ML17310A565
| ML17310A565 | |
| Person / Time | |
|---|---|
| Site: | Palo Verde  |
| Issue date: | 08/19/1993 |
| From: | Office of Nuclear Reactor Regulation |
| To: | |
| Shared Package | |
| ML17310A564 | List: |
| References | |
| NUDOCS 9309030066 | |
| Download: ML17310A565 (53) | |
Text
pe RECT~
Wp
+
0 Cy A0 4~*yW UNITED STATES NUCLEAR REGULATORY COMMISSION WASHINGTON, D.C. 20555-0001 Enclosure 1
SAFE Y EVALUATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION RELATED TO THE STARTUP AND OPERATION OF A 0 ERDE NUCLEAR GENERA ING ST TION UNIT 2 FOLLOWING THE STEAM GENERATOR TUBE RUPTURE OF MARCH 14 1993 ARIZONA PUBLIC SERVICE COMPANY DOCKET NO. 50-529 Executive Summary On March 14,
- 1993, a steam generator tube ruptured at Palo Verde Nuclear Generating Station Unit 2 while the unit was at 98 percent power.
This led to a reactor trip and cold shutdown for a steam generator tube inspection and a
refueling outage (the refueling outage was originally scheduled to begin on March 21, 1993).
Extensive eddy-current testing was conducted on the steam generator tubes in both steam generators.
Each steam generator has 11,012 tubes.
All of the unplugged tubes in each steam generator were examined with the bobbin probe.
In addition, about 3800 tubes in each steam generator were examined with the motorized rotating pancake coil probe.
Host of the eddy-current indications found were axial cracks and were contained within an'rc pattern on the tube sheet
- map, high in the steam generator where thermal-hydraulic analyses indicated that film boiling and dryout, and therefore chemical deposition, would be likely to occur.
The presence of deposits was confirmed by visual examination.
Eight tubes were removed from Steam Generator No.
2 for testing and metallurgical examination.
The eddy-current inspection resulted in plugging 74 and 175 tubes in Steam Generator No.
1 and No. 2, respectively.
The rupture occurred high in the steam generator in the free span and the root cause was due to intergranular attack and stress corrosion cracking (IGSCC) from the outside diameter of the tube, probably stimulated by the deposited chemicals and perhaps also assisted by the inadvertent introduction of demineralizer resin during operation.
The licensee has taken or plans to take several corrective actions to improve steam generator chemistry control to reduce the potential for another steam generator tube to degrade to the point of rupture.
The staff, however, believes it is difficult to quantify the overall effects of these changes in preventing future steam generator tube ruptures.
The significance of this steam generator tube degradation is the possibility that the resultant cracks may propagate through-wall in one cycle of lp3pqppppp6 93p819 t PDR ADOCK"05000528,,
P
'DR~
'k I
operation.
For Palo Verde, a fuel cycle is normally 15 months of power operation.
Applying the guidance contained in Regulatory Guide
- 1. 121, "Bases for Plugging Degraded PWR Steam Generator Tubes," the licensee proposes to operate Unit 2 for six months at power and then shut down for inspection, as opposed to the normal 15-month period of operation.
The staff believes that substantial uncertainty exists in the licensee's estimate of crack detection threshold with eddy current testing, crack growth rate, and burst strength as a function of crack size.
In view of these uncertainties, it cannot be assured that the burst strength margins specified in Regulatory Guide 1. 121 will be met for all tubes for the duration of the proposed six-month operating cycle.
Even with consideration of these uncertainties, the staff believes that a steam generator tube rupture (SGTR) is unlikely to occur as an initiating event.
- However, some tubes may become sufficiently degraded towards the end of the proposed six-month operating period that they could fail if a major secondary system depressurization should occur.
- The licensee has proposed enhanced primary-to-secondary leakage monitoring and will shut down the plant should leakage in a steam generator exceed 50 gallons per day.
While this enhanced leakage monitoring increases the likelihood that a degraded steam generator tube which leaks may be detected prior to a steam generator tube rupture occurring, the form of cracking observed in the Unit 2 steam generators may not always provide significant precursor leakage prior to rupture.
The licensee has improved its emergency operating procedures to eliminate problems found during the Harch 14, 1993, rupture event (as discussed in NRC Information Notice 93-56) and provided extensive additional training to all crews for steam generator tube rupture.
The licensee has also reduced the variable overpower trip setpoint and is instituting more restrictive limits on primary coolant activity to limit the potential consequences should a
SGTR event occur.
Given the potential degraded conditions of the steam generator tubes at the end of the proposed operating period, the consequences of SGTRs induced by a major secondary system depressurization were assessed.
It was concluded that sufficient refueling water tank inventory is available for the operator to maintain core cooling while depressurizing and cooling down the reactor coolant system to terminate the event.
Realistic calculations of potential offsite doses and control room operator doses were presented.
These realistic calculations met the acceptance criteria of 10 CFR Part
- 100, and 10 CFR Part 50, General Design Criterion 19, respectively.
Given the uncertainties that exist in evaluating tube integrity near the end of the proposed six-month operating period, the staff evaluated the risk associated with operating with potentially degraded tubes.
The staff estimates that the overall (integrated) core damage probability and associated containment
- bypass, for SGTR events is about 1.5xl0 over the proposed operating period.
The staff has concluded that operation of Palo Verde Unit 2 for the proposed operating interval of six months does not pose an undue risk to the public health and safety.
This conclusion is based on the numerous licensee actions
4
to address the root cause of the Unit 2 SGTR event to minimize the potential for future SGTR events, actions taken to enhance primary-to-secondary leakage monitoring to provide early indication of tube degradation, improvements in emergency operating procedures for SGTR events, and the reduction in the variable overpower trip setpoint and limits on primary coolant activity to limit potential consequences should an SGTR occur.
Further, while uncertainty exists in estimating the structural integrity of the steam generator tubes toward the end of the proposed six-month operating period, the risk evaluation confirms that the potential risk to the public is small.
1.
INTRODUCTION On March 14,
- 1993, a steam generator tube ruptured in Palo Verde Unit 2, Steam Generator No.
2 while the unit was at 98X power.
In response to the rapidly falling pressurizer level caused by the flow through the ruptured tube, the operators shut the reactor down (reactor trip) and entered the unit's emergency procedures to mitigate the event.
Steam Generator No.
2 was isolated in about three hours and the unit taken to cold shutdown thereafter.
The tube ruptured about one week prior to a scheduled refueling outage that would have ended a planned 15-month period of operation.
The ruptured tube was identified as R117C144 (row 117, column 144 on the tube sheet map), near the edge of the tube bundle.
The rupture occurred high in the tube in the free span between tube supports designated as "OSH" (the eighth tube support above the tube sheet on the hot leg side of the steam generator) and the "09H" support.
The rupture was about 2-inches long, axially oriented, and exhibited a
"fish-mouth" appearance.
It is estimated that the initial flow through the rupture was 240 gallons per minute (gpm)
(900 liters per minute).
Shortly after the event, the NRC dispatched an Augmented Inspection Team (AIT) to investigate the circumstances surrounding the event and the licensee's response leading to the cold shutdown of the unit.
The AIT found that the licensee's response was generally adequate, but also found a number of areas where changes and improvements could be made.
One of these areas was the time taken to diagnose the event as a steam generator tube rupture under the emergency procedures which was judged to be slow.
This was caused by the diagnostic procedure which did not consider either prior indications of radioactivity (the steam line monitors alarmed immediately after the tube ruptured) or increasing trends (the condenser air ejector monitor readings increased but did not alarm because the alarm point was set too high and the monitor was not working properly).
The existing diagnostic procedure required an alarm to be present on either the main steam line radiation monitor or the steam generator blowdown radiation monitor.
Since neither alarm was present, the diagnosis of a steam generator tube rupture could not be made under the licensee's emergency procedures, and therefore the optimal procedure for steam generator tube rupture could not be used.
The slower, more lengthy functional recovery procedures had to be used which delayed the diagnosis and isolation of the affected steam generator.
Improvements which the licensee has made in Emergency Operating Procedures (EOPs) and in procedures for determination of
1
primary-to-secondary (steam generator) leakage rates are discussed later in this Safety Evaluation.
Also, in response to this event, the licensee formed a special task force consisting of both licensee employees and outside industry experts to determine the root cause of the tube failure, to inspect the steam generator
- tubes, and to establish a basis for future operation.
The licensee has proposed to operate Unit 2 for six months and then shut down to perform steam generator tube inspections.
The basis for the proposed operating period and the NRC staff's evaluation of that basis is presented below.
This Safety Evaluation is organized as follows:
Section 2 describes the root cause for the SGTR which occurred and the licensee's actions to minimize future occurrences.
The eddy current inspection performed to detect and plug defects related to the root cause is provided in Section 3.
