ML18057B481

Forwards Addl Info to Support 910909 Request for Relief from Specific ASME Boiler & Pressure Vessel Code Requirements to Enable Repair of Containment Air Cooler,Per NRC Request

ML18057B481
Person / Time
Site:	Palisades
Issue date:	01/16/1992
From:	Slade G CONSUMERS ENERGY CO. (FORMERLY CONSUMERS POWER CO.)
To:	NRC OFFICE OF INFORMATION RESOURCES MANAGEMENT (IRM)
References
NUDOCS 9201230033
Download: ML18057B481 (11)

Text

~[]~SMimern

!PJ[]W~r GB Slade General Manager POWERiNii NllCHlliAN'S PRDliRESS Palisades Nuclear Plant: 27780 Blue Star Memorial Highway, Covert, Ml 49043 January 16,, 1992 Nuclear Regulatory Commission Document Control Desk Washington, DC 20555 DOCKET 50-255 - LICENSE DPR PALISADES PLANT -

REQUEST FOR RELIEF FROM SPECIFIC ASME B&PV CODE REQUIREMENTS - CONTAINMENT AIR COOLER CODE REPAIR - , ADDITIONAL INFORMATION

.

Consumers Power Company's letter, dated September 9, 1991, entitled, "Request for Relief From Specific ASME B&PV Code Requirement - Containment Air Cooler Code Repair," requested relief from code repair requirements for the Palisades Plant containment air coolers. NRC staff reviewers have requested additional information. Their questions and Consumers Power Company responses are provided as the attachment to this letter.

~~

Gerald B Slade General Manager CC Administrator, Region III, USNRC NRC Resident Inspector - Palisades Attachment A CIVISEIVERGY COMPANY

ATTACHMENT Consumers Power Company Pali sades Pl ant Docket 50-255 REQUEST FOR RELIEF FROM SPECIFIC ASME B&PV CODE REQUIREMENTS CONTAINMENT AIR COOLER CODE REPAIR ADDITIONAL INFORMATION January 16, 1992 9 Pages

1

1. Are the containment air coolers (GAG}, supposed to be seismically designed as part of the licensing basis? Are the CACs supposed to function post LOCA? If not, have the effects of flooding been addressed and what are the postulated size break conditions?

Yes, the design specification for the containment air coolers identifies the seismic design requirements. The coolers are seismically designed to withstand seismic loads concurrent with the normal and maximum design operating loads. The original design specification M-60, " Specification for Cooling and Heating Coils, Palisades Plant Consumers Power Company,"

for the containment air coolers identifies the following requirements:

"For normal operating loads, the seismic load is 0.46g in the horizontal direction and 0.067g in the vertical direction. The resultant cooler coil stress shall not exceed the allowable stress permitted by the Code for Pressure Piping, ASA 831.1, Section 1 with addition of Nuclear Code Case N-1.

For maximum design loads, the seismic loads will be 0.90g in any horizontal direction and 0.134g in the vertical direction. The stress shall not exceed the material minimum yield strength."

However, we do not have any calculations or test reports that specifically address the seismic qualification of the coolers. We recently requested that both the plant architectural engineer and the cooler vendor search their files for documentation on the seismic qualification of the coolers, but none could be found.

Yes, the coolers are designed to function post LOCA and post MSLB to condense steam to reduce post accident containment pressure.

Since the coolers are seismically designed and function post accident, no post accident flooding effects due to cooler failure have been addressed.

During power operation the on-line cooler operation and the methods for identifying cooler leakage have been addressed as each cooler is designed with a float switch capable of alarming when cooler leak rate exceeds 20 gpm. We have also stated in our previous relief request, dated January 16, 1991 and updated February 11, and 27, 1991, the capability to detect cooler leakage as low as 1 gpm.

2. Can the licensee demonstrate analytically that the braze joints in question are able to withstand a design basis seismic event (SSE)?

Analysis has been considered. However, there is some concern about the viability of such a finite element analysis where significant geometric continuities exist and workmanship strongly controls structural integrity.

In addition, because of the high cost associated with the analysis we've decided to pursue the response to Question 4 below to support the seismic qualification of the coolers. Since neither of the accident scenarios

2 that the coolers are designed to mitigate (LOCA and MSLB), are seismically initiated events, then a cooler failure as a result of a seismic event would likely result in a controlled plant shutdown, but would not result in a challenge to the coolers design basis operation.

3. Can the ljcensee demonstrate by physjcal test on the retjred header that the two most serjous fajlure modes are unljkely? These are: 1) tube

, pull-out from the manjfold, and 2) separatjon of the mjter jojnt.

Yes, a test method could be developed, but could not be performed on one of the retired headers. During the 1990 refueling outage all of the coils in the number four cooler were replaced with new coils. All of the used coils, manifolds, and headers were scrapped with the exception of a select few coil sections or parts used for root cause analysis and research.

