ML090400786
| ML090400786 | |
| 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: ML090400786 (15) | |
Text
ENCLOSURE 2 APPENDIX I 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 Travel Dates:
May 5-9, 2008 Travelers:
John Lehning, Reactor Systems Engineer, NRC/DSS/SSIB Paul Klein, Senior Materials Engineer, NRC/DCI/CSGB Robert Litman, NRC Contractor Location:
Chalk River Laboratories Chalk River, Ontario Canada Organizations: Dominion Energy (Dominion)
Atomic Energy of Canada, Limited (AECL)
Sensitivity:
Determination to be made pending review by Dominion and AECL for proprietary information.
Background/Purpose In response to Generic Letter 2004-02 (GL 2004-02), Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Pressurized-Water Reactors, pressurized-water reactor (PWR) licensees are evaluating the performance of their containment recirculation sumps and making any plant modifications necessary to achieve regulatory compliance according to approved mechanistic sump performance criteria.
Atomic Energy of Canada, Limited (AECL), is one of five vendors supplying replacement sump strainers to U.S. PWRs in support of their GL 2004-02 resolution activities. In the U.S. market, AECL supplied replacement sump strainers to 7 PWR units, including Millstone Power Station, Units 2 and 3; Surry Power Station, Units 1 and 2; and North Anna Power Station, Units 1 and 2, which are all operated by Dominion, as well as V.C.
Summer Nuclear Station, which is operated by South Carolina Electric and Gas. AECL became involved with sump strainer performance issues in the mid-1990s and has since completed the design and testing of sump strainers for reactors located both in Canada and abroad.
Prior to the NRC staffs May 2008 trip to Chalk River, the staff understood that AECL had already completed non-chemical testing for its client PWR licensees in the United States. However, the chemical effects testing protocol had not been developed by the time of the staffs previous trip to observe head loss testing at Chalk River in June 2006, or by the time of the staffs audit of North Anna Power Station corrective actions for Generic Letter 2004-02 in July 2007.
2 In late 2007 and early 2008, the NRC staff held several discussions with Dominion and AECL concerning their plan for completing chemical effects testing. The staffs primary objective was to ensure that technically adequate testing would be completed by Dominion on a schedule commensurate with the overall plan for completion of the GL 2004-02 review activities. To ensure that the chemical effects testing would be completed in a timely manner, Dominion and AECL constructed a new multi-loop test rig that would allow chemical effects head loss testing for its PWRs to be performed in parallel and developed procedures for performing tests in this new rig. Dominion invited the NRC staff to observe this testing.
The chemical effects testing protocol developed by AECL that was used for the Dominion test program was different than the approaches ultimately used by many other test vendors, which were typically based on the WCAP-16530-NP methodology.
Therefore, the purpose of the NRC staffs trip to Chalk River in May 2008 was to observe chemical effects head loss testing in the recently constructed multi-loop test rig and to discuss the bench-top testing results used to justify the chemical effects head loss testing procedure.
Desired Outcome The NRC staffs trip was intended to support the staff=s resolution of issues associated with Generic Safety Issue 191 (GSI-191), Assessment of Debris Accumulation on PWR Sump Performance, for several PWRs operated by Dominion. In particular, the staff=s observations of Dominion=s chemical effects head loss testing at the AECL test facilities support evaluations of the licensees supplemental responses to GL 2004-02 in the areas of strainer head loss and chemical effects. The trip benefits the resolution and closure of GSI-191 by presenting an opportunity for the staff to observe the execution of AECLs chemical effects head loss testing procedures and to discuss with AECL personnel the bench-top chemical testing results and other information that form the basis for the procedures developed by AECL.
Results Achieved The staff fulfilled the essential trip mission described above. Specific accomplishments include the following:
(1)
Obtaining and reviewing the test plan for the AECL/Dominion multi-loop integrated head loss testing including chemicals, (2)
Observing the preparation and addition processes for particulate and fibrous debris for one loop of the multi-loop test rig, (3)
Observing a demonstration of the planned chemical addition process for the multi-loop test rig, (4)
Observing four multi-loop chemical head loss tests and one reduced-scale chemical head loss test that were in progress during the staffs trip, (5)
Observing the laboratories, test equipment, and some test materials used to perform the bench-top chemistry experiments that AECL used to develop its chemical head loss testing methodology, and (6)
Discussion of results achieved from the bench-top chemical testing used to justify the chemical effects head loss test procedures.
