IR 05000483/1997019
| ML20203J863 | |
| Person / Time | |
|---|---|
| Site: | Callaway  |
| Issue date: | 12/18/1997 |
| From: | NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION IV) |
| To: | |
| Shared Package | |
| ML20203J851 | List: |
| References | |
| 50-483-97-19, NUDOCS 9712220119 | |
| Download: ML20203J863 (20) | |
Text
. -.
-..
... -
,
.,. _.._,.--.-,-......--. -.
_.-
.
_ -.. -.
.
.
.
.
-
)
_j
-. -
m
- ENGl.OSURE
'
,
U.S. NUCLEAR REGULATORY COMMISSION
^
REGION IV
.
'
iDocket No.:
50-483 Ucense No.:
-
Report No.:
.50-483/97-19 l
Licensee:
Union Electric Company r
Facility:-
Callaway Plant Location:
Junction Hwy. CC and Hwy. O Fulton, Missouri Dates:
November 17 through December 11,1997 Team Leader:
C. Paulk, Senior Reactor Inspector, Maintenance Branch Team Members:
M. Chatterton, Nuclear Engineer, Reactor Systems Branch
.,
. E. Kendrick, Reactor Engineer, Reactor Systems Branch K. Parczewski, Senior Chemical Engineer Materials and Chemical Engineering
J. Carew, Physicist, Brookhaven National Laboratory Approved By:
Dr. Dale A. Powers, Chief, Maintenance Branch Division of Reactor Safety
,.
Attachment:
SupplementalInformation i
r
e
. -
h 9712220119 971218 PDR. ADOCK 03000483
-
'
-PDR
,
.-
.-,
.
-- -.-
-
. - -
.
. -., -..
--
..
.... -.. -. - -
-
____-_______ ____ _ _ ____ _ _
i
.-
i
'
..
i:
>
l EXECUTIVE SUMMARY
- Callaway Plant NRC Inspection Report 50-483/97-19 This reactive team inspection was conducted in response to the Callaway Plant experiencing a significant reactor fuel axial offset anomaly. The inspection was conducted onsite, in Union
- Electric Company offices, and in NRC offices. The team concluded that the actions taken by-Union Electric engineers and management were conservative and witt.in regalatory requirements.
- The performance of this inspection met, in part, the requirements for the _ Office of Nuclear Reactor Regulation's Core Performance Act on Plan for Region IV facilities.
Enaineerina The audit / surveillance / inspection activities of the licensee were judged to bo of sufficient
-
depth (Section E1.1).
The vendor and licensee's parallel and interactive reload design process allowed the
development of a nighly optimized reload core for the ninth operating cycle (Section i
'
E1,2).
For the ninth operating cycle, both vendor and licensee engineers employed the same
methodology used in previous cycles (Section E1.3).
The program to monitor, mitigate, and correct fuel failures was appropriate
. (Section E1.4),
The actions taken in response to the axial offset anomaly were conservative and in
-
accordance with regulatory requirements (Section E8).
A weakness was identified with respect to certain engineering staff knowledge of rod
-*
swap methodology (Section E8).
e
\\%
,
(
S
_ _ _.. _.. _ _ _. _ _. _ _
[.
-3-Reoort Detallo Ill. Enaineerina E1 Conduct of Engineering Beckoround The Callaway Plant is a pressurized light water reactor, with a Westinghouse nuclear steam supply system, owned and operated by Union Electric Company. The core contains 193 fuel assemblies. Tne plant is licensed to operate at 3565 MWt.
Beginning with the fourth ' operating cycle, unexpected behavior in the power distribution at the plant was observed. This anomaly was characterized by a gradual, unexpected power shift toward the bottom of the core and was first detected at a core average burnup of approximately 7 GWD/MTU in the fourth operating cycle. The power shift continued until burnup effects became dominant and caused power to shift back to the top of the core near the end t,f the cycle, in addition to the anomalous power distribution, deviations were observed in the estimated critical position of the control rods. Although the estimated critical rod positions for plant restarts that occurred early in the cycle agreed well with measured critical positions, this agreement disappeared for restarts that occurred later in core life. During the fourth and fifth operating cycles, estimated critical rod position deviations increased to more than 0.5 percer.t ak/k. This anomaly continued in the sixth operating cycle, was absent in the seventh operating cycle, returned in the eighth operating cycle, and continued in the current ninth operating cycle, only worse.
L After analyzing relevant data, performing scoping calculations, and reviewing industry experience, licensee and Westinghouse engineers concluded that the power distribution anomaly was most likely caused by the formation of crud and deposition of lithium borate, initiated by subcooled nucleate boiling, on the fuel rods. The estimated critical rod position deviations were another effect of this anomaly. Incore detector indications of flux depressions between fuel grids in high power fuel assemblies, and visual examinations showing crud deposits on fuel pins, supported licensee and Westinghouse engineers' conclusions.
Axial offset is a measure of the difference between power in the upper and lower portions of the core. This difference must remain within limits established in the technical specifications to ensure that both shutdown margin and fuel rod local peaking factors are not exceeded. Exceeding the e limits could result in the reactor fuel exceeding the 10 CFR 50.46 limit on fuel cladding temperature of 2200*F (1204*C).
' Dates for operating cycles are included in the Attachment.
-
_ - _ _ _ - _ _ _ - _ _ - _
.,
-
.-
-.
-
_
.