Unfortunately all of the header arrangements that were saved have been destructively examined to one stage or another, and, therefore, we don't believe that a physical test on some of the remaining retired header parts could be accurately correlated to a header strength proof test.

4. Can the ljcensee demonstrate by showjng that the physjcal desjgn of the CAC prevents the two jtems above from happenjng due to physjcal constrajnts of the solder jojnts?

The following is an explanation as to why the tube pullout from the manifold and separation of the miter joint is unlikely. Figures 1 and 2 attached show cooler manifold and tube connection details to help visualize these joint configurations.

The physical arrangement of the tube-to-manifold joint prevents joint separation. The 1/2-inch tube (5/8-inch 0.0.) is inserted into the 1 1/2-inch manifold (1 5/8-inch 0.0.), approximately 1/2-inch. There are 16 tube-to-manifold joints on a single manifold spaced 1 1/2-inch on centers. The failure of a single tube braze would not cause the tube-to-manifold joint to separate since the tubes are restrained axially by the support plate and u-bend. The likelihood of failing all sixteen joints on a single manifold is considered extremely remote due to the design of the header manifold.

The physical arrangement of the miter and manifold-to-header joint does not lend itself to joint separation. The miter joint is fabricated with a 1/2-inch tube inserted into the side of the joint and there are three miter joints spaced 6 1/4-inches apart. Separation of the miter joint would likely only occur should one of the joints .fail completely with the other two joints collapsing, or if all three joints completely fail simultaneously. We believe it is extremely unlikely that conditions would exist which would cause an entire miter joint braze to fail. The thin wall nature of the tubing would tend to yield and deform after a certain amount of the braze fails. While leakage would occur, joint separation would not likely occur. The additional 1/2-inch tube tied into the side of the joint also makes complete joint failure unlikely.

3 The manifold-to-header joints consist of three 1 1/2-inch tubes inserted into the 3 1/2-inch header approximately 1/8-inch. The three manifold-to-header joints on the header are on 6 1/4-inch centers. Failure of any one of these joints would be similar to the 1/2-inch tube-to-manifold connection in that the installation prevents separation.

The entire header assembly, including the miter joints, is copper tubing that is a very ductile material. Significant structural distortion can be sustained before gross failure and miter joint separation is anticipated.

Joint separation would be expected only after the tubing has been subjected to fatigue or very corrosive environment. A fatigue environment does not exist and regular inspection and testing of the coolers provides assurance that significant corrosion damage does not exist.

A review of the analysis of the piping which connects to the flange of the cooler coil has been reviewed. This connection, as is typical of equipment connections, was modeled as an anchor point in the piping analysis. This produces conservative loads in the piping system due to the full translational and rotational restraint. However, these conservative loads are over predicted for the cooler coils since a certain amount of flexibility exists in the manifold configuration. Using these conservative postulated loads at the flange and a simplified analysis, the tubing is calculated to yield under seismic loadings.

As a result of other reviews and identified problems, Palisades is completing a Safety Related Piping Reverification Project (SRPRP). This effort will reconfirm that all of the work done in response the IE Bulletin 79-14 is correct. When piping or analysis deficiencies are found they will be programmatically corrected as part of this effort. Until this work is completed we cannot predict accurate loads at the coolers.

In conjunction with the redesign of the coolers, we have requested that the SRPRP project complete the walkdowns, reviews and analyses needed to confirm the design of the critical service water system in the containment, commensurate with the design input needs of the coil project.

The SRPRP schedule has been revised to assure that this system verification is completed as required.

A review of the piping configuration, which connects the flanges individually, reveals that the likelihood of developing loads which would cause gross failure is remote. The piping is in an area of low seismic accelerations and contains anchor points or lateral restraints near the flanged c~nnection. One 6-inch diameter header feeds four 3-inch diameter pipes connecting to the flanges. These configurations allow the sharing of the load from the header and would allow the transfer of load from one manifold to the other should a weakness develop in one of the cooler units.

4

5. In previous submittals you have addressed flooding in containment (MSLB, LOCA). This flooding was based on a 4 gpm leakage from small, service-induced, "wisping" leaks. You should look at a larger worse case leak, considering the antiquated design of the tube/header/miter joint design and:

A) Analyze potential for such leaks, and B) Assess consequences and mitigating actions which could be conservatively assumed.

Our February 27, 1991 letter to the NRC responded to the NRC question about the confidence we had that the brazed joints would not fail during a MSLB. In that letter we also described the amount of leakage (4 gpm) that our post accident analysis could accept and described the consequences and mitigating actions we could take if larger leaks appeared during a main steam line break.