3 Summary of Trip The main focus of the staffs trip was to observe the chemical effects head loss testing being performed in the new multi-loop test rig at AECLs Chalk River Laboratories. The construction of the multi-loop test rig was motivated by Dominions objective of conducting head loss testing in parallel for Millstone, Surry, and North Anna in order to complete activities associated with GL 2004-02 on a schedule consistent with the NRC staffs plan for closure of GSI-191. The multi-loop test rig consists of six individual test loops, which is a sufficient number to perform simultaneous testing of the strainers for Millstone, Surry, and North Anna. Currently, the Dominion PWRs mentioned above are the only plants that have completed chemical head loss testing in the multi-loop test rig using the AECL test protocol. However, it remains possible that additional U.S. or foreign plants could seek to perform similar testing at AECL in the future.
Overall, the staff was impressed with the quality of the multi-loop test rig, particularly in light of the compressed schedule for its design and construction. An example of one of the multi-loop test rig loops is shown below in Figure 1, with the key features labeled. The staff also noted that the AECL and Dominion personnel present at the test site appeared to have a high level of expertise concerning strainer testing and that the test facility and test procedure appeared to have been designed with an awareness of many NRC staff comments previously made concerning test setups and procedures used by other strainer vendors.
Chemical Injection Point Heater Debris Addition Tank Cooler Strainer Box Primary Pump Spare Pump (common)
Sight tube Figure 1: Multi-Loop Test Rig (1 of 6 Loops)
4 Key Head Loss Testing Observations Based on observing the testing in progress during the trip to Chalk River, as well as discussions with licensee and vendor personnel, the staff made several observations concerning the head loss testing performed by AECL for Dominion PWRs. The most significant staff observation was that, prior to the chemical additions, the head losses for the debris beds formed in the multi-loop test rig were significantly lower than the head losses for previous non-chemical head loss testing with similar debris loadings conducted in AECLs reduced-scale test rig. Comparisons of head loss results from selected tests for the six strainer design cases considered in the testing are shown in Table 1 below [5]. Note that, although multiple tests were conducted for a number of cases, the staff considered that the selection of alternate data points for comparison would not affect the overall conclusion that the multi-loop rig test results were systematically lower than results for similar cases conducted in the reduced-scale test tank. Since the multi-loop rig testing was in its early stages at the time of the staffs trip to Chalk River, the data in the table below was based on post-trip teleconferences mentioned below and information obtained during the North Anna chemical effects audit.
Table 1: Comparison of Head Loss Results in the Reduced-Scale and Multi-Loop Test Rigs [5]
- Note that the reduced-scale result shown for the Surry recirculation spray strainer was based on a homogeneous debris addition sequence. Higher head losses (e.g., 2-3 ft) in other reduced-scale tests had been achieved for similar loadings by adding the particulate prior to the fibrous debris, although the vendor and licensee believe these tests may have been more sensitive to biological fouling.
- Due to reductions in the plant debris loading following reduced-scale tank testing, the Millstone 2 multi-loop rig pre-chemical head loss was predicted to be only approximately 0.46 of the reduced-scale test value using the NUREG/CR-6224 correlation.
Due to the observation during the staffs visit that the measured head loss for the North Anna recirculation spray strainer test in the multi-loop rig was spuriously low, the licensee decided to shut down this test loop prematurely so that the same test could be repeated to provide additional head loss data for the multi-loop test rig prior to the addition of chemicals. (Note that the Millstone 3 strainer test had not yet begun at this time.) Following the staffs visit to Chalk River, the North Anna recirculation spray test was restarted and a new debris bed was formed.
The licensee subsequently informed the staff that a pre-chemical debris bed head loss was achieved similar to that of the initial test that was terminated prematurely.