.
.
.
'
Based on investigations by licensee and Westinghouse engineers, the caut.c of the axial offset anomaly at Callaway has been attributed to a cruo buildup on fuel assemblies in
- the upper portion of the reactor core. This crud buildup is especially pronounced in high-
'
power assemblies. High core power results in increased subcooled nucleate boiling in the upper core, which, in turn, causes greater crud accumulation on the fuel assemblies.
Lithium borate is absorbed and concentrated in the crud layer, reducing the fission rate in the upper portion of the core. The visualindications of crud buildup and flux depression between the grid straps support these conclusions. Because of the reduced fissioning in the upper coro, the power distribution shifts toward the bottom of the core. This resulting powcr shift causes a teduction in shutdown margin and an increase in local peaking factors. Near the end of cycle, excess burnup in the bottom of the core and reduced boron and lithium concentrations in the reactor coolant system cause the power distribution to shift back toward the upper portion of the core, partially restoring the burnup dis'ribution.
While several other plants have experienced an axial offset anomaly, the -15 percent axial offset at Callaway is the largest known to date. In response to the anomaly and continued erosion of shutdown margin, power at Callaway was reduced to 95 percent on July 15,1997, and to 70 percent in mid-August 1997. On September 6,1997, with the unit at 70 percent power and with the axial offset at approximately 0 percent, the licensee, in consultation with Westinghouse, reduced reactor power to 30 percent in the hope of releasing the lithium borate from the crud, thereby creating a more positive axial offset when power was retumed to 70 percent. (A similar result had been seen in a previous cycle following a down power event at Callaway.) The results of this maneuver were not as predicted, however. After 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> at 30 percent power, only 25 percent of the expected amount of lithium borate had been released from the crud. When power was increased to 70 percent, approximately 70 percent of the lithium borate released during the power reduction was reabsorbed into the crud. The axial offset at the conclusion of the maneuver was more negative than at the start. Westinghouse and licensee engineers surmised that the cause of the worsening axial offset was the release of depleted boron from the crud during the puer reduction, followed by absorption of fresh (less depleted) boron into the crud during the return to 70 percent power.
During July and August, shutdown margin was decreasing at the rate of 10-15 percent-
_
millitho (pcm) per day. As shutdown rrargin approached the technical specification limit of 1300 pcm, licensee perscnnel took several actions to restore it. These actions included reducing power, and reducing operational flexibility by modifying the rod insertion limits and relaxing the rod worth uncertainties in the shutdown margin calculations (based on an NRC-approved Westinghouse topical report). By reducing power and introducing operational restrictions, the licensee has continued to operate the
-
Callaway Plant within the limits of the safety analysi._________-_ _ ___ _ _
__
_
.
1 e
.g.
E1.1 Evaluation of Licensee and Vendor Inte-face Activities and Licensee Oversiaht Activities a.
Inspection Scooe (92903)
The team reviewed the licensee / vendor interface activities for the sixth through ninth operating cycles, covering the period from 1992 to present. The team also assessed licensee oversight activities inebding audits and inspections of the vendor and internal licensee interfaces, b.
Findinas and Observations The team noted that Union Electric personnel performed engineering oversight of Westinghouse primarily through independent and parallel analyses, supplemented with visits by licensee engineers dunng the core design phase and participation in vendor seminars. The team also noted that the t'unctional areas were scheduled to have all areas audited on a 3-to 4-year cycle.
The team determined that the licensee / vendor response to the ninth operating cycle axial offset anomaly was fragmented until the formation of the Callaway Axial Offset Anomaly Team, which included a Westinghouse core design engineer. This was evidenced in a long turnaround period (up to 2 weeks) from taking the actual incore flux measurements to the generation of power distributions and determination of a shutdown margin.
The team found thz audit reports identified in the attachment were generally complete and of suft.
't depth to identify engineering interface problems. The team also found that the cognizant engineers were aware of the corrective actions and appeared to understand the findings.
t c.
Conclusions The audit / surveillance / inspection activities of the licensee were judged to be of sufficient
'
depth.
E1.2 Review of the Lictnsee's Core Meload Desian Process a.
IDipection Scoce (92903)
The team reviewed the licensee's reload design process for completeness and depth of analysis. The ninth operating cycle design activities of the fuel cycle management group were specifically tracked from mid-September 1995 to the completion of the supplemental startup and operations report. As in Section E1.1, above, interface with the fuel vendor and interaction with the licensee's fuel performance, safety analysis and radiological engineerhg, regulatory operations, and reactor engineering groups were examined.
f
__
- _ - _ - -
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.
.
,
l
.
-6-b.
Finoings and Observations The team found that the ninth operating cycle reload design ptocess was comprehensive and highly interactive with the vendor, involving many iterations o reload fuel and core design parameters, including a split-enrichment reload batch with varying burnable poison inventory. Full use was made of the licensee's independent parallel reload core design models to arrive at an optimized (high energy, low peaking factor) design to meet the Callaway energy requirements with the lowest rmmber of feed assemblies.
The team found that the documents and calculations to support the core design for the ninth operating cycle were complete and showed adequate independent review and internal oversight.