Given the nature of the root cause,-
and limitations of the eddy current inspection, an assessment of the tube integrity at the end of the proposed operating period is provided in Section 4.
The licensee has also proposed additional operating constraints.
Section 5
discusses the enhanced primary-to-secondary leakage monitoring program and more restrictive leakage limits for continued operation.
Section 6 provides a
safety assessment of improvements in the SGTR emergency operating procedures, the reduction in the variable overpower trip setpoint, and reduced limits on primary coolant system activity.
A risk perspective associated with operation with potentially degraded tubes is provided in Section 7.
A summary of the staff's conclusions is provided in Section 8.
2.
ROOT CAUSE OF FAILURE Palo Verde Nuclear Generating Station (Palo Verde) Unit 2 has two Combustion Engineering (CE) System 80 steam generators.
The CE System 80 steam generators are recirculating U-tube steam generators which contain 11,012 high temperature mill-annealed Alloy 600 tubes with an outside diameter of 0.750 inch (19 mm) and a nominal tube wall thickness of 0.042 inch (1.07 mm).
Palo Verde Unit 2 experienced a tube rupture in Steam Generator No.
2 which resulted in primary-to-secondary leakage of approximately 240 gallons per minute (900 liters per minute) on March 14, 1993.
Primary-to-secondary leakage prior to the event was approximately 20 gallons per day (76 liters per day).
The ruptured tube, Row 117 Column 144 (R117C144),
was identified by eddy current and visual examination to contain an axial fish-mouth rupture of 2 to 2.5 inches (5.1 to 6.4 cm) in length with a total crack length of approximately 8 inches (20 cm).
The rupture occurred just below the ninth partial eggcrate support (09H).
Inspection of this tube with a bobbin coil probe in the previous refueling outage (approximately 15 months earlier) did not reveal any eddy current indications.
The licensee removed eight tubes from Steam Generator No.
2 for metallurgical examination in order to address the root cause of the failure, to assess eddy current inspection capabilities with regard to detection and sizing of flaws, and to address the nature of the cracking that has been observed.
The licensee reported that visual inspection of several tube sections under a
low power stereo microscope showed visual evidence of ridge deposit formation at free span locations.
Long axial free span cracks were found under the free span ridge deposits that are believed to have been formed as a result of reduced tube clearance and from the propensity of deposits to collect at crevices in the upper part of the tube bundle.
Free span ridge deposits were determined to be as thick as four mils (after pulling) on the outside diameter surface of the tube;
- whereas, normal scale deposits were found to be on the order of two mils thick.
Chemical analysis of the deposits showed a trend for increased concentration of normal deposit constituents and contaminants as the tube bundle height increased.
Based on the review of the deposit data, the licensee concluded that the concentration of the deposits and contaminants could facilitate the growth of outside-diameter-initiated intergranular attack (IGA) and intergranular stress corrosion cracking (IGSCC).
The licensee concluded based on the microanalytical analysis of the tube surface and crack surface films that the crack environment was alkaline (mildly caustic) with the presence of sulfates.
The conclusion that the IGA/IGSCC occurred in an alkaline-to-caustic environment was based, in part, on.pulled tube analyses that showed chromium depletion at the crack tips.
Such depletion would only be expected to occur in an alkaline environment.
Examination of the pulled tubes revealed the presence of sulfates and reduced sulfur on the crack surfaces, and the licensee concluded that the sulfur species contributed to the degree of IGA and IGSCC in the alkaline-to-caustic environment.
The evidence of some areas showing nickel depletion support the licensee's conclusion as reduced sulfur would precipitate nickel into solution.
The licensee reported that a small amount of lead was detected in the crack surface films but that it was not considered to have been a
significant factor in the tube cracking.
Metallic copper was also detected in high concentrations in both the deposits and crack surfaces associated with the upper tube bundle.
Although the oxidizing influence of copper was not detected, it is believed by the licensee to have had an influence on the rate of IGA attack.
Several tube sections were identified during the metallurgical examination to contain scratches on the outside diameter of the tube's surface.
Scratched areas could result in tube surface cold working with resultant surface tensile residual stresses.
The licensee believes that cold-worked areas can provide a
preferential site for IGA development and if thick deposits are present with these cold-worked areas, a concentrating chemical environment could be formed which could lead to more rapid crack initiation.
The source of the cold-work scratches has not been determined.
Examination of the lowest portion of the ruptured tube did not reveal the presence of cold work via scratched areas;
- however, the licensee believes the similarities to other tubes indicate the likelihood that the burst tube may have contained similar scratched areas and consequential cold working.
IGA and IGSCC were found in non-cold-worked areas on several tube sections;
- however, the depth of attack was not as severe as areas associated with cold working and ridge deposits.
Other scratches and grooves believed to be associated with the tube installation process were found under normal tube scale and the resultant IGA attack was less severe (i.e., approximately 6 mils deep).
I
Hicrostructural characterization of the ruptured tube,
- R117C144, revealed a
microstructure that would not be expected for a typical high temperature mill-annealed Alloy 600 tube.
The licensee believes that the less-than-favorable microstructure in the ruptured tube was probably a combination of the heat treatment/cooling process and a low carbon content.
A poor microstructure can reduce a material's resistance to stress corrosion cracking.
In summary, as a result of the metallurgical examination of the pulled tubes, the licensee has concluded that tube R117C144 ruptured due to IGA/IGSCC attack in an alkaline-to-caustic environment associated with free span deposits that contain sulfates.
Detection of cold working due to scratched areas on other tubes suggested to the licensee that a cold-worked surface area may have been present on the ruptured tube (although it was not observed since only the lower portion of the crack could be removed from the steam generator ) which when combined with the free span ridge deposits led to preferred IGA and subsequent early crack initiation.
The observed microstructure of the ruptured tube was less than ideal and may have reduced the resistance to IGSCC;
- however, the licensee believes this to be a secondary effect.
One of the contributing factors to the steam generator tube rupture was increased sulfate levels.
The increased sulfate levels were identified in the crack oxide analysis.
The licensee has identified two instances of abnormal resin intrusion into the steam generators of Unit 2 which may have contributed to the increased sulfate levels within the steam generators.
The first resin intrusion occurred in July 1991 but did not result in elevated sulfate concentrations in the bulk water of the steam generators;
- whereas, during the second resin intrusion in January 1992 the sulfate levels rose to approximately five times the levels in the EPRI secondary water chemistry guidelines.
In addition to the increased sulfate levels from the resin intrusion, resin fines were observed in this outage during a visual inspection of both steam generators.
Inspections of the service vessels containing resin during the current refueling outage revealed that several vessels had damaged retention elements.
The licensee has repaired the condensate demineralizer retention elements, as necessary, to ensure that they are in good working order prior to plant restart.
To further minimize the amount of contaminants (including sulfates) in the steam generators, they were drained and refilled.
The licensee has determined that the IGA/IGSCC occurred as a result of alkaline-to-caustic crevice conditions as evidenced by metallurgical examination of several pulled tubes.
A review of hideout return data over several operating cycles in conjunction with estimation of crevice pH by computer modeling indicated that local crevice conditions were highly alkaline to caustic.
In addition, the licensee believes a molar ratio (i.e., ratio of sodium to chloride) imbalance may have led to an aggressive chemical environment within the free span crevice.
The licensee currently plans to control the molar ratio after plant restart to minimize the potential of developing additional caustic crevices.
The formation of free span crevices was also identified by the licensee to be a contributing factor to the steam generator tube rupture.
The formation of a free span crevice was evidenced by video inspection which revealed bridging deposits in locations where the tube spacing appeared to be reduced.
In
- addition, evidence of tube bowing/bending was found on a few tube specimens during the metallurgical examination.
Eddy current testing also identified that several of the linear deposits occurred in tube pairs providing further evidence of the potential for bridging deposits.
These observations have led the licensee to believe that reduced tube spacing has occurred and that the potential for free span crevices exists in the upper portion of the tube bundle.
As discussed in Section 3, a thermal-hydraulic analysis indicated that the upper portion of the tube bundle is susceptible to contaminant concentration.
In summary, evidence indicates that the rupture of R117C144 was due to IGA/IGSCC which occurred as a result of tube-to-tube crevice formation.
The crevice and the heat flux across the tube lead to an aggressive environment under a ridge deposit.
As a consequence, a long deep crack initiated under the ridge deposit, leading to the loss of structural integrity under normal operating conditions.
Several additional factors such as increased sulfate levels due to resin intrusion, likelihood of cold working due to surface scratches, a less than standard microstructure, and increased susceptibility of contaminant concentration in the upper region of the tube bundle may have contributed to the loss of structural integrity of the tube.
To address the root cause of failure, the licensee has taken or plans to take the following corrective actions:
~
The condensate demineralizer retention elements were inspected and repaired as necessary.
~
The steam generator was drained and refilled to minimize contaminant loading.