Strainer Reduced-Scale Head Loss (ft)
Multi-Loop Rig Pre-Chemical Head Loss (ft)
Head Loss Ratio (Multi-Loop /
Reduced-Scale)
Surry -
Recirculation Spray 0.90 0.60 0.67
- Surry -
Low-Head Safety Injection 1.2 0.25 0.21 North Anna -
Recirculation Spray 4.8 0.69 0.14 North Anna -
Low-Head Safety Injection 3.2 1.7 0.53 Millstone 2 1.9 0.60 0.32 **
Millstone 3 15.6 1.0 0.064
5 The licensee suggested that a significant part of the systematic differences in measured head loss could be attributed to the presence of fine particulate and biological matter in the service water used for the reduced-scale testing, and that the accumulation of these contaminants on the debris beds formed in the reduced-scale tank led to significantly higher head losses relative to the multi-loop rig testing, which used deionized water [5]. The licensee stated that, because the plant coolant is generated from deionized water, the fine particulate and biological matter in the AECL service water (ultimately derived from the Ottawa River) were not prototypical of the expected plant condition in this regard. Another cause of the discrepancy suggested by the licensee was deaeration and air accumulation in the fins of the test strainers used in the reduced-scale tank [5]. The licensee considered deaeration to be particularly significant for the Millstone 3 reduced-scale tank tests.
Based upon the preliminary information available during the May 2008 trip to Chalk River, the staff concluded that the licensees hypotheses explaining the lower head loss values in the multi-loop test rig were not supported by a documented technical basis and recommended that the issue be discussed further after the completion of the multi-loop tests so that the licensee could perform additional comparative analysis.
In an effort to understand the basis for the systematic discrepancy noted above between the head loss results for the multi-loop test rig and the reduced-scale test tank, the staff subsequently reviewed the steps in the licensees test procedure for the multi-loop test rig based on observations made during the trip to Chalk River. (Previous reviews of the corresponding procedure used by AECL to conduct head loss testing in the reduced-scale tank are available in an NRC staff trip report from June 2006 [1] and in the staffs GL 2004-02 audit for North Anna in July 2007 [2].)
The debris loadings used for the multi-loop tests were typically thin bed cases, since these tests were shown to be the most limiting condition for previously completed non-chemical testing. In the AECL thin bed test protocol, the addition of particulate was performed first, followed by the addition of batches of fibrous debris. The target thin bed thickness, which AECL chose based on previous testing experience, was typically 1/4-inch.
The debris preparation procedures appeared adequate in general. Particulate and fibrous debris were mixed up in separate batches, and the concentration of the prepared debris slurries appeared appropriate. The fluid used to generate the test slurries was taken from the test loops, however, and since the particulate was added to the test loops first, the test fluid used to generate the fiber slurries contained suspended particulate debris. Therefore, the addition sequence for particulate and fibrous debris was not considered by the staff to be purely heterogeneous. The fibrous debris prepared by AECL for the tests observed by the staff appeared to be sufficiently fine. Photographs of the prepared debris slurries were taken, although there did not appear to be objective acceptance criteria to ensure adequate fibrous debris preparation. A thin layer of the prepared fibrous debris typically floated in a mat on the surface of the barrel used to mix the debris slurry, as shown in Figure 2, below.
6 Debris slurries were poured into the multi-loop test rigs debris addition tanks (see Figure 1 above). AECL test technicians prepared and transferred debris carefully and thoroughly. Once a slurry was poured into the debris addition tank, shown in Figure 3 below, the debris addition tank was closed, and valves were manipulated to allow debris to transport from the debris addition tank into the strainer box (Figure 1). AECL test technicians attempted to open the valves on the multi-loop test rig slowly so that the fibrous debris would gradually be transferred from the debris addition tank to the strainer box. It was unclear to the staff how gradual the transport process between the debris addition tank and the strainer box was in practice, however, since visual observation was prevented by solid tank walls and piping, as well as the opaqueness of the test fluid behind the strainer box observation window. Since the flow of debris to the strainer box was due, not only to flow, but also to gravity, it is not clear whether a significant part of the fibrous debris transport occurred only after the valve to the strainer box had been opened past a critical value (e.g., transport occurring as a slug of fiber rather than a fine slurry). A layer of floating fiber was typically present on the surface of the debris addition tank after the debris slurry was added, as shown in Figure 3.