The team observed that the maximum U-235 enrichment (4.6 wt percent) and the number (128) of burnable poison rods for this reload batch were both higher than previous Callaway reloads. This, with the low number of feed assemblies used (88),
made the ninth operating cycle's core a more challenging design, with potentially higher power peaking factors. The presence of many face-adjacent feed assemblies also led to a higher power ring of fuel as the burnable poison inventory was depleted. Although all
,
design constraints were met, the incrementally " advanced" design may have contributed to an increase in sites for subcooled boiling and crud buildup with increased boron concentration and precipitation, which was the postulated cause of the increased axial offset. However, use of both the vendor's analysis of record methodology and the licensee's independent models indicated that all design limits were met.
c.
Conclusions The vendor and licensee's paraHol and interactive reload design process allowed the development of highly optimized reload core for the ninth operating cycle.
E1.3 Adeauacy of Methods for Calculating Core Design Parameters a.
Insoection Scoce (92903)
The team reviewed the licensee and vendor's methods for ce!culating core power distribution safety and operating limits.
b.
Findinos and Observations The team verified that approved methodology was used by the vendor for all steady-state, transient, and safety analyses of record. The licensee's indepen4nt methodology was also used according to licensee procedures. However, neither th; vendor's nor the licensee's engineers could predict the potential for the increased axial offset.
- _ - _ _
_ _ _ _ _ _ - _ _ _ _ _ _ - _ _ _ _ _ _____
_ - _ _ _ _ _ _
.
.
c.
Conclusions For the ninta operating cyck, both vendor and licensee engineers employed the same methodology used in previous cycles.
E1.4 Licensee Resoonse to Fuel Defects a.
Insoection Scoce (92903)
The team r3 viewed the fuel rod failure history for Callaway and the fuel vendor's fuel performance to assess licensee engineers' ccpability to monitor and respond to fuel failures.
b.
Findinos and Observations The team found the licensee's program for detecting, monitoring, and mitigating the effects of fuel failures was consistent with industry guidelines and with vendor recommendations. The team noted that the licensee's engineers identified two fuel
-
bundles in the core for the si>.th operating cycle with defective or leaking fuel rods and either reconstituted the bundles with inert rods or discharged the defective bundles. The
'Mm determined that the licensee's engineers attempted to identify the root cause of blures and to followup on vendor corrective actions. The licensee's engineers identified the failure mechanism to be a manufacturing defect that resulted in the failure of the upper end cap welds. The team found that the guidance provided in Supplement 1 to Generic Letter 90-02," Alternative Requirements for Fuel Assemblies in the Design Features Section of Technical Specifications," and inpical Report WCAP-13061-P-A,
" Westinghouse Fuel Assembly Reconstitution Evaluation Methodology," July,1993, were implemented appropriately at the Callaway Plant.
c.
Conclusions The program to monitor, mitigate, and correct fuel failures was appropriate.
E8 Followup of Axial Offset Anomaly (92903)
a.
Insoection Scone The scope of this inspection was to review and evaluate the actions taken by Union Electric personnel with respect to an axial offset anomaly e,.,.erienced at the Callaway Plant This review included the rod worth uncertainty determinations, the computer code models, and reactor water chemistry controls. To accomplish this, the team reviewed the documents listed in the Attachment, and interviewed cognizant licensee and Westinghouse personnel.
l
- - __ -________-_____ _______________________________.
.
..
- -
-
.
-
.
.
>
C
.
.g.
.
b.
Observations and Findinos b.1 Rod Worth Uncertainties During the review of Revisions K, N, and Q to Request for Resolution 17906, " Revised
. Rodworth (sic) Uncertainty for Callaway Cycle 9," the team noted that the licensee's engineers provided their justification to change the uncertainty on the rod worth values from 10 percent to 3 percent.' While the reduction to a 3 percent rod worth uncericlnty was approved in Revision Q for use by the on-site review committee, it was never-implemented. The team observed that the licensee's engineers provided their justification, in Revision N, to reduce the rod worth uncertainty only from 10 percent to 7 percent by referencing a Westinghouse Topical Report WCAP 9217, "Results of the Control Rod Worth Program." This reduction was approved and carried out.
The team observed that the three revisions relied on the accuracy of the startup physics test data from the sixth through ninth operating cycles. The team noted that the licensee's engineers based their justification for a rod worth uncertainty of less than 7 percent on the " excellent agreement between predicted and measured rod worth for the recent cycles." The team tried to verify the difference (measured - predicted) in the rod worth values by reviewing the rod swap results. None of the values listed in Revision K for the seventh through ninth operating cycles could be validated. Measured values for the seventh and eighth operating cycles were under predicted by 7.6 and 3.3 percent, respectively. The ninth operating cycle data showed excellent agreement with the predicted value (0.15 percent); however, Revision K listed the difference as 1.2 percent.
The team verified all the values listed in Revision Q. The team found the inability to validate the values and the incorrect values in Revision K to be examples of a A ek of attention to detail by licensee personnel, both the engineers who prepared the document
,
and the onsite review committee who approved the document.
The team had many questions related to licensee engineers' justification for changing the uncertainty in the rod worth to a value lower than 7 percent. The questions centered on the assumptions regarding rod worth uncertainty. The team noted that the licensee's engineers' approach did not include uncertainty on the predicted value for the stuck rod.