~
Defective steam generator tubes have been removed from service (i.e.,
plugged).
~
The cation to anion molar ratio will be controlled to minimize the potential for forming alkaline-to-caustic crevices.
~
A resin monitoring program will be implemented to help in identification of a resin intrusion.
~
pH will be optimized to minimize the amount of iron transport into the steam generator.
Iron is the main constituent of the deposits.
~
Elevated levels of hydrazine will be used so as to minimize the electrochemical potential which should increase the tube's resistance to IGA/IGSCC.
~
The steam generator blowdown efficiency will be controlled to control steam generator contaminant levels and maintain proper molar ratio control.
~
Periodic power reductions will be performed, as necessary, to dissolve contaminants such as sodium, sulfate, and chloride.
Power reductions
will collapse the film boiling surface in the high quality region of the tube bundle wetting the previously dried out area.
The staff notes that the licensee has identified several plausible causal factors that may have led to the steam generator tube rupture.
The staff also notes that the licensee's proposed actions are aimed at reducing the potential for another tube to degrade to the point of rupture.
The staff,
- however, believes that it is difficult to quantify the overall effects of these changes and that appropriate conservatism should be taken when calculating an appropriate operating interval for the next cycle.
The staff also notes that the licensee has not taken any action during this outage to remove the deposits from the tubes (however, the licensee has drained and refilled the steam generator in order to minimize contaminant loading).
3.
EDDY CURRENT TESTING INSPECTION During the inservice inspection of the steam generator tubing, the licensee identified and characterized various forms of steam generator tube degradation by eddy current testing.
The eddy current examination primarily involved the use of two different types of eddy current test probes:
a bobbin coil probe and a motorized rotating pancake coil (MRPC) probe.
The bobbin coil probe was used to examine 100X of the unplugged tubes in both steam generators;
- whereas, the MRPC probe was used to provide better detection/characterization capability in examining specific areas of the steam generator tubing.
In addition to the bobbin coil and MRPC probe, alternative technologies and probes were investigated by the licensee during the refueling outage;
- however, the licensee primarily relied on the bobbin coil and MRPC probe for examining the tubing.
Alternative technologies/probes investigated included: (I) a rotating field eddy current probe, (2) a high resolution bobbin probe, (3) a magnetic indexing referencing
- probe, (4) a larger diameter bobbin coil probe, and (5) an ultrasonic testing probe.
The licensee modified the original inspection scope several times during the outage as additional information was gathered.
The additional information was gathered not only from the steam generator tube inspections but also from the steam generator tube rupture root cause analysis.
The final inspection scope including the approximate number of tubes inspected and the extent of the inspection is provided below:
~
100X bobbin coil examination of both steam generators.
~
100X MRPC examination in the arc region (described below) between 08H and the first hot leg vertical support.
~
-3X random MRPC sampling outside the arc region between 08H (i.e., the eighth eggcrate support) and the first hot leg vertical strap in Steam Generator No. 2.
~
-20X MRPC sampling between 07H and 08H.
~
Various other MRPC examinations at and below 07H.
~
J 1
I
As a result of the inspections outlined above, the licensee identified axial cracking at the following locations:
~
Axial cracks at the 01H support and the tubesheet.
~
Axial cracks at supports from the 05H to the 09H support.
~
Axial free span cracks in the tube sections between 08H and 09H.
~
Axial cracks at the batwing supports.
~
Axial free span cracking between the batwing support and the vertical tangent to the U-bend.
~
Axial free span cracking at the horizontal tangent to the U-bend.
~
Axial cracks at the vertical strap supports.
The eddy current inspection of Palo Verde Unit 2 resulted in the plugging of 74 tubes in Steam Generator No. I and 175 tubes in Steam Generator No. 2.
The vast majority of the axial indications referenced above were observed between the OSH support and the first vertical strap on the hot leg side.
As a result of both bobbin coil and HRPC examination of the steam generator
- tubing, a general area where the majority of cracking occurred was identified.
This area of interest, referred to as the arc region, encompasses approximately 4000 tubes located in the outer periphery of the tube bundle from the OSH support to the first vertical strap on the hot leg side.
The licensee has found approximately 150 indications in approximately l20 tubes in the arc region of the Unit 2 steam generators.
Approximately one-half of these indications were located with only the HRPC probe; therefore, in order to enhance detection capabilities, the licensee performed HRPC examination of the tubes in the area of interest (i.e., the arc region).
As mentioned previously, the majority of the axial indications identified during the inspection occurred within a specific radial and axial extent within the tube bundle (i.e., the arc region).
Since a disproportionate number of indications were being found in a general area within the tube
- bundle, the licensee performed a thermal-hydraulic analysis to determine if there was a physical explanation for this phenomenon.
A physical explanation could provide the basis for determining an appropriate inspection scope.
The thermal-hydraulic analysis indicated that in the outer periphery of the tube bundle there was a higher propensity for solid and contaminant deposition from the 07H eggcrate support to the tube bend on the hot leg side.
The analysis indicated that a transition to film boiling was occurring in this region.
The licensee believes that the higher propensity for contaminant deposition combined with other factors led to the tube rupture.
Indications,
- however, were found in portions of the tube bundle that were not predicted by the thermal-hydraulic analysis to be highly susceptible to deposit accumulation.
Therefore, due to the lack of preciseness of the thermal-hydraulic model, the licensee had to rely on the eddy current inspection program to identify the full extent of the area of interest (i.e., the arc region).
The thermal-
hydraulic analysis, however, did provide insight into the potential for concentrating contaminants in a general area of the System 80 steam generator tube bundle.
Since the licensee had to rely on the eddy current inspection to identify the full extent of the area of interest, the inspection scope included a buffer zone (i.e.,
an area where no indications were found) around any indications near the edge of the general area identified by the thermal-hydraulic model.
Although the thermal-hydraulic model indicated that the increased propensity for contaminant deposition could occur as low as the 07H eggcrate
- support, inspections by bobbin coil did not detect extensive cracking in this area (i.e.,
between 07H and 08H).
The licensee also performed inspections in this area of the tube bundle with the MRPC probe and did not discover any indications that were not found with the bobbin coil.
To ensure that the problem was limited to within the arc region, MRPC inspections were performed both radially outside the area of interest (i.e.,
the arc region) and above and below the arc region.
The licensee identified axial indications outside the arc region at 01H (4 indications),
05H (1 indication),
07H (2 indications),
and in the horizontal run past the first vertical strap (1 indication).
The licensee performed MRPC probe inspections below 07H.
One axial indication, which was identified with the bobbin coil, was found at the 05H eggcrate
- support, and confirmed with a MRPC probe inspection.
No other axial indications were found below the 07H support except for approximately four axial indications at 01H.
A 3X sample of tubes, radially inward from the area of interest, were inspected with the MRPC probe from 08H to the first vertical strap on the hot leg side.
No axial indications were identified during this random sampling.
The licensee concluded that due to the lack of indications found during these MRPC inspections that any significant degradation was contained within the arc region.
In summary, the two primary objectives of the MRPC inspection scope were (1) to perform a thorough inspection of the area of the steam generator in which a
disproportionate number of axial indications had been detected (i.e., the arc region),
and (2) to perform sufficient MRPC inspections outside the area of interest to demonstrate that the probability of having defects below the bobbin coil detection threshold was minimal.
The staff concludes that the MRPC inspections performed within the arc region and the bobbin coil and random MRPC inspections outside the arc region provide reasonable assurance that the structurally significant cracks have been identified and subsequently repaired (i.e., plugged).
The staff notes that the licensee indicated an enhanced bobbin coil detection methodology was developed during the inspection.
4.
PROPOSED OPERATING INTERVAL Subsequent to plant restart, the licensee is proposing to operate the plant for six months (based on elapsed time at power at or above Mode 2) at which time the plant will be shut down for another steam generator tube inspection.
The intent of the proposed six-month operating interval is to ensure that the 10
I
tubes will retain adequate margins against burst in accordance with Regulatory Guide 1. 121.
The limiting burst criterion for Palo Verde Unit 2 is that the tubes must retain a margin of 1.4 against burst under postulated accident conditions.
The limiting postulated accident was considered to be a main steam line break (HSLB).
The associated differential pressure across the tubes was assumed to be 2400 psi, corresponding to the peak reactor coolant system (RCS) pressure at the start of an'HSLB during full power operation with a concurrent loss of offsite power.
The secondary system was assumed to be totally depressurized.
The proposed six-month operating interval is based on consideration of the following input parameters:
~
The depth of cracks remaining in service at the time of plant restart (i.e., cracks which are below the detection threshold of eddy current testing),
~
Crack growth rates, and
~
The maximum crack size which will satisfy the limiting burst pressure criterion in Regulatory Guide
- 1. 121.
The licensee submitted two separate approaches to support its proposed six-month operating interval:
a "deterministic" approach and a "statistical" approach.