A stirrer was installed inside the debris Figure 2: Prepared Fibrous Debris Slurry for a Multi-Loop Rig Test Figure 3: Debris Slurry Poured Into Debris Addition Tank
7 addition tank to keep the debris slurry from agglomerating prior to its arrival into the strainer box, as shown in Figure 4, below. The staff observed that, when the debris addition tank was reopened after debris addition was thought to be complete, some small agglomerations of fibrous fines would occasionally remain on the surface of the debris addition tank. This fiber was subsequently broken up by AECL test technicians, and the debris addition tank was closed and restirred, allowing additional opportunity for the floating fibers to transport down to the strainer box. AECL test technicians ensured that the fibrous debris from the previous batch was transferred to the strainer box prior to adding the next batch of debris to the debris addition tank.
Once sufficient fiber had been added to the test tank, filtration of the suspended particulate occurred and the opacity of the water was reduced. The staff could eventually observe in several of the strainer boxes that the flow pattern appeared to result in a fairly uniform debris loading on the test strainers, as evidenced below by Figure 5. For a number of the test loops, the staff also observed that the flow in the strainer box was sufficient to keep the majority of the test debris in motion.
Figure 4: Debris Addition Tank Stirrer Figure 5: Debris Bed Formed in Multi-Loop Test Rig
8 The staff observed that the AECL test technicians used magnetic brushes to stir debris that initially settled on the strainer box floor, as opposed to transporting to the test strainer module inside the strainer box. A photograph of settled debris on the floor of one of the strainer boxes is shown below in Figure 6. Settling appeared to be more significant for the test loops that had smaller gaps between the strainer fins and lower recirculation flow rates. However, after the test technicians finished using brushes to stir the settled debris, the staff did not consider the small quantities of debris remaining on the strainer box floor to be of significance for any of the observed tests.
Following the addition of a batch of fibrous debris, the floors of the strainer boxes were typically brushed to resuspend settled debris. The staff observed that, in some cases, the head loss increase following the brushing of the tank was comparable to or greater than the increase associated with the fibrous debris addition itself. This observation indicated that a significant amount of the debris entering the strainer box likely first settled onto the strainer box floor and then was subsequently brushed back into suspension and drawn onto the strainer by the flow through the loop. The staff noted three possible effects of this transport sequence:
o First, the transport of part of the fibrous debris to the floor of the strainer box resulted in a splitting of each fibrous debris batch into two sub-batches, with the first sub-batch likely having an increased fraction of the finest debris fragments.
o Second, the intermediate step of settling part of the fibrous debris onto the strainer box floor prior to its arrival on the strainer could have provided an opportunity for fine debris to agglomerate into debris pieces of increased size.
However, the staff did not directly observe agglomeration of this debris, and it can be seen above in Figure 6 that the small clumps of fibrous debris that settled on the tank floor apparently remained relatively loose and fluffy.
o Third, the staff also observed that a small quantity of the particulate debris remained on the strainer box floor after the addition of several batches of fiber had clarified the test fluid. The cyclical resuspension of this settled particulate when the strainer box floor was brushed following the addition of batches of fibrous debris appeared to be equivalent to an addition sequence wherein a Figure 6: Fibrous Debris Settled in a Strainer Box
9 portion of the particulate loading was added to the debris bed in a staggered pattern with batches of fiber.
o AECL test technicians implemented rigorous practices to ensure sterilization of the test loop, the test fluid, and the debris added to the loop. For example, buckets used for debris preparation were wiped with a bleach solution and then rinsed with deionized water. Debris added to the test loop was autoclaved for a time and temperature considered sufficient to eliminate sources of biological growth (although it can be inferred from the floating fibers seen in Figure 2 and Figure 3 that the autoclave temperature was not sufficiently high to completely remove the fibrous binder). Such steps, which have not been observed at other strainer vendor test facilities, were motivated by AECLs observations during reduced-scale tank tests that biological growth during long-duration tests (e.g., multiple days) may have non-prototypically contributed to the increase in measured head loss [3].
Several power outages were observed to have occurred during the extended test runs in the multi-loop test rig. In discussions with the licensee near the conclusion of the multi-loop rig test runs, the licensee stated that the head losses for the affected test loops had returned to approximately the same level after power was restored.
Head tanks were installed on the multi-loop rig test loops to increase the static head of water above the surfaces of the test strainers in an effort to prevent deaeration across the debris beds. As shown below in Figure 7, the six head tanks were essentially large buckets located one floor above the multi-loop test rig, each connected to a test loop via piping. The staff did not perform a detailed review of the submergence level modeled for each test loop. However, it appeared that the enhanced submergence provided by the head tanks may have exceeded the actual plant strainers submergence levels in some cases.