This approach had not been previously reviewed and approved by the NRC as had the use of the 7 and 10 percent uncertainties. The team was aware of the generally accepted knowledge that the accuracy with which the worth of an individual bank of rods can be predicted or measured is less than that for the total rod worth. Likewise, it is well known that the accuracy with which an individual rod (i.e., the rod with the highest rod worth) can be predicted is less than that of an individual bank. Thus, failure to include the uncertainty of the stuck rod when reducing the total rod worth uncertainty to 3 percent for_use in shutdown margin calculations has not been justified, as wat, communicated to licensee engineers and management personnel at the October 7 and 8,1997, meeting (discussed in Section E8.b.2, below) and subsequent telephone conversations related to tnat meeting. However, the licensee's engineers did not violate any regulatory requirements in approving the calculation.
'
.
_
_
O'
.
n 9,
The team noted that the predictions for the sixth and seventh operating cycles had been recalculated using an updated version of the computer code SIMULATE. The team found the results of these recalculations in Calculation NFDC97-013, " Revised Cycle.6 and Cycle 7 Rod Swap Calculations," Revision 0.
'
The team observed that the person who was responsible for the rod swap topical report which was submitted and reviewed in 1992, was knowledgeable; however, he was no longerin the group responsible for the work. The engineer presently responsible for the calculations lacked an understanding of the calculations necessary to perform the rod swap. The second engineer presently responsible for the measurements was uncertain
,concerning a part of the measurement procedure that affected the values obtained. The team considered the lack of understanding on the part of these two engineers to be a weakness.
The team was also informed that there was a transition of the core reload analysis and design group from the corporate offices to the site, and efforts were underway to ensure appropriate capabilities were maintained.
b.2 Shutdown Margin Calculat!ons Due to the severity of the axial offset anomaly, the team noted that the licensee and Westinghouse engineers updated the core model to match the measured data after each flux map was taken. Licensee and Westinghouse engineers then used the updated model to predict values used in the shutdown margin calculation. The team determined t
that the shutdown margin calculations were conservative for the burnup at which the flux map was taken. However, the model, as it was being used between July and October, was anconservative with respect to future shutdown margin predictions. Since there was considerable delay, as much as 2 weeks between the flux map and the model update, shutdown margin calculations during this period were nonconservative. This was most important during July and August before operational flexibility and power were reduced. During that period, the shutdown margin technical specification limit of 1300 pcm was being approached. Following a meeting conducted on October 7 and 8, 1997, among Union Electric, Westinghouse, and NRC personnel arid documented in a November 19,1997, meeting summary, licensee and Westinghouse engineers revised the model such that the predictions would be conservativo if the axial offset does not
. become more negative.
The team discovered that, whenever the model was updated, licensee engineers did not recalculate the shutdown margin using the latest information to find out if the shutdown
margin had changed. This would have provided information about whether or not the technical specification limit was being approached faster than was previously determinedc in fact, using the data from the' August 14,1997, flux map, ti. 2 team noted that the shutdown margin on August 19,1997, was 1326 pcm, while the calculated
,
- shutdown margin was.1350 pcm. The team determined the engineers' failure to recognize this nonconservatism in the process was a weaknes..
.
.
- 10-b.3 Computer Codes and Modeling
,
The team noted that the reduction in the core axial offset at the Callaway Plant resulting from the presence of borated-crud required a substantial modification to the ANC computer code neutronic modeling. Since the borated-crud was in the top of the core (above 84 in (213 cm) from the bottom of the fuel), a shift in the core axial power distribution toward the bottom of the core and an anomalous change in the axial offset resulted. The measurec nial offset was considered an anomaly since the design predictions did not include the observed change in the offset. The boron inventory in the crud was believeJ, by licensee engineers, to increase with increasing heat flux, crud thickness; and coolant boron concentration, pH, and temperature. However, the team noted that the ANC computer model employed an empirical approach in which the. crud-boron inventory was adjusted to match the measured core axial offset and power distribution. The Callaway licensing calculations and the prediction of the axial offset were done with the Westinghouse ANC/ PHOENIX-P code system and the on-line core monitoring was done with moveable INCORE-3D detector system using nuclear data determined with the PHOENIX-P code.
The team noted that the adjustments to the ANC model were based on the periodic measurements of the axially dependent reaction rates made with the movable INCORE-3D detector system. The change in the core power distribution (a maximum of approximately -15 percent) introduced by the borated-crud was (to a good approximation) a core-wide shift toward the bottom of the core that increased with fuel burnup. While the change in the radial power distribution is generally of second order and the core-wide axial offset was of primary concern, both the assembly integrated power and axial offset were considered in matching ANC modeling to the measurements.
The team reviewed the measurement data to ensure the reliability of the data and the validity of the ANC model adjustments.
The team noted that the measurement data included an axial trace of the incore detector reaction rate (or flux) taken at each of the 58 instrumented assembly locations. The team observed that after approximately 4 GWD/MTU into the ninth operating cycle, each of these traces showed a reduction in power in the top of the core and an increase in power in the bottom of the core, resulting from the accumulation of boron in the crud. In addition, at higher fuel burnup there was local power suppression (s10 percent) due to the preferential buildup of crud immediately below the structural grids and the three intermediate flow mixing grids.
The team found that the Callaway three-dimensional core performance analysis was done with the approved ANC code. The team noted that the ANC nodal cross-sectional, nuclear input data were determined with the PHOENIX P two-dimensionallattice physics code. It correlated moderator and fuel temperature, coolant boron concentration, fuel burnup, and control rod insertion. The PHOENIX-P code was also used to calculate local pin powers and detector reaction rates. The ANC/ PHOENIX-P code system had been bench marked to a broad set of critical experiments and reactor measurements
!
