On 16 February, 2002, Davis-Besse plant began a refuelling outage that included inspecting the nozzles entering the head of the reactor pressure vessel (RPV). The inspection focused on nozzles associated with the control rod drive mechanism (CRDM).
The operator, FirstEnergy, found that three CRDM nozzles had indications of axial cracking, which had resulted in leakage of the reactor's pressure boundary. Crack indications were found in nozzles 1, 2 and 3, located near the centre of the RPV head. These findings were reported by FirstEnergy to the NRC on 27 February, 2002.
To identify the cause, FirstEnergy investigated the condition of the RPV head surrounding CRDM nozzle 3. This investigation included removing the CRDM nozzle, removing boric acid deposits from the top of the RPV head, and ultrasonically measuring the thickness of the RPV head.
On completing the boric acid removal, FirstEnergy conducted a visual examination, which identified a large cavity in the RPV head on the downhill side of CRDM nozzle 3. Ultrasonic testing indicated wastage of the low alloy steel RPV head material adjacent to the nozzle. The wastage area extended 5" downhill on the RPV head from the penetration for CRDM nozzle 3 and was 4-5" at its widest part. The minimum remaining thickness of the RPV head in the wastage area was found to be approximately 3/8". This thickness was attributed to the thickness of the stainless steel cladding on the inside surface of the RPV head, nominally 3/8" thick.
Axial cracking along the length of these nozzles had been observed, but not considered to be a safety concern requiring immediate attention. This issue was addressed by the NRC in Generic Letter 97-01, issued in 1997. Inspections and repairs have generally been made during plant maintenance outages. However, in 2001, circumferential cracking part way round the width of the nozzle above the structural retaining weld was discovered at several PWRs.
One function of the nozzles is to maintain the reactor coolant system pressure boundary. Cracking of the CRDM nozzles represent a degradation of the primary reactor coolant system boundary, and is potentially safety significant.
Nozzle cracking at Oconee 1 in November 2000 and Arkansas 1 in February 2001 was limited to axial cracking, of limited safety concern in the NRC's safety evaluation on the cracking of VHP nozzles. However, the discovery of circumferential cracking at Oconee 3 in February 2001 and Oconee 2 in April 2001 raised concerns about potential safety implications and prevalence of cracking in VHP nozzles in PWRs.
Duke Energy, the licensee at Oconee 3, performed a visual examination on the outer surface of the RPV head to inspect for indications of borated water leakage. This followed cleaning of the RPV head during the prior outage to remove all existing boric acid deposits that could mask the identification of subsequent deposits that would indicate new or ongoing leakage. It revealed small amounts of boric acid deposits (less than 1 cubic inch) where the CRDM nozzles exit the RPV head for 9 of the 69 CRDM nozzles.
After several examinations, Duke Energy identified crack indications, but these had been misinterpreted as inconsequential craze cracking with unusual characteristics.
Duke Energy concluded that the root cause of the CRDM nozzle cracking was primary water stress corrosion cracking (PWSCC). The cracking initiated at the OD of the nozzles after cracking of the J-groove weld or adjacent heat-affected zone metal permitted coolant leakage into the annular region between the CRDM nozzle and the RPV head.
For boric acid deposits from CRDM nozzle cracks to be detectable at the outer surface of the RPV head, sufficient reactor coolant has to leak through the primary pressure boundary into the annulus between the CRDM nozzle and the RPV head base metal, propagate up the annulus, and finally emerge onto the outer surface of the RPV head. Since PWSCC cracks in Alloy 600 and Alloy 182 welds are very tight, leakage from axial cracks in the nozzle and associated welds is expected to be small. In addition, possible restraint of pressure-induced bending of circumferential cracks in CRDM nozzles could minimise the leakage available even from CRDM nozzles with large circumferential cracks, as evidenced by small boric acid deposits identified at Oconee 3.
Most CRDM nozzles are installed into the RPV head with an interference fit at room temperature. Should these interference fits persist at plant operating conditions, they could provide an impediment to the flow of coolant leakage up the annulus, limiting the amount of deposit available on the RPV head for detection by visual examination.
CRDM nozzle degradation raises several issues:
• Cracking of Alloy 182 weld metal has been identified in CRDM nozzle J-groove welds. This raises an issue
regarding the adequacy of cracking susceptibility models based only on the base metal conditions.
• Cracking at Arkansas 1 raises an issue regarding adequacy of the industry's susceptibility model. The cracking was predicted to be more than 15 effective full power years beyond January 1997.
• Circumferential cracking of CRDM nozzles, located outside of any structural retaining welds, has been identified for the first time. This raises concerns about the potential for rapidly propagating failure of the CRDM nozzles and control rod ejection, causing a loss of coolant accident.