From these observations discussed above, the staff could not conclusively identify the step or steps in the multi-loop rig test procedures that were responsible for the significant differences between the measured head losses in the multi-loop test rig and the reduced-scale test tank.
Based on the information available, the staff considered several observations made concerning the test procedure as potentially contributing to the difference, including (1) the use of test fluid with suspended particulate to prepare fibrous debris slurries, (2) the formation of matted layers Figure 7: Multi-Loop Rig Head Tank Arrangement
10 of floating fiberglass during debris preparation and insertion into the debris addition tank, (3) the uncertainty as to whether slugs of fibrous debris could transport out of the debris addition tank and into the strainer box, and (4) the potential for staggered debris addition sequences and/or debris agglomeration to result from debris temporarily settling on the floor of the strainer box and later being brushed back into suspension. However, as noted above, the staff did not have evidence that any of these observations resulted in deficiencies in the AECL testing in the multi-loop rig. Additional sensitivity testing with variations in the test procedure would be necessary to identify conclusively whether aspects of the test protocol or test geometry were responsible for the observed differences in measured head loss between the multi-loop test rig and the reduced-scale test tank.
As discussed in the staffs report for the North Anna chemical effects audit [3], the staff did not agree that the licensee had sufficient basis to demonstrate that the differences in measured head losses for North Anna were primarily the result of conditions in the reduced-scale tank test setup the licensee considered non-prototypical, namely silt particulate from the Ottawa River, the presence of biological fouling, and the effects of deaeration and air accumulation inside the test module. Based on the similarity of the head loss test procedures used for all of the Dominion PWRs that tested at AECL, discussions with licensee and vendor personnel during the trip to Chalk River, and information incidentally reviewed for these plants during the North Anna chemical effects audit, the staff expected that a similar conclusion would likely hold for Millstone and Surry as well. However, as discussed below, the reduced-scale testing for Millstone 3 appeared to have experienced significantly more deaeration than the other tests.
A second significant observation associated with testing at AECL was the presence of anomalous results in the plots of measured head loss versus time for reduced-scale tests for several Dominion plants. In one non-chemical head loss test performed for Surry, the head loss steeply increased following the addition of the final two batches of fibrous debris to the test tank, and then unexpectedly ramped downward. In a chemical head loss test performed for Millstone 3, the head loss trace oscillated unpredictably throughout the test and also demonstrated a decreasing trend as the quantity of dissolved calcium added to the test rig was increased. This decreasing head loss trend was unexpected because the calcium added to the test tank was expected to react with the phosphate dissolved in the test fluid to form calcium phosphate precipitate, thereby increasing the head loss. Despite the decreasing head loss trend, the measured head loss for the Millstone 3 test exceeded the test acceptance criterion. The licensee attributed these testing anomalies to the release of dissolved air that had accumulated under the debris bed during the test and, therefore, did not consider the test results valid. It was not clear to the staff that all of the observed anomalies during the tests could be attributed to air effects or were non-representative of the plant condition. In particular, since a large head loss (relative to the strainer submergence) is first necessary to generate significant deaeration, it appeared to the staff that a substantial part of the head loss in the Millstone 3 reduced-scale test could not be attributed to deaeration.
The licensee did not consider the reduced-scale tests as strainer design qualification tests, and as such, focused its efforts primarily on the multi-loop test program. However, the staff briefly discussed with licensee and vendor personnel the potential for deaeration in the reduced-scale tank testing. During the staffs visit to Chalk River, the chemical test for Millstone 3 in the reduced-scale test tank was ongoing, and the staff observed the presence of air bubbles in a clear section of piping on the test pump suction line. However, licensee and vendor personnel had not yet had an opportunity to perform calculations to determine the expected void fraction downstream of the test strainer that was considered to be a significant contributor to the measured head loss for that test.