!
l
.
.
11-including incore detector reaction rates. The bench marking showed that the ANC/ PHOENIX-P code system predictions were well within the nuclear uncertainties established for Westinghouse analyses.
The team found that, as part of the Callaway core design quality assurance and verification, licensee engineers did independent core performance calculations with the SIMULATE-3 code system. The team noted that the same axial offset anomaly observed by Westinghouse engineers with the ANC/ PHOENIX-P code system was also observed by licensee engineers with the SIMULATE-3 code system.
The ANC/ PHOENIX P code system was used to perform Callaway core follow (trending)
and startup physics analyses and determine the key safety parameters for the analyzed transients. This included the calculation of the measured power distribution and anomalous axial offset throughout the cycle. The team found that the shutdown margin calculation done for compliance with the technical specifications was also calculated with ANC/ PHOENIX-P.
The team observed that the incore detector reaction rates were measured every 31 days (or more frequently depending on plant operation) and compared with the ANC predictions. The ANC prediction of the axial offset generally agreed with the measured axial offset early in the cycle (typically < 4-7 GWD/MTU) and tended to over predict the axial offset later in the cycle. To provide an accurate prediction of the core power distribution and the axial offset, the ANC model was adjusted to match the measured data. Based on the cenclusion presented in Report WCAP-14900," Root Cause of Axial Power Offset Anomaly," May 1997, that the anomaly was due to the presence of borated-crud in the top of the core, Westinghouse engineers analytically added a concentration of boron to the feed assemblies in the affected areas. Four distinct boron concentrations were used and the strength of each concentration was adjusted to match the observed core-wide axial offset, individual assembly offsets, and the assembly radial power distribution. The boron poison was added to most feed assembly locations (typically approximately 80 percent) and was uniformly distributed axially b6 tween 84 and 138 in (213 and 350 cm) from the bottom of the fuel. The B-10 microscopic cross-sections used to represent the si mil of borated-crud were the same cross-sections used to represent the boron coating on the fuel pellets in the bumable poison assemblies.
During the review of the ANC adjustment methodology, the team identified the following specific concerns that v.ere discussed with both licensee and Westinghouse engineers.
The team found that the engineers' responses addressed all concerns.
1)
ANC Accuracy. In response to the questior, of the accuracy of the adjustment, Westinghouse engineers showed thct the ANC adjustment resulted in agreement of better than 1 percent between the calculated and measured core-wide axial offset, and 1.5 percent between the nodal reaction rates. The team found that
'
this accuracy was comparable to the accuracy of the approved ANC model.
_ _
_ _ _ __-____ -_ _ _____ - __ __ -_ ____ _
--
.
- _
_ _.
_ _ _.
-
-_.
-
.
--
- l I
)
2)
PHOENIX P Reaction Rates. The team ncted that the PHOENIX P calculational code did not account for the borated crud and, consequently, over predicted the detector reaction rates, the reaction rate axial offset, and the boron required to
'
obtain agreement with the measured axial offset (for a given power distribution).
In response to this concern, Westinghouse engineers showed that this effect had beer, evaluated and was small (s 1 percent) and had been considered in the prediction of the coolant boron concentration by the ANC model.
3)
Axial Distribution of Boron. During the review, the team noted that the arsumed axially uniform distribution of borated-crud required a larger quantity of borori to produce a given change in axial offset. In response to this concern, the
'
Westinghouse 2ngineers did two additional calculations to evaluate this effect.
The team noted that the assumed uniform boron distribution resulted in an overestunation of the boron inventory by up to approximately 10 percent. This scWted in a conservative shutdown margin calculation.
4)
Reactivity Modelina. The team noted that the ANC modeling assumed that the axial offset anomaly was due to 'he accumulation of borated-crud on the outside of the fuel rods. However, in response to concems associated with the specific mechanism causing the axial offset anomaly, Westinghouse engineers evaluated the reactivity associated with two alternate mechanisms that alac produced a0reement with the observed axial offset. In the first case, the poison was modeled by increasing the absorber fast to-thermal cross section ratio by a factor of 10. In the second case, the fast to-thermal absorber cross section ratio was reduced by a factor of 10 (simulating the effects of voids). In both cases, the core reactivitv ' itroduced by these absorbers was essentially the same, resulting in equivalent predictions of the shutdown margin. This indicated the insensitivity of the shutdown margin calculation to the specific reac+.ivity modeling.
5; ANC Adjustment Freaueacy. Westinghouse engineers showed that the accuracy of the ANC prediction was comparaole to the approved ANC model after
,djustment, and during subsequent periods when there was no significant change in the operating state point (i.e., fuel burnup, power, flow, etc.). The team found that, when there was a substantial change in the core state point, the accuracy of the ANC prediction could degrade significantly and a model adjustment would be necessary. While the technical specification frequency requiremem for INCORE-3D flux maps was only once every 31 days, licensee engineers took flux maps and performed the ANC model adjustment approximately every 2 weeks.
The team noted that licensee engineers, in the determination of the shutJown margin, assumed that the boron that had accumulated in the crud was completely released to the coolant resulting in a substantial reduction in the core shutdown margin (approximately 2,000 pcm at 10 GWD/MTU into the ninth operating cycle). Since the crud boron inventary was determined by matching the measured axial offset and power distribution, l
- -
.-
, -,
- --
-
.
