facility. In particular, a multi-loop test facility identified as Rig 89 was fabricated to perform these tests. Design and operation of the Rig 89 test loops are discussed in further detail in Section 3.1, AECL Test Facilities.
The NAPS Rig 89 integrated chemical effects tests were performed in a simulated post-LOCA pool environment containing representative amounts of boron and scaled amounts of plant-specific debris. Test loop pH was adjusted to a representative value using NaOH. The test loop temperature was held constant at 104°F (40°C). Plant-specific particulate debris quantities and the quantity of fiber needed to develop a thin bed were added in increments to the test loop.
After a stable baseline head loss was established across the test strainer section, sodium aluminate was added in small batches with the objective of having the dissolved aluminum concentration in the Rig 89 test loop equal the predicted plant-specific calculated dissolved aluminum concentration. Since the Rig 89 loop aluminum addition is scaled according to the post-LOCA pool concentration (instead of scaling the aluminum precipitate mass to the strainer area), precipitation of an aluminum compound during the test could result in a non-conservative dissolved aluminum concentration in the test loop. Therefore, if dissolved aluminum measurements indicate precipitation of an aluminum containing compound occurred during the test, more sodium aluminate is added to the test loop, up to an amount that would represent the maximum amount of aluminum precipitate mass per strainer area for the plant.
The total head loss measured in the Rig 89 test loop represents the plant-specific, integrated head loss across the sump strainer for plant debris and chemical effects.
3.0 Integrated Head Loss As part of the chemical effects audit for North Anna, the staff performed a review of the non-chemical portion of the debris bed head loss testing methodology and results. The head loss from the non-chemical debris is pertinent to the chemical effects audit because the filtration and accumulation of precipitate in the debris bed, and hence the resultant overall head loss impact attributed to chemical effects, depends upon the formation of a prototypical non-chemical debris bed.
An NRC staff review of the head loss testing conducted for North Anna prior to the staffs audit for Generic Letter 2004-02 corrective actions in July 2007 was documented in the staffs audit report [1]. In light of this earlier review, the staffs head loss review for the chemical effects audit focused primarily upon systematic differences that had been observed for similar non-chemical debris loadings in two different AECL head loss test rigs used for testing Dominion PWRs.
These systematic differences in head loss were first identified during the staffs trip to observe chemical effects head loss testing at AECL in May 2008 [2], see Appendix I.
3.1 AECL Test Facilities Head loss tests for North Anna were performed by AECL in two different head loss test rigs.
The earlier tests for North Anna were performed in the reduced-scale tank (Rig 33, see Figure 1), and did not include chemical precipitates. The final tests for North Anna were performed in the multi-loop test facility (Rig 89, see Figure 2), and were longer-term tests that included the modeling of chemical precipitation and the measurement of the head loss impact of the precipitates that accumulated in the debris bed.
The reduced-scale test tank is a cylindrical tank roughly 7.5 ft in diameter and 5 ft high. The test fluid was service water supplied by the Ottawa River that had been filtered and chlorinated by AECL. The test fluid was maintained at a temperature of 104°F (40°C). Debris was typically added to the tank from buckets near a mechanical stirrer used to discourage debris settling.
Debris settling was further discouraged through the positioning of the pump discharge line, which induced turbulence along the tank floor. Baffles were positioned around the strainer to prevent the induced turbulence from disrupting the formation of a uniform debris bed.
In early 2008, the multi-loop test rig was constructed so that head loss testing for several Dominion PWRs (North Anna Units 1 and 2, Surry Units 1 and 2, and Millstone Units 2 and 3) could be performed in parallel. Each of the six loops of the multi-loop test rig consists primarily Figure 2: Multi-Loop Test Rig 89 (1 of 6 Loops)
Figure 1: Reduced-Scale Test Tank (Rig 33)
of a 16-inch by 16-inch by 36-inch box housing the test strainer, a 12-inch-diameter by 18-inch-long cylindrical debris addition tank, a pump, and associated piping, components, and instrumentation. The test fluid was deionized water maintained at 104 °F. Debris was added to the debris addition tank, where it was stirred with a mechanical stirrer until a valve was opened that would allow the debris to transport down to the box housing the test strainer. Prior to adding chemical precipitates to the multi-loop test rigs, the non-chemical debris bed head losses were allowed to stabilize.