The staff's evaluation below is based on the licensee's deterministic approach.
Details of the statistical approach were submitted only recently by letter dated July 30, 1993.
The staff could not complete its review of the statistical approach in the time available.
The staff's initial review of the statistical approach indicates that it is likely that additional information will be necessary from the licensee before the staff can reach any conclusions on this analysis.
The staff believes that a statistical approach can potentially be a useful tool for evaluating uncertainties and for providing a more realistic assessment of tube integrity margins than can sometimes be achieved through a "deterministic" approach.
Such an approach may also prove useful for purposes of evaluating future operating intervals for Palo Verde Unit 2 after the forthcoming steam generator inspection to be performed at the end of the proposed six-month operating interval.
For this
- reason, the staff plans to continue its review of the licensee's statistical approach.
- 4. 1 Crack Detection Threshold with Eddy Current Testing The proposed six-month operating interval is based on the assumption that the depth of the most limiting crack remaining in service at the time of plant restart will be just below the detection threshold at which a crack can be reliably detected.
The detection threshold was assumed to be 40X through-wall (T.W.) for portions of tubes which were inspected by NRPC probes and 50X T.W.
for tubes and portions of tubes which were only inspected by bobbin probes.
The licensee believes that the assumed detection thresholds are supported by the pulled tube data from Palo Verde.
The licensee reports that each of 9 cracks determined by destructive analysis to have maximum depths ranging
between 40 and 100X T.W., including 3 cracks with depths between 40 and 49X T.W., were detected in the field with HRPC.
None of three cracks with maximum depths less than 40X T.W. were detected by NRPC.
Regarding bobbin probes, the licensee reports'hat 4 of 6 cracks determined by destructive analysis to have maximum crack depths between 50 and 100X T.W. were detected in the field with bobbin probes.
The maximum depth for the two non-detected cracks were 61 and 57X T.W., respectively.
The average depth over the length of these cracks was 27 and 45X, respectively.
Only one of nine cracks with maximum depths less than 50X T.W. were detected in the field with bobbin probes.
The licensee considers burst pressure to be more a function of average depth than maximum depth.
Each of two cracks determined by destructive examination to have average crack depths in excess of 50X T.W. was detected in the field with bobbin probes.
In a recently published draft report dealing with interim plugging criteria (NUREG-1477, "Voltage-Based Interim Plugging Criteria for Steam Generator Tubes" ), the staff noted that there is considerable uncertainty over the probability of detection (POD) performance being achieved in the field for stress corrosion cracks.
These uncertainties
- stem, in part, from variabilities which exist in the field and in the laboratory with respect to crack morphologies, noise levels, inspection equipment, data acquisition and analysis procedures, and the performance of the data analysts.
The staff is examining the POD issue as part of on-going generic activities relating to industry proposals for alternate plugging criteria and for degradation specific management and as part of a cooperative international effort.
In the meantime the staff estimated in the draft report that the POD for cracks in excess of 80% T.W. was approximately 0.80 and that the POD for cracks less than 40% T.W. was much lower.
This estimate made no distinction between cracks of different rupture potential which varies for a given crack depth as a function of crack length and degree of segmentation.
Tubes with cracks that are not detectable are expected by the staff to be relatively short or highly segmented and thus to have burst strengths which substantially exceed the criteria of Regulatory Guide 1. 121.
This is true provided that the inspections are performed using the appropriate equipment and test and analysis procedures which the staff believes to have been the case during the recently completed inspection at Palo Verde Unit 2.
The licensee's assessment of the allowable operating interval is based on the allowable crack depth (consistent with Regulatory Guide 1. 121) for non-segmented cracks ranging to 1.4 inches in length.
The staff believes that the detection threshold values assumed by the licensee to be consistent with the Palo Verde pulled tube data and with the licensee's methodology for determining the operating interval.
However, the pulled tube data base for Palo Verde is limited and, thus, there is uncertainty associated with the licensee's detection threshold estimates which the staff believes to be on the order of 10X of the wall thickness based on its experience.
For the bobbin probe, this uncertainty may exceed 10X in the bend region of the tubing; however, this is not a concern for the proposed six-month operating interval since HRPC inspections have been performed for all tube bends in the region of the bundle which the staff believes to be of most concern.
12
4.2 Crack Growth Rates The licensee estimated potential crack growth rates based on eddy current results between the last two inspections.
The licensee considered the cumulative probability distribution of the bobbin depth measurements for each quantifiable indication found during the current outage inspection and which was confirmed to be a crack-like indication by HRPC.
For each location currently exhibiting an indication, the licensee reanalyzed the data from the previous inspection;
- however, none of these locations was found to exhibit a detectable bobbin indication.
The licensee assumed that cracks were present at these locations at the time of the previous inspection with depths ranging between 0 and 50K T.W. (i.e., the assumed detection threshold with the bobbin probe) with a most likely value of 1/2 of this range or 25K.
For purposes of estimating crack growth at each location during the last operating cycle, the licensee used a beginning of cycle (BOC) crack depth of 25X T.W. at locations where the end of cycle (EOC) indication exceeded 50X T.W.
For EOC indications less than 50% T.W., the licensee assumed the corresponding BOC depth to be 1/2 the EOC value.
Based on these assumptions, ninety-five percent of the growth rates were found to be less than 2 mils (4.6X T.W.) per month for indications at or above the OSH support.
The average depth for indications below the OSH support were found during the current outage inspection to be 2/3 that above the OSH support.
- Thus, the licensee considered the operative growth rate for indications below the OSH support to be 2/3 of the 2 mils/month value; or 1.4 mils/month.
The staff believes that there is significant uncertainty associated with the licensee's corrosion rate estimate.
Bobbin probe depth measurements for cracks tend to exhibit significant scatter, although the scatter tends to be random rather than systematic.
There is no direct information on the crack depth distribution at BOC and, in addition, there is no direct confirmation that the cracks found at EOC had, in fact, initiated at the time of the BOC.
Thus, the assumed distribution of crack depths at BOC is somewhat arbitrary.
However, the absence of direct information aside, it is the staff's judgement that many of the cracks found at EOC are likely to have initiated at or before BOC in view of the fact that three tubes with indications of axial cracks were found during the previous inspection of these steam generators (these three tubes were plugged at that time).
The presence of cracks above the detection threshold can reasonably be expected to be accompanied by cracks which have initiated but which are below the detection threshold.
It is also likely that the crack in the tube that ruptured was one of those that had already initiated at or before BOC rather than one which initiated after BOC.
The staff believes that a reasonably conservative approach is to assume that the crack in the ruptured tube initiated at BOC.
The staff estimates that the tube that ruptured had an effective crack depth of 92K T.W. at the time of rupture.
On this basis, the staff estimates a corrosion rate of 2.8 mils/month (6.7X/month) as a reasonably conservative estimate for the tube that ruptured.
This estimate illustrates the potential uncertainty in the
., licensee's estimate of 2 mils/month.
Laboratory data for mill-annealed Inconel 600 capsule C-ring specimens indicate crack growth rates as high as 7 mils/month in caustic (IOX NaOH) environments at 600 degrees F.
Data from pressurized capsules in a 10X NaOH 13
0 J
I
environment, which appear to the staff to be more directly applicable for assessing crack growth rates in steam generator
- tubes, indicate crack growth rates of 4.5 mils/month in a 10X NaOH environment at 600 degrees F.
The licensee repo} ts that the hoop stress in these capsules was approximately twice that in steam generator tubes under normal operating conditions.
The staff has not had the opportunity to review the material microstructures included in this laboratory data set as they compare to the relatively susceptible microstructures in some of the Palo Verde tubing.
However, it is the staff's judgement that this data is conservative compared to likely crack growth rates at Palo Verde Unit 2 because of the higher stresses and the chemical environment of the laboratory tests.
4.3 Maximum Allowable Crack Size/Regulatory Guide 1.121 The licensee has estimated the maximum allowable crack size (i.e.,
68X T.W.
for crack lengths not exceeding 1.4 inches long) using an expression developed by Framatome.
This expression relates tube hoop stress as a function of flaw depth and flaw length, tube geometry, and the yield and ultimate strength of the tube material.
The staff has not had the opportunity to review the detailed development of this expression.
However, burst strength data submitted by the licensee suggests that the expression provides more of a best estimate than a conservative estimate of tube burst strength.
The staff does not agree that the burst strength margins specified in Regulatory Guide 1. 121, by themselves, provide adequate justification for use of a best estimate burst strength correlation, either for purposes of determining plugging limits or for determining operating intervals without further justification.
For deterministic assessments such as that performed by the licensee, the staff believes that use of a best estimate burst expression is inappropriate unless it is accompanied by a statistical analysis demonstrating that use of the expression will ensure adequate tube integrity margins considering all potential sources of uncertainty.