-_
.
_
---.
- -.
-
-_
o t
13
!
the team questioned whether the indirect boron determination would result in an accurete prediction of core shutdown reactivity.
In response to this question, licensee end,neers provided the results of recent boron measurements in these measurements, the change in boron inventory was inferred
,
from the change in coolant lithium concentration, assuming the boron left the crud as lithium borate (see E8.b.4, below). The measured increase in axial offset per gram decrease in crud-boron inventory was generally in agreement with the ANC calculation, confirming the boron added to tha ANC model, b.4 Reactor Water Chemistry Controls As noted above, licensee and Westinghouse engineers attributed the axial offset ancmaly occurring in the Callaway Plant to a non-uniform crud distribution on the fuel elements. The team noted that this theory was based on the fact that crud tends to deposit preferentially in the top of the core due to the increased temperature. The team also noted that ncensee and Westinghouse engir'eers concluded that the resulting crud deposits can, under certain circumstances, absorb lithium borate from the reactor coolant.
The team concluded that the transport and deposition of lithium borate in the crud deposits could be controlled by two factors: the amount of reactor coolant boiling and the temperature-dependent solubility limit of lithium borate. Boiling in crud deposits will occur due to high heat fluxes. This boiling will concentrate lithium borate in the incoming reactor coolant, and, along with high temperatures existing in these deposits, will cause the precipitation of the lithium borate. Both effects will be exacerbated in high duty fuel that operates with higher heat fluxes. Since the primary cause of axial offset is a difference in the temperatures existing between the upper and lower parts of the core, it is mainly controlled by the core design parameters. However, since the crud generated in the plant plays some : ole, proper primary coolant chemistry control could contnbute to reducing this phenomenon. Keeping reactor coolant clean will reduce crud deposition, and precipitation of lithium borate h the crud deposits can, to some extent, be controlled by the coolant pH. These considerations prompted the NRC inspection team to include review of the mmary coolant chemistry programs in the scope of the inspection.
The main sources of impurities in the reactor coolant are the corrosion products generated by corrosion of stainless steel and inconel 600 components in ihe primary coolant system. The purpose of the nrimary chemistry program is to reduce generation of these corrosion products, reduce the deposition in the core, and to remove them from the primary coolant system by the chemical and volume control system demineralizers.
The team noted that tne licensee achieved the first two objectives through controlling pH, minimizing oxygen concentration, and keeping dissolved hydrogen concentrations in the
,
i reactor coolant at a recommended value. Removal of impunties from the coolant was
!
achieved by operating the chemical and volume control system demineralizers during
i
,
-, _ - _
-. -,.
-.,
..,,
-
- - - -.
-
__
_ _ _ _
. _..
_ __
_
_
_ _ _ _ _. _ _ _ _ _.
____
-
o
.
l normal plant operation and during a shutdown, when crud was being removed from the system.
The primary coolant chemistry program in the Callaway Plant was based on the j
recommendations provided in the EPRI generic guidelines for pressurized light water i
reactor plants and in the Westinghouse plant specific guidelines.
The pH in the reactor coolant was controlled by adjusting lithium concentration, in the first through fifth operating cycles, the licensee maintained a coordinated pH control program with pH,(pH at the operating temperatures)just above 6.9. Initially, a considerable swing in lithium concentration was observed, but, in the later cycles, the lithium control band was considerably tightened. Tighter control of lithium reduced fluctuations of pH, and contributed to lower radiation fields. The team noted that, at the recommendation of Westinghouse, in the sixth operating cycle, pH control was changed to a modified pH control program that would allow the pH to increase during the second i
half of the cycle to 7.2. Due to a severe axial offset, this value wcs subsequently changed to pH, of 7.4 to provide increased crud dissolution. Later in the cycle, it was lowered to 6.9. In the subseg Jent two operating cycles (the seventh and eighth), a modified pH, control program was niaintained with 2.2 ppm :ithium and upper value of pH, of 7.4.
However, in the ninth ope';ating cycle, the team noted that there was a departure from this procedure. Since it was found that the axial 0ffset began to appear early in the operating cycle, licensee chemists thought that increasing pH at the beginning of the i
operating cycle was logical, to reduce general corrosion and release of corrosion products that later would redeposit in the upper core areas. Also, licensee chemists considered that maintaining a more constant pH, throughout the cycle may reduce the transport of crud in the core to out of core locations. Therefore, the lithium concentration wu increased slightly at the beginning of the ninth operating cycle tv near 2.5 ppm with the intcot to operate at a constant pH of 7.1. However, later in the ninth operating cycle, i
to reduce the axial offset, licensee chemists reduced pH, to 6.9. The team noted that, for the se')e reasons, licensee chemists were tentatively planning to operate at a constant value of pH, of 7.2 at the beginning of the tenth operating cycle.
The team concluded that the hydrogen concentraticri in the primary coolant of the Callaway Plant had been controlled within the recommended limits provided by EPRI and Westinghouse for all operating cycles. During the first through seventh operating cycles, hydrogen concentration was limited to a range of 25 35 cc/kg. This was due to the susceptibility of some materials in the primary coolant system to stress corrosion cracking This recommendation was subsequently revised, and, in the eighth and ninth operating cycles, the range was increased to 30-45 cc/kg. The higher amount of dissolved hydrogen helped to increase corrosion product solubility. The team found that the hydrogen concentration in the coolant was mainta:ned within the prescribed limits, with a very few depaitures The team was informed that, in the future, to increase crud
-.