A comparison of selected parameters for the two test rigs is provided in the table below. Note that the reduced-scale tank test protocol underwent revisions during the course of the North Anna testing, and that the table below is intended to reflect the revised procedure used for the (non-chemical) design case tests.
Table 2: Comparison of Selected Test Rig Parameters Parameter Reduced-Scale Tank (Rig 33)
Multi-Loop Rig (Rig 89)
Test Fluid Filtered and chlorinated water from Ottawa River Deionized water Test Fluid Volume (L) 5000 230 Temperature (°F) 104 104 pH Not controlled 7.0 Test Strainer Area (ft2)
RS1 Strainer LHSI2 Strainer 9.4 16.9 5.74 5.74 1 Recirculation spray 2 Low-head safety injection 3.2 Safety Systems Drawing Suction from Containment Sump Both North Anna units are Westinghouse 3-loop PWRs with subatmospheric containment designs. During the recirculation phase of a design-basis accident, low-head safety injection (LHSI) pumps and recirculation spray (RS) pumps draw suction from the containment recirculation sump. The RS system provides long-term containment heat removal by passing sump water through a heat exchanger and then spraying it into the containment atmosphere.
The NRC staff recently approved a license amendment to change the start signal for the RS pumps to the coincidence of signals for high-high containment pressure and a wide-range RWST level of 60%. The LHSI pumps provide low-pressure, high-flow-rate cooling to the reactor core and are aligned to the containment sump when the refueling water storage tank (RWST) reaches its low-low level setpoint.
At North Anna, each unit has a single recirculation sump that provides the common suction for the LHSI pumps and the RS pumps of both trains. A photograph showing a section of the AECL Finned Strainers' installed at North Anna is provided below as Figure 3. In the photograph, the upper fins of the strainer belong to the LHSI system and the lower fins belong to the RS system.
Separate strainers are provided for the RS and LHSI systems because the RS pumps begin drawing water from the containment sump significantly earlier than the LHSI pumps. This design allows the RS pumps to take suction through strainer fins that are fully submerged for
' Trademark of Atomic Energy of Canada Limited
the reduced water level conditions at the time the RS pumps are actuated, while also allowing the upper strainer fins used by the LHSI pumps to take advantage of the increased water level available at the time their suction is switched to the sump.
Figure 3: A Section of North Annas Containment Sump Strainers 3.3 Observed Systematic Non-Chemical Head Loss Differences As mentioned above, systematic differences in non-chemical debris bed head loss were observed between tests conducted for North Anna in the reduced-scale tank and the multi-loop test rig. The strainer design case head loss results for similar debris loadings in the reduced-scale tank and the multi-loop rig (prior to the introduction of chemical precipitates) are shown in the table below.
Table 3: Comparison of Head Loss Test Results at 104 °F [3]
Strainer Reduced-Scale Tank Head Loss (ft)
Multi-Loop Rig Head Loss (ft)
Ratio (Multi-Loop /
Reduced-Scale)
RS 4.8 3.2 0.69 0.14 0.22 LHSI 3.2 3.0 1.7 0.53 0.57 The LHSI measured head loss in the single multi-loop rig test was slightly more than half of the measured head loss values for the two tests conducted in the reduced-scale tank. The discrepancy was even more significant for the RS strainers, for which the head loss measured in the multi-loop rig test was only 14-22% of the value measured in the reduced-scale tank tests.
As noted in Appendix I, similar systematic differences in non-chemical debris bed head loss were also observed for other Dominion PWRs, for which a similar series of tests had been conducted. These other plants head loss results were beyond the scope of the North Anna
chemical effects audit, and the staffs conclusions in this report are not intended to be applied to these plants directly. However, due to the similarity of the strainer testing methodologies, some results for other Dominion PWRs were reviewed as part of this audit in order to gain insights into the evaluation of the North Anna head loss testing results.