11 Following the trip to Chalk River, the staff briefly analyzed the Millstone 3 test conditions in the reduced-scale tank using the deaeration model in the NUREG/CR-6224 Correlation and Deaeration Software Package in order to estimate the magnitude of the downstream void fraction. Based on the fluid conditions, minimum strainer submergence, strainer head loss, and other parameters that were considered representative for the Millstone 3 reduced-scale tank test referred to above in Table 1, the staff estimated that the void fraction that occurred through the test debris bed likely remained below an approximate peak value of 1.75% and likely was closer to 1% for much of the test. In two earlier Millstone 3 tests in the reduced-scale tank that the licensee also considered to have been influenced significantly by deaeration effects, the staff estimated that the void fraction had been on the order of 1% or less. Based on subsequent discussions of the deaeration model with AECL personnel during the North Anna chemical effects audit, it was realized that the results calculated by the staff should be considered upper bound values, since the deaeration model in the NUREG/CR-6224 software package appears to model the strainer as a horizontal flat plate and the staff calculation used the minimum submergence rather than a strainer-averaged value. The staff estimated that a more representative average submergence value may have decreased the void fractions calculated above by several tenths of a percent.
The staff noted that air ingestion is discussed in Regulatory Guide 1.82, which recommends that a 2% limit be imposed on air ingestion in the pump suction line for the purpose of ensuring adequate pumping performance. As described in NUREG-0897, Revision 1, and NUREG/CR-2792, this limit was derived based on testing that measured the degradation of pump head as a function of ingested air at the pump suction. Thus, the specification of the 2% recommended air ingestion limit did not account for the impacts on net positive suction head (NPSH) margin resulting from two-phase flow through a debris bed or the accumulation of air inside the strainer resulting in an imbalance in the static head across the strainer surface. Based upon the experience from the Millstone 3 test above, the 2% limit of Regulatory Guide 1.82 may not be sufficiently stringent to address all of the means through which air may affect sump performance. The licensee indicated that the Millstone 3 test conditions evaluated are not representative of the plant condition, because the reduced-scale tank could not accommodate the full strainer submergence for the plant condition. Based on a calculation using an increased submergence the licensee stated would exist for the plant strainer prior to the onset of the peak strainer head loss, the staff expected that a significant reduction in the calculated downstream void fraction would occur. In light of the discussion above, the staff concluded that the upcoming revision to Regulatory Guide 1.82 should address the additional means by which air could adversely impact strainer performance that appeared to be present during the Millstone 3 testing.
Considering the anomalous results mentioned above that were observed in the traces of head loss versus time for some of the AECL tests performed for Dominion PWRs, the staff questioned whether differential pressure phenomena had resulted in the disruption of the debris beds. In addition to performing a visual scan of the post-test debris bed to identify bore holes or other bed disruptions, the staff considered it beneficial to use other means to demonstrate conclusively that temperature-based scaling of test head loss results to the plant condition is justified, particularly when unexpected behavior is observed in a test. The staff noted that carefully performed flow sweeps or other means could be used to test the head loss response of the debris bed to a change in hydraulic conditions.
12 Key Chemical Effects Observations Most of the staffs observations during the trip to Chalk River focused upon non-chemical aspects of the multi-loop rig test procedures because most of these tests had not progressed to the point of chemical addition at that time. However, on the afternoon prior to the staffs departure from Chalk River, the licensee performed a chemical addition in one of the multi-loop rig test loops. The test loop to which the chemical addition was performed was that for the North Anna recirculation spray strainer that the licensee subsequently planned to restart with a new debris bed, as mentioned above. The staff observed that the licensee prepared the chemical solution by draining fluid from the test loop and mixing in sodium hydroxide flakes and sodium aluminate powder. The resulting chemical solution was subsequently metered into the test loop over a half-hour period using a cylinder/piston device that resembled an oversized syringe, as shown below in Figure 8. The chemical injection did not have an obvious impact on the test head loss, which was expected because the quantity of aluminum added to the test loop was relatively small. The licensee stated that the chemical addition protocol observed by the staff was for demonstration purposes only, and that the procedure for chemical addition in the formal design-basis tests had not been finalized.
When designing the multi-loop test rig, the licensee realized that simultaneously scaling quantities of chemicals to both the test fluid volume and the test strainer area was not feasible.