.
-
-
-
.
.
. _ _ _
- __
-.
O
.
+ 15-solubility further, licensee chemists were also considering the use of ammonia in the primary coolant.
Corrosion products can be released from the crud deposits into reactor coolant whenever there is a sudden change in reactor power, temperature, or coolant chemistry Three large crud bursts occurred in the third through ninth operating cycles. The largest one, occurring during a startup in the fourth operating cycle, produced an increase of 1.38 ppm of particulates and caused Co-58 to spike to 0.75 uCl/gm.
Controlled crud removal activities were performed by the licensee during refueling operations. Aftct initial addition of boric acid, the crud was oxidized by injecting hydrogen peroxide. Released corrosion products were then removed from the system by circulating the coolant through the chernical and volume control system demineralizers.
The efficiency of the operation was monitored by the Co 58 activities in the coolant immediately after oxidation and at the end of the operation. By optimizing different steps in the operation, the licensee succeeded in significantly improving its efficiency.
Refur.L5 RefueLQ Refuel 7 Refuel 8 Curies of Co 58 Removed 602 1047 557 1098 Pounds of Ni Removed 4.8 12.6 8.4 9.3 The chemical and volume control system demineralizers remove impurities (mainly corrosion products) during normal plant operation and during shutdown, when special procedures induce crud release to the coolant. The reported decontamination factors, which were based on total activity measurements, varied over a very wide range due to the difficulty in measuring very low levelt of these activities. However, looking at all tt.e available data, the team found that the ovs rail performance was satisfactory with few demineralizer failures reported. The team did not identify any demineralizer failures, in the data reviewed, due to resin overheating. The few domineralizer failures that occurred were due to resin exhaustion for certain anions, sulfate ions released due to improper procedure used in replacing the resin, and, in a few cases, mixed bed domineralimrs experienced high pressure drops across their beds. It was also noticed that to improve efficiency of the purification system, licensee chemists downstzed purification filters from the originally used 30 microns to 0.1 micron and increased the chemical and volume control system flow rate from 75 gpm (284 Lpm), used in the earlier operating cycles, to its maximum value of 120 gpm (454 Lpm).
In an attempt to explain the occurrence of the axial offset anomaly, licensee chemists initinted an effort to characterize the crud deposited on the fuel rods. Samples of this crud were removed from fuel rods after the sixth operating cycle, which had experienced
.
.
-
-
-
_.
..
.-
. - - _ -
.
16-an axial offset of -12 percent, and the seventh operating cycle, which did not experience any axial offset anomaly. The semples of crud were obtained by scrapping preselected areas on six fuel bundles. The chernists and engineers found that most of the crud was deposited on the upper portions t he fuel roads and were inicker on twice bumt assemblies than on once burnt assemblies.
The scrapped crud was fransferred to a laboratory where it was weighed and its composition determined. No significant differences in crud composition were found in samples from the sixth and seventh operating cycles, although the nickel to-iron (Ni/Fe)
ratios were 0.84 and 0.7 in the sixth and seventh operating cycle, respectively. Higher Ni/Fe ratios are typical for the cores exhibiting an axial offset anomaly.
In the sixth operating cycle, crud weight per unit area in the upper portion of once burnt fuel was more than three times the crud weight per unit area on the corresponding fuelin the seventh operating cycle. This ratio increased to seven for the twice bumt fuel assernolies.
The team found that the thickness of crud was an important parameter in evaluating the capability to absorb lithium borate, Unfortunately, the thickness could not be measured directly, and only indirect estimates from the crud weight per unit area were made by using a crud density of 1.2 g/cnf. Since crud densities may vary for different crud deposits, the thickness of crud calculated in this way can be only considered as a very rough estimate. Also, the calculated thickness represented an average over the scraped area. The actual crud thickness at specific points could have been much higher, A comparison of the amounts of crud on the upper ponion of fuel rods in the sixth and seventh operating cycles are given below.
S.idti Seventh Operatino Cvele Ooeratin2_CJ&la Once Bumt Fuel Tw',ce B.urnt Fuel Once Burnt Fuel 112 i 60 239*59 30128 t
e d
Thickness (microns)
9i5 20 i 5 3*2 In an attempt to reduce the axial offset, licensee chemists performed certain manipulations of reactor coolant chemistry aimed at increasing the solubility of lithium borate and, thereby, reducing the amount in the crud deposits. Two such operational maneuvers were made.
The first one consisted of reducing reactor power from 70 to 30 percent. This reduction caused the boiling of reactor coolant in the crud deposits in the upper portion of the core to decrease and cooled the temperature of the reactor coolant, Both effects !ncreased the dissolution rate of lithium borate because of its higher solubility at lower
O o
17-temperatures. Release of soluble lithium horate from the crud deposits was monitored by observing the increase of its concentration in the coolant. Hnwever, when power was returned to 70 percent, most of the dissolved lithium borate redeposited.
The second maneuver consisted of lowering the lithium borate concentration and therefore dropping pH, from 7.1 to 6.9. Lower concentration of lithium borate in the coolant increased the dissolution of lithium borate, but tha effect on the axial offset was small. The results of these manipulations of coolant chemistry suggested to licensee engineers that the improvement of axial offset through modifications of chemistry was not very effective and otner methods, more related to core design, should be attempted.
c.