The licensee presented several possible reasons to explain the systematic differences in measured debris bed head loss between the reduced-scale tank and multi-loop test rig for the Dominion PWRs. The possible reasons included the following:
The potential for filtration of fine particulate from the Ottawa River suspended in the service water used as the test fluid in the reduced-scale tank.
The potential for biological growth in the debris bed due to organisms from the Ottawa River suspended in the service water used for the reduced-scale tests.
The potential for excessive deaeration across the debris bed due to the inability to model prototypically the full strainer submergence for some plant configurations in the reduced-scale tank.
These reasons are discussed in further detail below as pertaining to the observed differences in the head loss tests conducted for North Anna.
3.3.1 Ottawa River Particulate AECL observed that, even after being filtered, the service water taken from the Ottawa River that was used for the reduced-scale tank tests contained fine suspended particulate. The licensee hypothesized that this suspended particulate was filtered out in the debris beds formed in the reduced-scale tank, resulting in a significant head loss impact that was not prototypical of the plant condition. The multi-loop test rig used deionized water as the test fluid to minimize the potential influence of suspended impurities on the test results.
The service water used for the reduced-scale tests for North Anna had been successively filtered through 200-m and 10-m filter bags prior to the initiation of the test. After being filtered, measurements of the remaining suspended particulate were made and compared to similar measurements made for the multi-loop rig tests. An example of the total suspended solids (TSS) measurements for tests conducted for the RS strainers is shown below in Table 4.
Table 4: Example of Total Suspended Solids Measurements [3]
Test Rig TSS (mg/L)
Standard (1.5-m filter)
Fine (0.1-m filter)
Reduced-Scale Tank 3
5 2
6 Multi-Loop Test Rig
< 0.2 2
< 0.2 2
From the quantities of fine particulate measured to be present in the test fluid, the licensee calculated the total mass of the fine particulate and a particle number based upon the assumption of a 0.2-m particle size [3]. Based on this assumption, the licensee stated that the number of silt particles from the Ottawa River was several orders of magnitude larger than the number of walnut shell particles added to the test to simulate failed coating and other sources of particulate debris [3]. The licensee further compared the impact of the fine river particulate to
that from Microtherm insulation debris, which was added to one of the early head loss tests for North Anna and resulted in a rapid head loss increase (the licensee subsequently replaced the Microtherm in question with a different insulation [1]).
After reviewing the licensees analysis of the Ottawa River silt that is summarized above, the staff concluded that the presence of this fine particulate did not provide an adequate basis to explain the discrepancy between the head loss results in the reduced-scale tank and multi-loop test rig. In particular, many assumptions made in the licensees calculations appeared to significantly overestimate the impact of the river water particulate. The primary factors leading to the staffs conclusion are as follows:
The licensee had not performed head loss testing to directly examine the effect of the river particulate in the absence of other variables. Without such testing, the influence of river particulate could not be reliably estimated. Also, in the analytical calculation of the significance of the river particulate, the licensee had not validated many important assumptions that had substantial uncertainty associated with them, the most significant of which are elaborated upon below.
The licensee assumed that the river particulate was uniformly 0.2 m in diameter. In actuality, the staff expected that much of the fine particulate would be distributed more evenly in an approximate range of 0.1-1.5 m. Images AECL took of several samples of river particulate that had been analyzed with a scanning electron microscope confirmed the staffs expectation; however, this information had not been considered in the estimation of the significance of the river particulate. Assuming complete or essentially complete filtration (as the licensee did), head loss correlations would predict that a more even particulate size distribution with a larger average size would lead to a reduced impact on head loss compared to the licensees assumptions.
The licensee assumed that debris beds would be capable of effectively filtering 0.2-m particulate. The staff expected that much of the 0.2-m particulate would actually be capable of repeatedly passing through the pores in the debris beds. Had the debris beds formed by AECL been capable of effectively filtering 0.2-m particulate, the staff expected that the measured debris bed head losses would have been significantly in excess of the values shown in Table 3.
The mass of Microtherm added to the early test where the significant head loss increase occurred was approximately 2.4 lbm; whereas, the mass of river particulate present in the reduced-scale test tank typically ranged from approximately 0.06-0.08 lbm. Due to the substantial difference in mass, the staff considered it very unlikely that the river particulate could have a similar effect to that observed for Microtherm. Furthermore, the staff noted that microporous insulations such as Microtherm have been shown to be more effective at increasing debris bed head loss than equal masses of other typical particulate sources.