This includes uncertainties in the HRPC and bobbin detection threshold estimates, uncertainties in crack growth rate estimates, as well as uncertainties associated with use of a best estimate burst expression.
Although the staff has been unable to review the licensee's statistical
- analysis, the analysis does not appear to satisfactorily address detection threshold uncertainties or burst strength uncertainties.
The licensee's estimate of the allowable crack depth (68X T.W.) is based on the assumption that cracks will not exceed 1.4 inches in length.
Given that cracks well in excess of 10 inches have been found on the pulled tube specimens and that the length of the fish-mouth on the tube that ruptured was 2-1/2 inches, the staff concludes that there is little basis to support the crack length assumption.
The licensee has also determined the allowable operating interval using an empirical burst strength expression developed by Battelle for the NRC staff in NUREG/CR-0718.
Use of the Battelle correlation in lieu of the Framatome correlation leads to an allowable depth estimate-of 60X for cracks greater than 1.6 inches long which, in turn, translates to an operating interval of 4.5 months.
Data submitted by the licensee shows that this expression may not bound all the data for short cracks, but that it does bound the data for long cracks.
The staff believes this estimate can be used to approximate the
,I
uncertainty associated with the licensee's estimate of burst strength as a
function of crack size.
4.4 Overall Assessment of Operating Interval It cannot be assured by a deterministic evaluation that the burst strength margins specified in Regulatory Guide 1.121 will be met for all tubes for the duration of the proposed six-month operating cycle in view of the significant uncertainties associated with the assumed crack detection thresholds, corrosion rates, and burst strength as a function of crack size.
Any attempt to conservatively bound all of these uncertainties for purposes of establishing an operating interval would in the staff's opinion lead to excessive conservatism and an excessively short operating interval.
The staff has estimated the potential for tube rupture during this six month period based on the following assumptions:
~
An initial crack depth at plant restart which is just below assumed detection threshold of 50N T.W. (for NRPC),
~
A crack growth rate of 2.8 mils/month, and
~
The Framatome expression for burst strength as a function of crack depth for long cracks (i.e.,
>2 inches).
The dominant uncertainty affecting the potential for tube rupture during the proposed six-month interval is the maximum length and depth to which cracks may grow.
The potential for having a small number (e.g.
one to three) tubes susceptible to failure will be governed by the tail of the distribution of crack growth rates.
For this reason, the staff has utilized conservative estimates of initial crack size and crack growth rate and a best estimate of burst strength as a function of crack depth for long cracks.
Based on these assumptions, the staff estimates that it will take about 4 months for a crack to grow to the point where it could potentially rupture under steam line break and 6-1/2 months before it could rupture during normal operation.
Based on this conservative assessment, a steam generator tube rupture is unlikely under normal operating conditions during the proposed six month operating interval.
Given the uncertainties that exist in evaluating tube integrity near the end of the proposed operating period, the staff evaluated the risk associated with operating with potentially degraded tubes.
This assessment is provided in Section 7.
5.
PRIMARY-TO-SECONDARY LEAKAGE MONITORING Unit 2 had been monitoring small primary-to-secondary leakage since July 1992.
Secondary system radiation monitors began to alarm in February
- 1993, although the licensee reports that these alarms were not long in duration but were consistently received during small reactor coolant system pressure transients, such as when shifting charging pumps.
In early March 1993, the primary-to-secondary leakage became large enough to calculate and track the leakage by the concentration of Iodine-131 in the secondary system.
The reported leak 15
0
rate during the two weeks prior to the rupture was nominally 20 gallons per day (gpd).
A post-event review of the leak rate data was performed by the licensee to validate the actual leak rate as well as to assess the various leak rate determination methodologies.
During this review, deficiencies were identified with the iodine method of leak detection and trending.
It is the licensee's current position that gas grab samples taken at the condenser vacuum exhaust will provide the most accurate estimation of the actual primary-to-secondary leak rate.
The primary-to-secondary leak rate monitoring program at Palo Verde consists of both continuous on-line monitoring capability and non-continuous (i.e.,
grab samples) monitoring.
The continuous on-line monitors in use at Palo Verde include the steam generator blowdown radiation monitors, the main steam line radiation monitors, and the condenser vacuum exhaust radiation monitor.
The non-continuous monitoring of primary-to-secondary leakage involves analyzing grab samples for radionuclides.
The grab samples can be obtained from various sampling points including the steam generator downcomer, the steam generator hot leg blowdown, and the condenser vacuum exhaust.
The steam generator downcomer and hot leg blowdown samples are liquid samples; the condenser vacuum exhaust sample is a gas sample.
Routine chemistry sampling of the secondary system (i.e., non-continuous monitoring) can, alert personnel to low level concentrations of activity that may be below the sensitivity of the continuous monitors.
The sensitivity of the different continuous on-line radiation monitors vary.
The steam generator blowdown radiation monitor which detects liquid radioactivity and the condenser vacuum exhaust radiation monitor which detects gaseous activity from noble gasses are sensitive to leakage on the order of 10 gpd.
The main steam line radiation monitors,
- however, are only sensitive to large amounts of leakage on the order 10 to 20 gpm (14 to 28 thousand gallons per day).
The actual sensitivities of all these monitors depend, in part, on the activity of the primary coolant system.
As a result of the steam generator tube rupture event, the licensee has made or will be making several enhancements to their primary-to-secondary leakage monitoring program including:
~
The steam generator blowdown radiation monitor sample point has been moved from the hot leg blowdown to the downcomer.
The downcomer is a
more concentrated stream (since it is not diluted with the incoming feedwater) and offers greater overall sensitivity to detect primary-to-secondary leakage.
~
The alert set points of the condenser vacuum exhaust radiation monitor have been reduced to provide an earlier alarm to the operators for increasing primary-to-secondary leakage.
The previous set points were based on off-site dose limits.
~
The preferred hier archy for measuring the primary-to-secondary leak rate will be a gas grab sample from the condenser vacuum exhaust analyzed for 16
noble gasses, a bulk water sample from the steam generator analyzed for
~
The monitoring frequency for measuring primary-to-secondary leakage will be increased as the leakage increases.
A formal evaluation for continued operation will be conducted when a
10 gallon-per-day leak rate increases by more than 50X in a 24-hour period, or a stable leak rate of 25 gallons per day is reached (i.e., evaluation of the rate of increase of the leakage will be performed).
~
An administrative leak rate limit of 50 gallons per day has been adopted and if exceeded will require an orderly plant shutdown.
~
Consideration is being given to the addition of Nitrogen-16 (N-16) monitors to the leak rate monitoring program.
The staff notes that N-16 monitors can enhance a leak rate monitoring program if properly designed and installed.
Potential advantages of N-16 monitors include a more rapid identification of the faulted steam generator during a primary-to-secondary leak event, an on-line confirmatory measurement to the condenser vacuum exhaust monitor readings, and a more timely notification of an increase in leak rate.
The staff concludes that the above-listed changes will enhance the licensee's leak rate monitoring program and thereby increases the likelihood that a
degraded steam generator tube which leaks may be detected and the plant shut down for inspection prior to the occurrence of a steam generator tube rupture.
However, the staff notes that the form of cracking observed in the Unit 2 steam generators may not always provide significant precursor leakage prior to rupture.
Therefore, appropriate training and other measures should be taken to ensur e proper handling of such an event.
6.
STEAM GENERATOR TUBE RUPTURE SGTR MITIGATION
- 6. 1 Emergency Operating Procedure (EOP)
Enhancements During the SGTR event which occurred at Palo Verde Unit 2 on March 14,
- 1993, the reactor operators experienced difficulties in entering SGTR procedures using the existing EOPs even though most operators had concluded early in the event that a
SGTR had probably occurred.
As a result, the event was identified about 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> after the SGTR initiation and the affected steam generator was isolated in about 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> after the event occurred.
The procedure steps in the Functional Recovery Procedures (FRPs) for mitigation of an SGTR were used to cooldown and depressurize the RCS and take the reactor to cold shutdown thereafter.
One of the causes of the time delay prior to effective mitigation of the SGTR event was the diagnostic logic tree which did not consider either past conditions of radiation monitors (the main steam line radiation monitors alarmed immediately after the SGTR occurred but cleared) or abnormal rising trends at the condenser vacuum exhaust radiation monitor (the condenser air ejector monitor readings increased).
The diagnostic logic tree required alarms to be present on either the main steam line radiation monitor or the steam generator blowdown radiation monitor at the time that the operators enter the step in the diagnostic logic tree which uses the Radiation 17
Honitor System Flowchart in order to enter the optimal procedure for SGTR.
The other cause of the time delay in isolating the ruptured steam generator was licensee management's conclusion that more rapid action was not required and therefore a shift change was made.
This conclusion was based on the fact that:
the FRPs permitted, but did not require, rapid isolation of the affected steam generator; existing plant conditions were stable; and a fresh operating crew was desirable.