In response to Question 4 above, we described the reasons we don't believe that tube pull-out from the manifold or separation of the miter joints will occur. We, therefore, do not believe that the original design of the coolers or the design of the coolers today will allow tube pull-out or separation of the miter joint.

As stated in our February 27, 1991 submittal, based on the low probability of a catastrophic failure of a tube during a MSLB or LOCA and the low probability of a design basis MSLB or LOCA requiring emergency operation of the containment air coolers, there is enough margin in time to conclude that a hypothesized increase in cooler leakage during a MSLB or LOCA could be addressed and curtailed prior to flooding of required long term environmentally qualified equipment.

Although we previously assumed the design of the coolers to be "antiquated," we have learned, from discussions with the vendor, that the tube-to-manifold joint is a design currently used and has experienced few problems throughout the industry. The Palisades miter joint design has been changed on current designed units and will be modified at Palisades as addressed in response to Question 6.

Regarding consequences and mitigating actions which could be assumed, it is noted that current procedures and operating philosophy have increased the awareness of the operator to possible cooler leaks. Installed instrumentation is capable of indicating leaks as low as 1 gpm and if any leakage is detected, the technical specifications action statement related to an inoperable cooler would be entered, which would require repair or plant shutdown within 7 days.

Given the repairs, replacements and modifications which have occurred since 1978, CPCo feels the material condition of the containment air coolers has improved and using the engineered clamps until the coolers are modified (addressed in response to Question 6) is a reasonable resolution to the current situation.

5 In our relief request, dated January 16, 1991, and updated in our February 11 and February 27, 1991, submittals, we requested relief from the zero leakage requirements of the ASME B&PV code. As part of the information we supplied supporting the request, we reviewed our leak detection equipment and methods, and showed that we could detect leaks in the containment air coolers as small as 1 gpm. We also showed that margins were available in the containment flood elevation (4 gpm for 70 hours8.101852e-4 days <br />0.0194 hours <br />1.157407e-4 weeks <br />2.6635e-5 months <br /> of leakage), and how a proposed total cooler leakage of 4 gpm could be accommodated in a post accident LOCA condition.

In this relief request, we are not requesting relief from the zero leakage requirements of the code. We intend to operate the coolers without leaks.

As stated in the relief request, dated September 9, 1991, if we discover a containment air cooler leak during power operation,*the limiting condition for operation in Technical Specification 3.4.2 for containment cooling will be entered. This allows one cooler to be out of service for up to seven days before a plant shutdown is initiated. If the cooler leakage cannot be stopped within the allowable LCO time frame, the plant will be shutdown in accordance with the requirements of the technical specification. Since we have shown we can detect leakage as low as 1 gpm during normal operation, if larger leaks occurred during normal operation, they, too, could also be easily detected.

Again, as discussed in our January 16, 1991 relief request and as supplemented February 11 and February 27, 1991, containment air cooler leakage can occur following pressure transients in the service water system which generally occur as a result of testing during plant shutdown.

During operation, the containment air coolers normally operate without these transients occurring.

During a MSLB or LOCA scenario, a safety injection signal initiates the opening of the 8-inch outlet valves on the containment air coolers. This allows for maximum flow through the ~oolers in order to remove the additional accident heat load, and would result in a lower system pressure in the air coolers. If a loss of off-site power were also to occur, the service water pumps would be load shed and then sequenced back on the diesel generators. The A and B service water pumps start at 9 seconds and the C service water pump starts at 26 seconds after diesel start. Because the 8-inch cooler outlet valves are opened before the pumps start, we expect that the pressure surges in this accident condition would be smaller than surges from the pump starts under normal testing conditions.

Variations in service water system pressure would be dependent on whether a loss of off-site power occurs, and the number of diesel generators and service water pumps starting in an accident scenario.

The maximum 4 gpm leak rate proposed by the above mentioned relief request assumes that flooding of required long term post accident environmentally qualified equipment will occur if the 4 gpm service water leakage were to continue for more than 70 hours8.101852e-4 days <br />0.0194 hours <br />1.157407e-4 weeks <br />2.6635e-5 months <br /> following a LOCA. Containment flooding is more limiting for a LOCA due to the amount of water released inside the containment verses the main steam and feedwater released during a MSLB.

This scenario assumes one emergency diesel and two service water pumps (A

6 and C) operating and three containment air coolers and one spray pump operable for containment heat removal. If, during this 70 hours8.101852e-4 days <br />0.0194 hours <br />1.157407e-4 weeks <br />2.6635e-5 months <br />, off-site power were returned or the second diesel generator was operable, or another source of emergency power established, (to provide power for containment spray pump operation), then the service water to the containment could be isolated, if necessary, (assuming the containment water level was showing an increasing trend and service water isolation was necessary) and containment sprays used to remove heat in containment.