The licensee stated that, with an assumed particulate size of 0.2 m, the quantity of silt particles was 3 to 4 orders of magnitude greater than the number of 10-m walnut shell flour particles added to the North Anna tests. The staff noted again that, on a mass basis, the river particulate was essentially negligible (0.06-0.08 lbm), whereas the walnut shell flour masses ranged from approximately 1.3-8.8 lbm. Although the number of particles was computed to be greater for the river silt (based on the licensees assumptions evaluated above), the staff noted that the number of 0.2-m pores in the debris bed may exceed the number of 10-m pores by a similar factor or more. The staff considered it very unlikely that the minute quantity of river silt suspended in the test fluid had a significant effect on the final head loss relative to the walnut shell flour.
3.3.2 Biological Fouling The head loss tests performed by AECL in the reduced-scale tank typically lasted several days to a week. The licensee attributed part of the long-term head loss increase experienced in these reduced-scale tests to the growth of organisms in the debris bed which slowly reduced the bed porosity and hence resulted in a gradual increase in measured head loss. The origin of the biological organisms was thought to be the service water from the Ottawa River, and the licensee considered the biological fouling phenomenon to be closely associated with the river silt discussed above. The licensee considered this biological growth to be non-prototypical of the plant and, in the multi-loop test rig used deionized water as the test fluid and disinfected the debris used for bed formation to preclude the potential influence of biological growth on these tests.
In order to mitigate the potential head loss impact due to biological growth in the reduced-scale tank tests, the licensee added bleach to the test tank to achieve an initial chlorine concentration over 10 ppm during the heating and filtering of the test fluid prior to the start of testing. However, most debris bed samples taken following the completion of head loss testing still showed evidence of some biological growth. Based on a comparison to shorter strainer pass-through tests that had lasted roughly 6-8 hours and did not show evidence of biological growth, the licensee suspected that the biological growth had predominately occurred after the chlorination had lost its potency (e.g., after 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />).
After reviewing the licensees analysis of biological fouling that is summarized above, the staff concluded that the growth of biological organisms in the debris bed did not provide an adequate basis to explain the discrepancy between the head loss results in the reduced-scale tank and multi-loop test rig. The primary factors leading to the staffs conclusion are as follows:
The licensee had not performed head loss testing to directly examine the effect of biological fouling in the absence of other variables. Without such testing, the influence of biological fouling could not be reliably established.
The staffs examination of the licensees head loss versus time traces for the reduced-scale testing showed that the most significant part of the head loss increases appeared to be fairly rapid, as opposed to the gradual increases that would be expected from biological fouling. In other cases, gradual increases appeared more consistent with the filtration of particulate from the test fluid following the addition of debris to the test rig, with a leveling of the head loss as the filtration process was completed. The staff could not conclude that the head loss versus time traces indicated a significant impact from biological fouling.
No metric had been developed to determine what quantity of biological fouling was necessary to contribute significantly to the measured strainer head loss.
After being filtered and chlorinated, it was unclear that the service water used for the AECL reduced-scale tests was fundamentally different than the tap water used for head loss testing by other strainer vendors. The staff considered it possible that a similar degree of biological growth to that experienced at AECL may occur in other test vendors debris beds during long-term tests, but without being considered a significant contributor to the measured head loss.
3.3.3 Deaeration Across the Debris Bed The submergence of the test strainer in the reduced-scale tank was not modeled prototypically for all head loss tests conducted for the Dominion plants. As a result, the licensee noted that the effect of deaeration resulting from the test fluid undergoing a pressure drop at the debris bed would be more severe for the test condition than for the plant condition. In addition, the containment pressure credited in North Annas net positive suction head (NPSH) margin analysis would also reduce the potential for deaeration to occur for the plant condition.
Two effects of deaeration in the reduced-scale tank tests were noted by the licensee during the audit: (1) deaeration as the test fluid passes through the debris bed that increases the differential pressure due to the two-phase flow through the debris bed porous medium and (2) the accumulation of air in the fins of the test strainer that creates an imbalance in the static head of water across the strainer, thereby increasing the differential pressure across the strainer.