In response to the lessons learned from the Palo Verde Unit 2 event, the licensee has modified the EOPs at Palo Verde for more effective diagnostic and mitigation of SGTR as follows:
- 1. Procedural changes have been made to ensure that the operators are able to quickly diagnose and mitigate an SGTR.
These changes will be in place at restart:
~
A note has been added to the Radiation Honitor System Flowchart for the operator to consider past and present conditions of radiation monitors.
~
A step has been added to the Radiation Honitor System Flowchart for the operator to consider high alarm or abnormal rising trend at the condenser vacuum exhaust radiation monitor as activity in the steam plant.
~
Additional checks have been added in both the SGTR Procedure and FRP for identifying a steam generator with a tube rupture.
The added checks were to look for differential feedflow between steam generators to maintain level.
~
Hade the checks for an SGTR within the FRP continuously applicable.
2.
Changes are being incorporated into the EOPs to enhance overall radiological management of an SGTR event.
These changes will be made effective October 31, 1993:
~
Transfer secondary system sump discharge promptly to liquid radwaste systems to minimize the spread of contamination following an SGTR.
~
Transfer the affected unit's auxiliary steam source to an unaffected unit to minimize the spread of contamination following an SGTR.
~
Hodified feedwater system operation to allow continued use of steam generator blowdown demineralizers throughout the recovery of an SGTR.
~
Hodified feedwater and condensate system operation to allow use of main feedwater and condensate throughout the recovery of an SGTR to minimize the amount of makeup water added to the secondary system.
18
I
~
Add directions that effluent releases must be monitored and maintained below limits set by the TSC during blowdown of the affected steam generator for level control during the cooldown.
In consideration of induced SGTRs as a result of a major secondary side depressurization, the licensee is also making two enhancements to the EOPs for more effective diagnostic and mitigation of this event scenario.
These enhancements will be made effective October 31, 1993:
~
The Safety Function status check of the Excess Steam Demand Procedure will be modified to guide the operators to exit to the FRP if either of the following indications of induced SGTRs from a secondary side depressurization exist:
If safety injection (SI) flow is needed to maintain pressurizer water level after SI throttling criteria have been met, or If radiation monitors and steam generator samples are not available, then direct radiation and contamination surveys of the steam generator release points will be initiated (steam generator safety valves and atmospheric steam dump valves).
~
The status checks within the FRP for indications of possible SGTR will be broadened to include the checks described immediately above.
The staff considers that the licensee's planned modifications to the EOPs at Palo Verde will provide sufficient enhancement for both diagnosis and mitigation of various SGTR scenarios including induced SGTRs from secondary side depressurization.
Specifically, the procedural enhancements at restart will ensure that a
SGTR event can be promptly identified during use of the diagnostic logic tree and will direct the operator to the SGTR procedure.
The changes made in the Radiation Monitor System Flowchart will continuously present steam plant radiation activity status when a large primary-to-secondary system leakage exists.
This information will significantly improve the effectiveness of identifying a single or multiple SGTR event using the diagnostic logic tree following a reactor trip.
During the Unit 2 SGTR event, the diagnostic logic tree did not guide the operators to enter the SGTR procedure promptly because the Radiation Monitor System Flowchart did not provide the needed confirmation for an SGTR event.
With the enhanced flowchart, the time delay in identifying an SGTR should be precluded and the operator is expected to take the actions necessary to mitigate an SGTR event promptly.
- Further, should the operator initially misdiagnose the event and enter the FRP, the enhancements to the FRP will allow the operator to more quickly identify the SGTR and cause the operator to enter Attachment 3 of the FRP which has similar steps for event mitigation as provided in the SGTR procedure.
These actions address the staff's concerns about operator actions during the March 14,
- 1993, tube rupture event outlined in NRC Information Notice 93-56.
The planned additional improvements to the EOPs to evaluate if HPSI flow is needed for pressurizer level control when the SI throttle criteria are met, to dispatch operators for local radiation surveys, and to look for differences in 19
S H
feedwater flow rates to the steam generators will provide alternative means of identifying an SGTR event.
- Thus, these changes are expected to further enhance the operator's ability to identify an SGTR and enter either the SGTR procedure or the SGTR portions of the FRP for effective mitigation of the event.
The staff has evaluated the delay, until October 31, 1993, in implementation of the procedural enhancement for induced SGTR events.
As discussed in Section 4,
even under conservative assessments of crack growth rate, the steam generator tubes will have adequate structural margin to withstand a main steam line break without rupture for approximately the first four months after restart.
Therefore, the staff considers the delay in implementation of the procedural enhancement for the induced SGTR events to be acceptable.
6.2.
Operator Training In response to the staff request, the licensee in its letter dated July 30, 1993, stated that the Palo Verde reactor operators have been briefed on the Unit 2 SGTR event and the changes that were made to the EOPs.
Also, the operators have participated in simulator scenarios during requalification training that closely parallel the Unit 2 event.
The simulator has been modified to more realistically model the plant, particularly the response of the Radiation Monitoring System to an SGTR.
For the modifications that will be made effective October 31, 1993, the operators will be trained on those changes during requalification training.
This will be completed before these procedure changes are effective.
6.3.
Safety Analysis of SGTR Events In response to the staff request, the licensee has performed an assessment of SGTR scenarios which included 1) a single SGTR event, and 2) induced single and multiple tube ruptures due to a major secondary side rapid depressurization transient such as would be caused by a main steam line break or main feedwater line break.
The assessment was performed using the approach described in Section 4 of draft NUREG-1477, "Voltage-Based Interim Plugging Criteria for Steam Generator Tubes."
In its assessment, the plant specific conditions at Palo Verde were used such as:
- 1) lack of a pressurizer power-operated relief valve,
- 2) lower pressure ECCS pumps
- and,
- 3) plant-specific EOPs.
In determining the most limiting event for dose consequences and loss of reactor coolant inventory outside containment, the licensee considered the multiple SGTR events caused by either a main steam line break (NSLB) or a main feedwater line break.
While the licensee concluded that a main feedwater line break would result in higher pressure differential across the steam generator tubes and thus produce higher leak rate across the steam generator
- tubes, there is a check valve in each of the main feedwater lines inside containment to prevent an unisolable break outside containment.
Therefore, an induced SGTR caused by a NSLB outside of the containment but upstream of the main steam line isolation valve is the most limiting event with respect to radiological consequences and concerns regarding Refueling Water Tank (RWT) inventory.
The analysis of induced SGTRs caused by an HSLB is considered to bound other possible induced SGTR scenarios caused by secondary side depressurization such as stuck open safety or atmospheric dump valves.
20
In a letter dated July 25, 1993, the licensee provided the results of its assessment of the radiological consequences for the SGTR scenarios analyzed.
To reduce the radiological consequences during these
- events, the licensee committed to make two changes:
~
reduce the variable overpower trip (VOPT) setpoint from IOX to 8X so that a reactor trip on a secondary side depressurization with offsite power available will occur earlier to preclude fuel damage during the transient, and
~
provide administrative limits on primary coolant iodine activity (0.6 pCi/gm, 12 pCi/gm-peak) to reduce calculated offsite dose.
The licensee did not reanalyze the single SGTR event.
"Rather, the licensee has confirmed that the procedural modifications made, as discussed in Section 6.1, will provide prompt identification of the event and direct appropriate operator actions which are bounded by the analysis in the Final Safety Analysis Report (FSAR).
Further, with the proposed administrative controls on primary coolant activity, the radiological consequences for a single steam generator tube rupture provided in the FSAR are bounding.
Detailed results of a best estimate computer calculation of a HSLB with three double-ended guillotine tube ruptures were provided.
The operator actions given in the Palo Verde EOPs as well as other plant-specific conditions were factored into this calculation.
This analysis provided plant system responses and radiological consequences during the event, and the results were also used to assess the adequacy of the RWT capacity to allow the reactor to be brought to stable depressurized condition following a postulated NSLB-induced multiple SGTR event.
The licensee's analysis of the event concluded that there was sufficient time for the operator to cooldown and depressurize the plant following a NSLB/SGTR event.
The analysis showed that entry conditions for shutdown cooling system operation would be reached in approximately 75 minutes.
Cooldown to cold shutdown conditions could be reached approximately four hours later.
The licensee's calculations assume that the operator maintains a more conservative (higher) subcooling margin to maximize the cooldown time.
At the end of the calculation (8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />),
the licensee's analysis indicates a continuing primary-to-secondary leakage rate of 300 gpm.
Total RWT inventory injected over this period is 256,500 gallons.
Since the Unit 2 Technical Specifications require a minimum volume of 600,000 gallons in the RWT, approximately 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br /> would be available to accomplish the cooldown to cold shutdown before the RWT would empty.