Conclusions The actions taken in response to the axial offset anomaly were conservative and in accordance with regulatory requirementi,. Wh!!e licensee engineers' actions supported their conclusions regarding the cause of the axial offset condition, no conclusions were reached by the team with respect to the root cause determined by the licensee and vendor. A weakness was identihed with respect to certain engineering staff knowledge of rod swap methodology.
E.ManagenlentMeitlings X1 Exit Moeting Summary The team presented the inspection results to members of licensee management during a telephonic exit on December 11,1997. The licensee's management representatives acknowledged the findings presented.
The team asked the licensee's representatives whether any materials examined during the inspection should be considered proprietary. While some materials were considered to be proprietary, no proprietary information is included in the report.
.
O
,
o ATTACHMENT SUPPLEMENTAL INFORMATION PARTIAL LIST OF PERSONS CONTACTED Licensen r
D. Bettenhausen, Supervising Engineer, Quality Assurance H. Bono, Supervising Engineer, Quality Assurance and Regulatory Support R. Davis, Engineer, Quality Assurance R. Irw;n, Supervising Engineer, Licensing and Fuels J. Knaup, Reactor Engineer, Nuclear Engineering J. Laux, Manager, Quality Assurance T. Michalek, Core Design Engineer, Licensing and Fuels J. Moose, Senior Core Design Engineer, Licensing and Fuels T. Moter, Supervising Engineer, Nuclear Engineering C. Nastund, Manager Nuclear Engineering D Shafer, Supervising Engineer, Licensing and Fuels W. Witt, Superintendent, Nuclear Engineering
,
Westinchouse B. Johnson, Core Design Manager J. Secker, Senior Core Design Engineer NBC D. Powers, Chief, Maintenance Branch B. Weistreich, Project Manager, Office of Nuclear Reactor Regulation INSPECTION PROCEDURES USED IP 92903 Followup - Engineering DOCUMENTS REVIEWED Procedure OSP-SF 00001," Shutdown Margin Calculation," Revision 18 Recuests for Resoluti2D No.17096, " Revised Rodworth (sic] Uncertainty for Cycle 9," Revision K No.17096, * Revised Rodworth [ sic] Uncertainty for Cycle 9,* Revision N No.17096," Revised Safety Evaluation for Cycle AO Anomaly," Revision P Ho.17096, * Revised Rodworth (sic] Uncertainty for Cycle 9," Revision Q No.17096," Revision of Cycle 9 Rod Insertion Limits," Revision R i
__
__
_ _ _
_
_
.
.
.
..._-._.
O
.
C 2-Calculations NFDC 92-005, * Simulate 3 Cycle 6 Startup aredictions," Revisien 0 NFDC 92 007, * Simulate 3 Cycle 5 Adjusted Axial Offset Model,' Revision 0 i
NFDC 93-005, * Simulate 3 Cycle 7 Stcrtup Predictions,' Revision 0 NFDC 95-008, " Cycle 8 Startup Predictions," Revision 0 NFDC 96 003, 'Cydes 18 Simulate 3 Version 4.02 Models," Revision 0 NFDC 96-024, 'Cy:le 9 Startup Predictions," January 9,1997 NFDC 97-003, ' Cycle 9 ESP Correction Factors for April /May 1997," Revision 0 NFDC 97 013," Revised Cycle 6 and Cycle 7 Rod Swap Calculations,' Revision 0 ARCP 7301, " Calculation of Nuclear Fuel O'
Quahlv ' 13urance Deoartment Audit Reoorts AP 08, " Technical Specification and Reload Design Control Sections," May 14,1993 SPw J39,"Burnup Distribution Effects on INCORE3D," May 26,1993 SQL 04100, " Union Electric Audit of Westinghouse - NTD " July 22,1994 Ippical Recort$
EPRl:
TR 108320, * Root Cause Investigation of Axial Power Offset Anomaly," June 1997 Westinghouse:
WCAP 7308-L, " Evaluation of Nuclea; Hot Channel Factor Uncertainties," April 1969 WCAP 7308-L ' Update to WCAP 7308-L, Evaluation of Nuclear Hot Channel Factor Uncertainties," March 1984 WCAP 9217,'Results of the Control Rod Worth Program *
WCAP 10842, " Qualification of the PHOENIX /POLCA Nuclear Design and Analysis Program for Boiling Water Reactors," June 1985
$NCAP-10966-P-A, "ANC-A Westinghouse Advanced Nodal Computer Code," December 1985 WCAP-13061 P-As, * Westinghouse Fuel Assembly Reconstitution Evaluation Methodology,"
July 1933 WCAP 14900," Root Cause of Axial Power Offset Anomaly,' May 1997
_
e
. -
O 3-CALLAWAY OPERATING CYCLES First December 19,1981 - February 28,1986 Second April 18,1986 September 10,1987 Third November 15,1987 March 31,1989 Fourth May 23,1989 - September 20,1990
>
Fifth November 19,1990 March 20,1992 Sixth May 18,1992 - September 30,1993 Seventh November 22,1993 - March 25,1995 Eighth May 11,1995 - October 12,1996 Ninth November 13,1996 - April 4,1998
,
b t
A
_ _ _ _ _
_ _ _ _. -. _ _ _ _ _. _. - _. _ _ _