The licensee performed air accumulation calculations for a number of reduced-scale tests and concluded that the effect of air accumulation for the North Anna tests was minor (e.g., 25-30%).
However, for several tests conducted in the reduced-scale tank for other Dominion PWRs, air was considered to have had a significant impact, and the presence of air downstream of the strainer was observable through a transparent section of piping.
After reviewing the licensees analysis of deaeration that is summarized above, the staff agreed with the licensees assessment that the effect of air accumulation on the North Anna tests was likely minor, and potentially somewhat less than predicted by the licensee. The primary factors leading to the staffs conclusion are as follows:
Although conservative means exist for determining the deaeration the test fluid would experience after undergoing a pressure drop across the debris bed (e.g., Henrys Law),
the dynamics of air accumulation inside a strainer volume is not considered amenable to accurate prediction. The licensee stated that calculations indicated that air bubbles larger than a critical size (e.g., on the order of tenths of millimeters, but which ultimately depends on the orientation of the strainer fins) would move to the tops of the strainer fin channels, whereas smaller bubbles would be entrained in the flow toward the pump.
Some of the calculations performed by the licensee estimated significant voiding in the strainer fins, to the point of assuming almost the entire fin was filled with air. Yet without being able to evaluate such complex effects as the dynamics of bubble coalescence, the rates at which air bubbles would enter and leave air pockets as a function of the size of the pockets, and the impact of the strainer and suction line geometry on the transport and accumulation of air, the quantity of air that accumulates in the strainer fins cannot be reliably calculated. As a result, the staff could not determine that the licensees estimates of the differential pressure effect due to the accumulation of air inside the strainer were reliable. Furthermore, the licensee had not adequately demonstrated that air would not fill the fins of the strainer under plant conditions in a manner similar to that for the test strainer.
Regarding the effect of deaeration increasing the differential pressure from the flow through the debris bed porous medium, the staff expected that this phenomenon would not be significant until a certain head loss threshold (related to the submergence of the test strainer) was exceeded. However, some reduced-scale test results displayed potential symptoms of air effects only at relatively large head losses (e.g., 8-10 ft),
whereas other tests displayed fairly similar symptoms at head losses that were less than the strainer submergence. Based on the interactions during the audit concerning these
results, the licensee did not appear to have identified a threshold for air effects that could consistently explain the range of behaviors observed in the head loss test results.
In addition, as described in Appendix I, the staff performed confirmatory deaeration calculations for several cases for different Dominion PWRs using the deaeration model in the NUREG/CR-6224 Correlation Software Package [2]. These calculations suggested that the void fraction downstream of the strainer for the test conditions in the reduced-scale tank typically should not have been excessive, particularly for conditions applicable to North Anna.
3.4 Additional Audit Issues The licensee assumed that the non-chemical debris loading for the LHSI strainers at North Anna would be 50% of the RS strainers loading. The staff reviewed the report from the July 2007 audit of North Anna and determined that this debris loading was accepted by the staff at that time based upon information from the licensee that the maximum flow percentage through the LHSI strainers assuming at least two RS pumps in operation would be 46% [1]. During the November 2008 chemical effects audit of North Anna, the licensee showed the staff reviewer a copy of the plant procedures that directed that at least two RS pumps remain in operation post-LOCA to support the debris-distribution assumptions made in the sump performance analysis.
However, based upon calculations received by the staff during the chemical effects audit, the staff observed that the LHSI strainer could draw up to 62% of the total recirculation flow (and debris), even with two RS pumps operating, rather than the maximum of 46% that had been assumed by the licensee in July 2007. Because this issue was not directly related to chemical effects and the time available to discuss issues with the licensee during the onsite portion of the audit was limited, the staff deferred this question to the RAI process on the Generic Letter 2004-02 supplemental responses.
3.5 Chemical Effects Head Loss Test Results Once the non-chemical debris beds in the multi-loop test rig had reached a suitably stable head loss value, AECL proceeded to introduce chemical debris in batches over an extended period of time. The total duration of the multi-loop rig tests was roughly three months. The results of the multi-loop rig tests for the North Anna RS and LHSI strainers are provided in the table below.