Specifically, because the ratio of the test strainer area to the test fluid volume in the multi-loop rig is much larger than the corresponding ratio for the plant, adding an amount of dissolved chemicals that would create a representative test fluid condition for modeling precipitation reactions would not lead to the generation of a sufficient quantity of precipitate when scaled to the test strainer area. The staff and licensee discussed how the multi-loop test protocol would compensate for this scaling issue. Although a finalized chemical addition procedure had not been developed by the licensee at the time, the licensee stated that, if indications of precipitation were observed in a given test loop, additional dissolved chemicals would be added to maintain a representative chemical concentration in the test fluid. The licensee further stated that the test fluid would be periodically monitored for evidence of precipitation and that Figure 8: Injection of Chemicals into a Multi-Loop Rig Test Loop
13 additional chemicals would be added to maintain a representative chemical concentration in the test fluid until an appropriate quantity scaled to the test strainer area had been added to the test loop.
Licensee personnel also described efforts to reduce the calculated quantity of dissolved aluminum in the post-accident sump fluid for Surry and North Anna. The licensee planned to reconsider existing containment analyses to determine whether the potential for reduced containment temperature and pH conditions could lower the aluminum concentration in the post-accident sump fluid into the range of 10 ppm. The licensee also noted that some aluminum equipment, such as ladders, could be removed from these plants containment buildings if necessary.
Post-Trip Interactions Due to the extended duration of the AECL multi-loop chemical effects head loss tests, the staff could not observe the execution of all key steps of the test procedure and critical aspects of the testing, such as the behavior of the measured head loss when significant quantities of chemicals were added to the test loops. Therefore, the NRC staff held two follow-on phone calls with Dominion and AECL to discuss the results of the multi-loop rig testing, on June 25, 2008, and August 6, 2008. The date of the first call was selected based on the expectation that the multi-loop tests would be completed within roughly 30 days from the start of the addition of chemicals. However, all six of the test loops were still running into the month of August, and a second phone call was arranged for August 6, when the multi-loop tests were essentially complete. Key points discussed in these teleconferences have been incorporated into the foregoing discussion.
Conclusions As described above, during the May 2008 trip to Chalk River Laboratories, the NRC staff identified several issues concerning the testing performed at AECL for Dominion PWRs which were not adequately understood. The primary issues identified included the systematic discrepancy between the head loss results for similar debris loadings in the reduced-scale tank and the multi-loop test rig and anomalous behavior in some of the reduced-scale tank tests.
Based on the evaluation of these issues during the North Anna chemical effects audit, the staff considers these issues to be resolved for North Anna. These issues will be considered for Millstone and Surry during the review of their generic letter supplemental responses and dispositioned appropriately at that time.
Notwithstanding the issues discussed above, the staff found the licensee and vendor personnel present during the trip to the Chalk River Laboratories to be highly knowledgeable regarding head loss testing and strainer performance in general. The staff was further impressed with the thoroughness and attention to detail of the AECL engineers and test technicians. Finally, the staff also noted that the licensee and vendor had both expended considerable effort in constructing a high-quality multi-loop test rig on a compressed schedule in order to complete the chemical effects head loss testing in a timely manner.
14 References
[1]
U.S. Nuclear Regulatory Commission, Foreign Travel Trip Report-NRC Staff Visit to Chalk River Laboratories to Observe Sump Strainer Head Loss Testing Performed by Atomic Energy of Canada, Limited, July 31, 2006, NRC ADAMS Accession No. ML062020596.
[2]
U.S. Nuclear Regulatory Commission, Audit Report, North Anna Power Station Corrective Actions for Generic Letter 2004-02, November 15, 2007, ADAMS Accession No. ML073100167.
[3]
U.S. Nuclear Regulatory Commission, North Anna Power Station Audit Report Corrective Actions for Generic Letter 2004-02 Chemical Effects, (Attached), Issuance pending.
[4]
U.S. Nuclear Regulatory Commission, Development and Implementation of an Algorithm for Void Fraction Calculation in the 6224 Correlation Software Package, January 2005.
[5]
AECL, Discussion of the Results of Head Loss Tests Conducted in Rigs 89 and 33, GnP-34325-AR-001, Revision 0, October 2008.
15 Attachment I: List of People Contacted Name Organization Title Addison Hall Dominion Lead, Strainer Testing Martin Legg Dominion Engineer Michael Henig Dominion Project Manager, GSI-191 Dave Guzonas AECL Senior Scientist Dave Rhodes AECL Principal Engineer Qingwu Cheng AECL Testing Engineer Jason Deadman AECL Design Engineer Shelly Maves AECL Project Manager Walter Hahn AECL Project Manager