Therefore, it was concluded that adequate RWT inventory, and time for operator action, is available at Palo Verde to mitigate induced SGTR events caused by a secondary side depressurization.
In a letter dated July 25, 1993, the licensee analyzed the radiological consequences of the MSLB/SGTR event.
The licensee provided additional clarifying information on the analysis in a supplemental letter dated August 9, 1993.
The analysis calculated the radiation doses at the Exclusion Area Boundary (EAB), at the Low Population Zone (LPZ) boundary, and in the control room which could result from the event.
The analysis included 21
l I
II II 1
I
calculations for the pre-existing radioiodine spike case and the event-generated radioiodine spike case.
The licensee used three calculational methods.
First, the licensee used the method in the staff's Standard Review Plan (SRP),
which is a conservative method used for initial plant licensing.
- Second, the licensee used the SRP method substituting the reduced primary coolant radioiodine concentration limits which have been instituted at Palo Verde through administrative controls.
Third, the licensee calculated the doses using the new radioiodine concentration limits, realistic atmospheric dispersion
- values, and dose factors from International Council on Radiation Protection Report 30.
10 CFR Part 100 provides acceptance criteria for the doses analyzed during initial plant licensing at the EAB and LPZ for design basis accidents.
The dose in 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> at the EAB should not exceed 25 rem to the whole body or 300 rem to the thyroid.
The dose in 30 days at the boundary of the LPZ should not exceed 25 rem or 300 rem to the thyroid, For the event-generated radioiodine spike case, the staff reduces these acceptance criteria to 5 rem whole body or 30 rem thyroid.
10 CFR Part 50, General Design Criterion 19 provides acceptance criteria for doses to control room operators.
The dose in 30 days should not exceed 5 rem to the whole body or the equivalent to any part of the body (the staff uses 30 rem for the thyroid).
The licensee's calculated doses using the SRP method (first method) were higher than these acceptance criteria for the EAB and control room.
However, the doses calculated using the realistic methods were within these acceptance criteria for all'ases.
The staff reviewed the analysis presented by the licensee and concluded that the methods used by the licensee are acceptable.
The realistic doses were calculated using methods and realistic factors consistent with those employed by the staff in NUREG-1477.
Also, the staff performed check calculations of the bounding cases and obtained doses consistent with those calculated by the licensee.
Recognizing that there is substantial uncertainty in the prediction of tube integrity at the end of the operating interval, the staff concludes that a realistic assessment is appropriate for evaluating the dose consequences for induced SGTR scenarios.
Since the doses calculated with realistic parameters were less than the acceptance criteria discussed
- above, the staff concludes that the radiological consequences for induced SGTR events (up to 3 ruptured steam generator tubes) caused by a secondary side depressurization are acceptable.
7.
RISK ASSESSMENT The licensee has proposed to operate Palo Verde Unit 2 for six months and then shut down for inspection of the steam generator tubes.
As noted in Section 4, the staff has concluded it cannot be assured that all tubes will retain the margins against burst specified in Regulatory Guide 1.121 for the full duration of the proposed six-month operating interval in view of the significant uncertainties involving, in particular, crack detection thresholds associated with eddy current testing and crack growth rates.
To assess the potential risk implications of these uncertainties, the staff evaluated the potential for tube ruptures based on the conservative estimates of crack detection and growth rate as discussed in Section 4.
This section provides a
22
risk perspective of the proposed operation with potentially degraded steam generator tubes.
The staff's conservative assessment indicated that a steam generator tube rupture is unlikely under normal operating conditions during the proposed six-month operating interval.
For this reason, the staff concludes that the generic frequency of an SGTR as an initiating event, approximately 2x10 per reactor year, applies to the proposed operating interval of Unit 2.
The staff recognizes the licensee has implemented more restrictive primary-to-secondary leakage monitoring, as discussed in Section 5, which is expected to provide additional assurance that there is little potential for an SGTR over the proposed operating interval.
- However, due to the uncertainties in estimating the potential benefit of this monitoring, the staff has given no credit for this benefit in this risk assessment.
The probability of failing to successfully mitigate a steam generator tube rupture is estimated as 1.8xlO in the NUREG-1150 analysis for Surry, and
- 1. 13xlO in the Palo Verde IPE.
Based on the improved emergency operating procedures discussed ip Section 6, the licensee now estimates this failure probability as 7.7xlO The staff's assessment used a failure probability of 10
, which results in an estimated frequency of a core damage accident at Unit 2 due to an SGTR initiating event of 2xlO per reactor year.
Therefore, the overall (integrated) core damage probability due to an SGTR initiatinp event over the proposed six-month operating period is estimated to be 10 The staff notes that the Palo Verde IPE concluded that all SGTR core melt sequences would lead to a containment bypass.
The staff also evaluated the probability that a secondary side depressurization event (e.g.,
a break in a main steam line or a stuck-open safety valve) would induce rupture of one or more degraded tubes.
The frequency of the initiating steam line break or stuck-open SRV was estimated by the licensee as 7x10 per reactor year, based on safety valves that stuck at least 20K open.
However, this value apparently includes cases where the steam generator depressurization was not very severe.
Using the estimated values for a steam line break or stuck-open SRV from NUREG-1477, the staff estimated the initiating event frequency for a significant secondary side depressurization as 2x10 per reactor year.
Using the conservative estimates of crack detection and growth rates in Section 4, the staff estimates that there could be a tube vulnerable to rupture in the event of a significant secondary side depressurization after about 4 months of operation and that multiple tubes could be vulnerable to rupture after about 5 months of operation.
Because of the uncertainty in steam generator tube conditions, the staff's analysis assumed that a
significant secondary side depressurization would result in a single tube rupture during the fourth month and a multiple tube rupture (2-4 tubes) during the fifth and sixth months of the proposed operating period.
In NUREG-1477, the staff estimated the probability of the operator failing to depressurize and cool down the plant to mitigate induced SGTR scenarios caused by a significant secondary side depressurization.
The estimated failure probability for leak rates corresponding to the assumed numbers of tube 23
~
~ >
ruptures for Palo Verde Unit 2 was 10 Using these estimates>
the frequency of a core damage event caused by induced tube ruptures is 2x10 per reactor year.
Since the tubes would only be vulnerable to rupture for three months, the overall core dpmage probabi)ity for induced tube rupture events is estimated as 5x10 over the proposed operating period.
The above estimates of the integrated core damage probability are subject to significant uncertainty due to uncertainty in tube conditions during the proposed six-month operating period.
As noted above, the conditional tube rupture probability given a secondary side depressurization has been assumed to be 1.0 for the last three months of the proposed six-month operating period.
However, the staff believes that its assessment of the steam generator tube integrity is conservative.
Therefore, the staff concludes that the overall core damage probability, and associated containment
- bypass, due to SGTR events is about 1.5xlO during the proposed six months of operation and that the potential risk to the public is small.
8.
CONCLUSION The staff has concluded that the March 14,
- 1993, SGTR occurred high in the steam generator in the free span and was due to intergranular attack and stress corrosion cracking from the outside diameter of the tube, probably stimulated by the deposited chemicals and perhaps also assisted by the inadvertent introduction of demineralizer resin during the operating cycle.
The licensee has taken or plans to take several corrective actions to improve steam generator chemistry control to reduce the potential for another steam generator to degrade to the point of rupture.
The staff, however, believes it is difficult to quantify the overall effects of these changes in preventing future steam generator tube ruptures.
The significance of this steam generator tube degradation is the possibility that the resultant cracks may propagate through-wall in one cycle of operation.
For Palo Verde, a fuel cycle is normally 15 months of power operation.
Applying the guidance contained in Regulatory Guide 1. 121, "Bases for Plugging Degraded PWR Steam Generator Tubes," the licensee proposes to operate the unit for six months at power and then shut down for inspection, as opposed to the normal 15-month period of operation.
The staff believes that substantial uncertainty exists in the licensee's estimate of crack detection threshold with eddy current testing, crack growth rate, and burst strength as a function of crack size.
In view of these uncertainties, it cannot be assured that the burst strength margins specified in Regulatory Guide 1. 121 will be met for all tubes for the duration of the proposed operating cycle.
Even with consideration of these uncertainties, the staff believes that an SGTR is unlikely to occur as an initiating event.
- However, some tubes may become sufficiently degraded towards the end of the operating period that they could fail if a major secondary system depr essurization occurred.
The licensee has proposed enhanced primary-to-secondary leakage monitoring and will shut down the plant should leakage exceed 50 gallons per day.
While this enhanced leakage monitoring increases the likelihood that a degraded steam generator tube which leaks may be detected prior to the occurrence of a steam generator tube rupture, the form of cracking observed in the Unit 2 steam 24
J~
~
I
generators may not always provide significant precursor leakage prior to rupture.