Therefore, even if a pressure transient during the MSLB or LOCA were to increase the amount of leakage to twice the limit (8 gpm), 35 hours4.050926e-4 days <br />0.00972 hours <br />5.787037e-5 weeks <br />1.33175e-5 months <br /> would still be available to return off-site power or start the second diesel generator to service or possibly provide an alternate source of emergency power to power the containment spray pumps, and isolate the coolers to prevent further flooding of equipment in containment.

The LOCA analysis was reviewed and shows that the containment sprays alone could control containment temperature and pressure 70 hours8.101852e-4 days <br />0.0194 hours <br />1.157407e-4 weeks <br />2.6635e-5 months <br /> after the event. At this time, service water could be isolated to the containment to stop any service water leakage.

Operation with a leak-rate of 4 gpm would result in the addition of approximately 16,800 gallons of water to the containment building over a 70 hour8.101852e-4 days <br />0.0194 hours <br />1.157407e-4 weeks <br />2.6635e-5 months <br /> period. This amount is considered insignificant with regard to dilution of boron concentrations considering; the length of time over which the dilution would occur, our ability to inject concentrated boric acid, and the fact that control rods would be inserted.

The above information notwithstanding, it is important to note that use of the engineered clamps will attain zero leakage on tube-to-manifold joints and miter joints which warrant repair. During the 1992 refueling outage, the coolers will be inspected for leakage and clamps applied where needed.

We intend to complete the cooler leakage inspection after the SIS testing, which we believe is the worst pressure transient the coolers see. This testing will, therefore, reveal leak locations for repair.

During power operation our intent is to monitor for leaks with the current instrumentation (which can detect leak rates as low as 1 gpm), and if any are detected, we will enter containment, inspect the coolers and seal the leaks as necessary.

6. Schedule Recognizing the need for proper engineering walk-downs and the desire to "do it right", does not justify your extended cooler replacement proposal.

Any non-code repair allowable will be based, in part, on time till cooler rep 1acement.

The staff would like a commitment to replace:

1 cooler - 93 outage 2 coolers - 94 outage

7 The staff feels the degraded condition of the coolers, and the fact that the Staff desired cooler replacement when the Last relief was granted -

warrants this replacement schedule.

We acknowledge the NRC staffs concern for timely modification of the containment air coolers. Based on the staffs request, we have reviewed our previously.proposed schedule. For the reasons outlined below, we believe that the following schedule will allow us to develop the best alternatives for containment air cooler modification.

• Verification of new cooler design during 1993 Refout.

• Installation of 2 cooler modifications during the 1994 refout.

• Installation of the 1 remaining cooler during the 1996 Refout.

Our September 9, 1991 relief request provided some reasons for our suggested schedule. The NRC's schedule suggests that one cooler be replaced in the 1993 refueling outage. The 1993 refueling outage is scheduled to begin in May of 1993. To assure that actual outage plans go as scheduled, certain pre-outage milestones are established for outage preparation and planning purposes. The outage milestone for major equipment delivery is usually 1 or 2 months prior to the start of the outage, which would be March or April of 1993. As noted in our September 9, 1991 letter, all reasonable estimates for delivery of replacement coolers are 30 to 34 weeks, which would mean that purchase orders would have to be issued in July or August of 1992. Once a design is finalized it would take about three months to complete a final specification for new coolers, therefore, a design would have to be finalized in April or May of 1992. Conceptual engineering and walkdowns are scheduled to be completed by the end of the 1992 refueling outage which is scheduled for the end of April 1992. We don't consider there to be enough time to develop an adequate, verifiable design for a new cooler and be ready for the 1993 refueling outage. We are planning on using the 1993 refueling outage to verify that our final design can be implemented, and complete modifications on the schedule noted above.

Also, as noted in our response to Question 4, we need to determine what the loads applied from the piping system to the cooler are. The SRPRP project will complete walkdowns during the 1992 refueling outage to obtain data needed to verify the design of the service water system inside of the containment. The analysis of this data will need to be completed before we can finalize our cooler design. We have recently requested the SRPRP project to adjust their schedule for completion of this in containment work and they still need to integrate the analysis part of this work into their schedule.

FIGURE 1

• NON-CLEANABLE U-BEND GALVANIZED STEEL 3Yt o. D.

COPPER HEADER FINS

%* o. D. TUBE .

M~NIFOLD I '

f TURBULENT I FLOW 118 1 0. D. TUBE 1gs9 MODEL

... * \ I- ..

FIGURE 2

• TUBE WELD !UBE* TO-MA~IFOLD MAM\FOLD 151e* O.D.TUBlNG TU&4 WELD EMSlON 1 0 3 'z O.D. COPPER TUBING

. HEA'DE.R Ml1ER JOINT l'1G8 MODEL