The first value provides the stabilized head loss for the non-chemical debris, and the second value provides the final head loss measured after the completion of the chemical effects portion of the testing.
Table 5: Multi-Loop Test Rig Results for North Anna at 104 °F System Non-Chemical Debris Bed Head Loss (ft)
Final Head Loss (ft)
RS 0.69 6.0 LHSI 1.7 6.7 Based on the Rig 89 test results, the licensee recognized that reducing the aluminum inventory in containment would be necessary to ensure the conservatism of the limiting aluminum concentration assumed for the post-LOCA sump pool. Therefore, aluminum ladders were removed from the North Anna containment. Since aluminum is an important contributor to chemical effects at North Anna, the NRC staff was interested in comparing the predicted plant-specific aluminum release between the AECL method and the WCAP-16530-NP spreadsheet.
The licensee provided a comparison of aluminum release for the two different methods as a
function of pH. For a pH of 8.5, which was used to calculate the North Anna aluminum release, the AECL method predicted a slightly higher aluminum release than the WCAP method.
3.6 Analytical Conservatisms In light of the considerations discussed above, the staff did not agree that the licensee had developed a sufficient technical basis to fully address the observed differences in the non-chemical debris bed head loss results for similar debris loadings added to the reduced-scale tank and the multi-loop test rig. Therefore, the staff suggested that the licensee document significant conservatisms that were incorporated into the strainer performance analysis that could potentially mitigate the uncertainties associated with the differences in the measured head losses between the two test rigs.
Near the end of the onsite audit, the licensee provided the staff a five-page list of conservatisms that were incorporated in the sump performance analysis. The conservatisms covered a range of different aspects of the strainer performance analysis, including the following:
Debris generation Debris transport Latent debris Chemical effects Downstream effects Head loss testing Pump net positive suction head After reviewing the list of conservatisms, the staff concluded that the licensee had incorporated significant conservatism in many areas of the sump strainer performance analysis. Some of the conservatisms that the staff considered to be particularly significant included the following:
Conservative zones of influence from Nuclear Energy Institute (NEI) 04-07 were used to estimate debris generation.
Full transport was assumed for miscellaneous debris materials, which resulted in a sacrificial strainer area of 150 ft2.
Conservative debris transport fractions were assumed for all debris types, with 100%
transport assumed for most debris types.
All failed coatings were assumed to be in the form of fine particulate debris.
The head loss testing protocol prepared the fibrous debris into a relatively fine size distribution, whereas the plant debris distribution also included small and large debris pieces.
Conservative debris sequencing was used for the thin bed tests.
The calculated post-LOCA pool equilibrium pH for North Anna is 8. Plant specific aluminum release was calculated at a pH of 8.5, and the Rig 89 multi-loop rig tests were performed at a pH of 7. These values provide for a conservative amount of aluminum release and a conservative amount of aluminum precipitation in the test loop relative to that projected for the plant-specific environment.
The licensee has added margin into their calculations for plant-specific aluminum.
No credit was taken for long-term subcooling of the sump fluid in the calculation of pump NPSH.
The staffs review of the licensees list of conservatisms resulted in increased confidence that the uncertainties associated with the differences in reduced-scale tank and multi-loop test rig head loss results were bounded. However, due to the difficulty in quantifying the conservatism
13 inherent in these assumptions, the staff initially considered the difference in the head loss results between the two test rigs to be a draft open item at the conclusion of the onsite audit.
To address this draft open item, the licensee provided additional information to the staff directly following the onsite portion of the chemical effects audit to quantify the long-term increase in NPSH margin resulting from the decreasing temperature of the sump fluid as a design-basis LOCA progresses. The additional information is summarized in the table below.