The licensee has improved its emergency operating procedures to eliminate problems found during the Harch 14, 1993, rupture event (as discussed in NRC Information Notice 93-56) and provided extensive additional training to all crews for steam generator tube rupture.
The licensee has also reduced the variable overpower trip setpoint and is instituting more restrictive limits on primary coolant activity to limit the potential consequences should a
SGTR event occur.
Given the potential degraded conditions of the steam generator tubes at the end of the proposed operating period, the consequences of SGTRs induced by a major secondary side depressurization were assessed.
It was concluded that sufficient refueling water tank inventory is available for the operator to maintain core cooling while depressurizing and cooling down the reactor coolant system to terminate the event.
Realistic calculations of potential offsite doses and control room operator doses were presented.
These realistic calculations met the acceptance criteria of 10 CFR Part
- 100, and 10 CFR Part 50, General Design Criterion 19, respectively.
Given the uncertainties that. exist in evaluating tube integrity near the end of the proposed six-month operating period, the staff evaluated the risk associated with operating with potentially degraded tubes.
The staff estimates that the overall (integrated) core damage probability, and associated containment
- bypass, for SGTR events is about 1.5xl0 over the proposed operating period.
The staff has concluded that operation of Palo Verde Unit 2 for the proposed six-month interval does not pose an undue risk to the public health and safety.
This conclusion is based on the numerous licensee actions to address the root cause of the Unit 2 SGTR event to minimize the potential for future SGTR events, actions taken to enhance primary-to-secondary leakage monitoring to provide early indication of tube degradation, improvements in emergency operating procedures for SGTR events, and the reduction in the variable overpower trip setpoint and limits on primary coolant activity to limit potential consequences should an SGTR occur.
Further, while uncertainty exists in estimating the structural integrity of the steam generator tubes toward the end of the proposed operating period, the risk evaluation confirms that the potential risk to the public is small.
Principal Contributors:
R.C. Jones, Jr.
E. Hurphy K. Karwoski K. Eccleston C. Liang R.
Emch C. Trammell Date:
August 19, 1993 25
l I~
'F l
f
Enclosure 2
COMNEN S
ON HE APS ETTE OF JU Y 25 993 The APS letter of July 25, 1993, provided the licensee's safety assessment of continued operation of Palo Verde Units 1 and 3 with potentially degraded steam generator tubes.
This assessment included a comparative assessment of the Palo Verde Unit 1 and 3 steam generators with the Palo Verde Unit 2 steam generators.
The licensee has identified several differences between the operational performance of the Unit 2 steam generators versus the Unit 1 and 3 steam generators.
The licensee believes that these differences, when considered collectively, contributed to the rapid propagation of flaws in Unit 2.
In addition, the licensee believes that these differences lead to the conclusion that the rate of crack propagation or initiation in Units 1 and 3 is significantly less.
The differences cited by the licensee include the following:
~
Eddy current test data from past inspections has shown degradation rates in Steam Generator No.
2 of Unit 2 to exceed the rates observed at Units 1
and 3.
~
Unit 2 experienced an acute intrusion of secondary resin beads into the steam generators.
The resin was discovered on the can deck in the steam generators and subsequent investigation revealed previous failures of the demineralizer retention elements.
The resins would be expected to contribute to the accumulation of sulfates in the steam generators.
~
Unit 2 has experienced the hideout return of sulfates during shutdowns and power reductions, indicating that a larger inventory of sulfates are present in the Unit 2 steam generators.
~
Steam Generator No.
2 of Palo Verde Unit 2 (where the tube rupture occurred) has had the highest peak sodium return as well as the highest molar ratio of sodium to chloride as calculated from the peak return concentrations compared to all other Palo Verde steam generators.
The NRC staff has the following comments on these cited differences:
~
The steam generators (SGs) and secondary plants are similar among all three units in terms of design, construction, and operational practice.
Bulk water chemistry experience among the three units has also been similar.
~
At the end of their current operating cycles, Units 1 and 3 will have accumulated a comparable number of operating years as that accumulated at Unit 2 at the time of the rupture.
~
Prior to the Unit 2 rupture event, degradation observed by eddy current testing at the three Palo Verde units was dominated by mechanically-induced degradation (e.g.,
wear), not corrosion-induced cracking.
Thus, the licensee's trending analysis of degradation rates at the three units is not relevant to assessing crack growth rates at the three units.
I~
(
~
~
~
Past eddy current inspections provide little evidence that cracking activity at Units 1 and 3 are lagging cracking activity at Unit 2.
Bobbin probe inspections of 19400 tubes during the previous refueling outage at Palo Verde Unit 2 revealed only one indication characterized at the time as a potential crack.
Two additional indications at Unit 2 characterized as wear indications at that time are now believed to have been crack indications based on a recent reevaluation of the previous inspection results.
The previous bobbin probe inspections at Units 1 and 3 included 15100 and 8700 tubes, respectively, with no reported indications.
It is our understanding that no reanalysis of the previous inspection results has been performed for Units 1 and 3.
Nevertheless, it can be shown with simple statistics that the previous inspection results at Units 1 and 3 are entirely consistent with the hypothesis that cracking is occurring at similar rates at all three units.
(Of course, the previous inspection results do not ~ro e the hypothesis.)
~
As noted earlier, isolated instances of detectable cracks first occurred at Unit 2 during the 1991 refueling outage.
Thus, the environmental conditions promoting. crack initiation and propagation predate that time.
Hideout return data indicate that alkaline or caustic conditions were present in the crevices (including under-deposit crevices) at all three units dating back to at least 1987 and suggest that the aggressiveness of the crevice environment was not unique to Unit 2.
Based on averages of peak hideout return data since
- 1987, sodium and sulfate concentrations among the three units have been similar.
Molar ratios of sodium-to-chloride were highest at Unit 3 during this period with a ratio of 106 for SG 31 and 89 for SG 32.
Molar ratios at Unit 2 were 2.9 at SG 21 and 19.2 at SG 22.
Molar ratios at Unit 1 were 2.8 for SG 11 and 8.1 for SG 12.
Hideout return data for the period 1991 to 1993 indicate that sodium and sulfate inventories in the steam generators have been approximately twice as high for Unit 2 as for Units 1 and 3.
Molar ratios of sodium-to-chloride were 7 for Unit 1, 75 for Unit 2, and 22 for Unit 3.
- However, as noted in Revision 3 of the EPRI Secondary Water Chemistry Guidelines, prediction of crevice chemistry at a high confidence level is not a simple process that can be achieved by relying on simple ratios.
- Rather, recognition must be given to the relative hideout rate of each species and the interactions of species within the crevice.
This point is reinforced by the observation that the predicted pH by MULTEg is about the same for all three units; 10.2 for Unit 1, 10.23 for Unit 2, and 10.35 for Unit 3, suggesting an equally caustic condition at all three units.
MULTEg predicts the chemistry of the solution which results from interaction of different species in concentrated crevices with consideration of the different tendencies for individual species to remain in solution during the boiling process.
- However, MULTEg predictions themselves incorporate significant uncertainty by virtue of the uncertainties of many of the important input parameters.
~
The only definitive indicator of the relative amounts of corrosion crack activity at Units 1 and 3 relative to that at Unit 2 will be the forthcoming tube inspection results from Units 1 and 3.
4 (y ~
~
Enclosure 3
RE UES FOR ADDITIONAL INFORMATION PURSUANT TO 10 CFR 50.54 f PA 0 VERD UNITS I 2
AND 3 Provide your plans and schedules for installing N-16 monitors at Units I, 2, and 3.
Include a physical description of the system to be inst'alled, its location on the steam line and the procedures governing its use.
Discuss the potential benefits and impacts of the use of automatic pressurizer spray initiation for prevention and mitigation of a steam generator tube rupture event.
Describe any analyses performed for this evaluation.
Provide the results of the ATHOS III analysis and your assessment of these results with respect to the previously submitted root cause analysis.
Discuss the implications of these results with respect to potential further corrective actions, together with any specific plans and schedules to implement these additional corrective actions.
Discuss the potential benefits and impacts of changing the steam generator recirculation ratio, including any plans and schedules for making such a change.
Describe the options for achieving such a change.
Discuss the potential benefits and impacts of reducing T0 including any plans and schedules for making such a change.
Discuss the potential benefits of chemical
- cleaning, including any plans and schedules for implementation.
Provide a detailed description of the inspection program to be implemented at the next steam generator inspection outage for each unit.
For Units 2 and 3, submit this description at least six weeks prior to the scheduled start of the inspection.
In the description for each unit address plans for:
a.
eddy current test scope and methods; b.
UT examinations; c.
visual examinations; d.
tube pulls, including any plans and schedules for examining these tubes; and e.
use of advanced or alternative technologies.
Provide the results of the inspection of each unit at least two weeks prior to restart and discuss the proposed operating interval until the next inspection.
In addition, discuss the implications of these results on the other two units.
l y ~