• Circumferential cracking from the CRDM nozzle OD to the ID has been identified for the first time. This raises concerns about increased consequences of secondary effects of leakage from relatively benign axial cracks.
• Circumferential cracking of CRDM nozzles was identified by the presence of relatively small amounts of boric acid deposits. This increases the need for more effective inspection methods to detect the presence of degradation in CRDM nozzles before the nozzle integrity is compromised.
Because visual examination of the RPV head or volumetric examination of the VHP nozzles occurs only periodically, crack growth rate in VHP nozzles is an important consideration in providing assurance that VHP nozzles will maintain their structural integrity between examination opportunities. Crack growth should be low enough to ensure that VHP nozzles which are unflawed during an examination do not have critical flaw sizes prior to the next scheduled examination.
From the susceptibility ranking model proposed, PWR plants can be divided into several groups:
• Those which have demonstrated the existence of PWSCC in their VHP nozzles through detection of boric acid deposits, and for which cracking can be expected to recur and affect additional VHPs.
• Those which have a moderate susceptibility to PWSCC based upon a susceptibility ranking of less than 5EFPY from the Oconee 3 condition.
• Those which have a moderate susceptibility to PWSCC based upon a susceptibility ranking of more than 5EFPY, but less than 30EFPY from the Oconee 3 condition.
• The balance of plants which have low susceptibility based on a susceptibility ranking of more than 30EFPY from the Oconee 3 condition.
For plants with a low susceptibility to PWSCC, the anticipated low likelihood of PWSCC degradation indicates that enhanced examination beyond the current requirements is not necessary because there is a low likelihood that the enhanced examination would provide additional evidence of the propensity for PWSCC in VHP nozzles.
For plants with a moderate susceptibility to PWSCC based on a susceptibility ranking of more than 5EFPY but less than 30EFPY from the Oconee 3 condition, an effective visual examination of 100% of the VHP nozzles that is capable of detecting and discriminating small amounts of boric acid deposits from VHP nozzle leaks may be sufficient to provide reasonable confidence that PWSCC degradation would be identified prior to posing an undue risk. This effective visual examination should not be compromised by the presence of insulation, existing deposits on the RPV head, or other factors that could interfere with detection of leakage.
For plants with high susceptibility to PWSCC based on a ranking of less than 5EFPY from the Oconee 3 condition, the possibility of VHP nozzle cracking indicates the need to use a qualified visual examination of 100% of the VHP nozzles. This examination should be able to detect and accuractely characterise leakage from cracking in VHP nozzles considering two characteristics. One characteristic is a plant-specific demonstration that any VHP nozzle exhibiting through-wall cracking will provide sufficient leakage to the RPV head surface. Secondly, similar to the effective visual examination for moderate susceptibility plants, the effectiveness of the qualified visual examination should not be compromised by the presence of insulation, existing deposits on the RPV head, or other factors. Absent the use of a qualified visual examination, a qualified volumetric examination of 100% of the VHP nozzles may be appropriate to provide evidence of the structural integrity of the VHP nozzles.
For plants which have already identified the existence of PWSCC in the CRDM nozzles, there is a sufficient likelihood that cracking of VHP nozzles will continue as the facilities continue to operate. Therefore, a qualified volumetric examination of 100% of the VHP nozzles may be appropriate to provide evidence of the structural integrity of VHP nozzles.
The repair plan
FirstEnergy has submitted a plan to repair the RPV head. It calls for cutting out the corroded area of the reactor head and welding in a 15cm thick, corrosion-resistant nickel alloy plate. The plate will be about 45cm in diameter and cover the CRDM nozzle opening adjacent to the corroded area, as well as another nozzle opening near the corrosion. The repair cost is estimated at about $25 million. FirstEnergy said it hopes that Davis-Besse will return to service during the third quarter of 2002, subject to the approval of the NRC.
FirstEnergy said the root cause analysis report of the incident had been submitted to the NRC. The statement makes the following points:
• The conclusions are consistent with those of the probable cause summary report provided to the NRC and those of the NRC augmented inspection team.
• The damage resulted from boric acid seeping through cracks in two of the CRDM nozzles and corroding carbon steel in the RPV head. The cracks in the stainless steel nozzles probably occurred over a period of four years or more.
• Several other factors contributed to the corrosion, including missed opportunities to detect the problem earlier.
• Several corrective actions are listed, including some already implemented, such as modifying the service structure on the head to aid inspection and maintenance. An inspection of all reactor coolant system equipment to ensure there is no other corrosion damage is underway. Some 40 plant systems are being reviewed to identify and resolve any long-term concerns. The boric acid corrosion control programme is also being enhanced.
• Relevant management oversight will in future occur at a much higher level within FirstEnergy.