Table 6: Short-Term and Long-Term NPSH Margin Values at 104 °F System Short-Term Debris Bed Head Loss Acceptance Criterion (ft)
Long-Term NPSH Margin to Offset Debris Bed Head Loss (ft)
RS 6.3
> 25 LHSI 7.5 12.5 The short-term debris bed head loss acceptance criterion bounds the non-chemical debris head loss results for both the reduced-scale tank and the multi-loop test rig (see Table 3). The short-term acceptance criterion similarly bounds the multi-loop rig final results with chemical effects, although the staff considers the final multi-loop test rig results to be affected by uncertainties associated with the formation of the non-chemical debris beds. The additional long-term NPSH margin for the RS and LHSI pumps shown in Table 6, however, provides confidence that uncertainties associated with debris bed formation and the subsequent impact of chemical precipitates are bounded by the available margins. Based on information provided by the licensee, the long-term margins shown above would be present soon after the switchover to sump recirculation (e.g., within 2 or 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />). Based on the existing knowledge developed from chemical effects testing, aluminum-containing precipitates are not expected to occur immediately after a LOCA, since there is a time dependency associated with aluminum corrosion in the post-LOCA environment, and the elevated pool temperatures immediately following a LOCA favor the aluminum remaining in solution rather than immediately forming a precipitate. As a result, by the time the peak chemical effects head loss occurs, the staff expects that the additional long-term margin will be available to ensure functionality of the RS and LHSI pumps. Furthermore, test data generated at the Argonne National Laboratory [4]
indicates that, for a constant aluminum concentration, the North Anna multi-loop tests performed at a pH of 7 would be expected to have significantly more aluminum hydroxide precipitate compared to a test pH of 8 that would be more representative of the projected post-LOCA pool pH.
Therefore, based on the discussion above, the staff determined that the uncertainties associated with debris bed formation in the multi-loop test rig, which may have affected the final debris bed head losses with chemical precipitates, were adequately addressed by the conservatisms associated with additional NPSH margin gained in a relatively short time after the initiation of containment sump recirculation, as well as the other conservatisms in the licensees sump performance analysis that were reviewed by the staff.
3.7 Head Loss Summary The licensee concluded that the influences of river particulate, biological fouling, and deaeration were sufficient to explain the increased head loss of the reduced-scale tests relative to the multi-loop rig tests prior to the addition of chemicals. The licensee further concluded that, because these phenomena were not expected to be present in the plant containment pool, the multi-loop rig tests were more representative of the plant condition than the reduced-scale tank test.
14 Based upon the discussion above, the NRC staff does not concur with the licensees conclusions. A definitive cause of the head loss difference between the two head loss rigs could not be identified during the chemical effects audit or the staffs earlier trip to Chalk River to observe head loss testing in May 2008. The staff expected, however, that a significant part of the systematic difference in head loss could be attributed to differences in the debris preparation, addition, transport, and accumulation on the test strainers in the two test rigs. These differences are described further in Appendix I [2]. The staff considered the debris preparation, addition, and accumulation for the reduced-scale tank to be more prototypical of the plant condition than the multi-loop rig. Although the staff did recognize that the influences of river particulate, biological fouling, and dearation likely affected the measure head losses in the reduced scale test tank, based on the information provided by the licensee, the staff did not conclude that they were of primary importance for the North Anna test conditions.
4.0 Conclusions After considering the significant conservatisms incorporated into the licensees sump performance analysis, the NRC staff concludes that the uncertainties associated with the formation of debris beds in the multi-loop test rig are bounded. As a result, the draft open item discussed with the licensee during the onsite portion of the chemical effects audit is resolved, and the staffs chemical effects audit of North Anna is complete with no open items or requests for additional information.
5.0 References
[1]
U.S. NRC Audit Report, North Anna Power Station Corrective Actions for Generic Letter 2004-02, November 15, 2007, NRC ADAMS Accession No. ML072740400.
[2]
U.S. NRC, Appendix I (Attached), Report of the NRC Staffs Visit to Chalk River, Canada, to Observe Integrated Chemical Effects Head Loss Testing Performed for Pressurized-Water Reactors Operated by Dominion.
[3]
AECL, Discussion of the Results of Head Loss Tests Conducted in Rigs 89 and 33, GnP-34325-AR-001, Revision 0, October 2008.
[4]
Argonne National Laboratory Technical Letter Report on Evaluation of Long-Term Aluminum Solubility in Borated Water, ADAMS Accession Number ML081550540.