ML19323H916

Summarizes Questions Transmitted to Bechtel as Result of 800410 Meeting Re Util Embed Rept

ML19323H916
Person / Time
Site:	Callaway
Issue date:	04/24/1980
From:	Schnell D UNION ELECTRIC CO.
To:	Gallagher E NRC OFFICE OF INSPECTION & ENFORCEMENT (IE REGION III)
References
ULNRL-349, NUDOCS 8006170260
Download: ML19323H916 (5)

Text

,. ... 7)(1

• C

• • g UNION EttcTmc COMPANY 19 09 G R ATIOT S T R C C T - ST. LO UIS April 24, 1980 ~o *

• atss-a aP' '. 'O . B O M e49 ST. LOUIS. MO. 6 3les Mr. E. Gallagher U. S. Nuclear Regulatory Commission Region III hM U bc /

799 Roosevelt Road Glen Ellyn, Illinois 60137 ULNRC- 349

Dear Mr. Gallagher:

1 NRC QUESTIONS ON UE EMBED REPORT CALLAWAY PLANT The following generally summarizes the questions transmitted to Bechtel as a result of our April 10, 1980 meeting on the subject report:

1. NRC takes exception to statement in J. K. Bryan cover letter, P.1: "As noted in the Bechtel specification for these embeds, even AWS requirements limiting undersize, profile and other weld characteristics cannot be applied to manually-welded embeds and are unnecessary to assure their ability to carry design loads."

Mr. Gallagher has talked to Moss Davis of AWS plus NRR and I&E HQ people expert in AWS requirements. lie posed this question:

"Are AWS weld profile, undercut, etc. requirements applicable to manually-welded studs?" The answer from these people is "YES".

2. Ilow does Bechtel support deviations from Dl.1 on undercut, profile & allowable underthickness in fillet lengths as allowed in Rev. 9 of Specification Cl31?

It was unclear to the NRC from specification Cl31 whether AWS Dl.1 is a full requirement. Eechtel is to clarify specification intent that AWS Dl.1 is applicable with excep-tions clearly defined.

3. Even though Bechtel specification allows 1/16" under on vertical leg, can we confirm that minimum weld leg is 5/16"?

Apparently anything less than 5/16" violates AWS code for fillet welds in this application.

4. NRC noted that current (1980) AWS Dl.1 section 8.15 does not allow 1/16" undercut.

8006170pGO ]

Mr. E. Gallagher April 24, 1980 Page Two

5. Did Cives Corp. inspect the same mechanically-welded embeds reported in Daniel Data Package under cover DLUC-2399 (Enclosure 7 to ULNRC-238)?

6. Why was section 9.6 added to Bechtel Specification Cl31?

Prior to this revision (9) didn't Cives bend test machine-welded studs with less than 3600 weldflash? Wasn't Cives following Dl.1 4.30.1 prior to June 8, 1977? If Cives was not bend testing questionable studs prior to this time, how or what were they inspecting prior to shipment? If they were performing bend tests, is there any documentation?

Note: Several Cives & Bechtel letters in NRC files indicate that bend tests were not being done prior to Rev. 9 (e.g.: Cives letters to Bechtel, SL:126, July 12, 1977, and SL:134 dated July 14, 1977 and Bechtel letters BLSM-5959, dated Aug. 8, 1977 and BLSE-5195, dated Nov. 21, 1977). Was there any Bechtel follow-up to these letters?

7. Reference Cives letter to Bechtel SL:134, dated Aug. 18, 1977.

NRC is concerned about several things in this letter--References to "new" inspection criteria; statement about "no other sections of AWS apply to our work" and " tolerance for welds having less than 3600 flash". Also, was request in this letter to use '77 version of AWS acted upon by Bechtel?

See also BLSE-5227 dated Nov. 29, 1977 which may respond to concerns in SL-134.

8. Enclosure 1 to UE report (Bechtel probability study) :

Messrs. Gallagher & Landsman of NRC are not expert in probability analysis and will not personally support an analysis of this type to defend acceptability of machine-welded embeds.

NRC concerned not with probability of stud failure but number of defective studs per embed and probability of embed failure.

D. Schnell pointed out the study does cover probability of failure of plate for each area of plate. NRC was suspicious of this analysis. NRC would like to see complete dialogue for probability report.

Do we have evidence of multiple defects per embed? How many plates were involved in the 66 stud defects? Bechtel is to provide a list of plates which contain the 457 defective studs (including 66 failed studs) and determine how many plates had more than 1 defective or failed stud.

9. Bechtel states in Enclosure 1 (pg. 1 in Bechtel Report) that

l

. ,e G, $

Mr.-E. Gallagher

-April 24, 1980 Page Three the " majority" of manually-welded plates did not meet dimensional / profile requirements of specification.

.How many plates are in this " majority"?

How many plates were considered acceptable (meeting spec) after_ relaxations (items a, b, c & d) listed in Bechtel Report?

In-a general statement the NRC questioned whether AWS Dl.1 was fully-implemented prior to June 8, 1977. Bechtel is to verify that AWS was fully utilized as criteria before specification Cl31 was relaxed and determine how effectively it was imposed.

• 10. DIC memo PQWP-152, dated October 26, 1977 has results of inspections of machine-welded embeds which differ markedly from numbers in Enclosure 1. How is difference explained?

Was different criterial used in inspection?

11. Enclosure 1, Appendix B, Table I,:

Why were different. allowable stress levels used (27ksi vs. 36ksi) to. calculate embed " load capacity" and " reduced load capacity" because of assumed undersize?

Using Nelson " cookbook" rules for determining load capacity, it appears capacity of EP-512's should vary for the different configurations A thru F. (NRC could not verify any Bechtel capacities for any configuration of 512's.) How did Bechtel compute capacities and why don't capacities vary for different configurations of EP-512's? Would also like to see calculations to determine how " reduced load capacity" for all types of plates was arrived at.

12. Enclosure 1, Appendix B, Table 3:

Rationalle followed by Bechtel in Table 3b on allowing

'1/16" undercut is based on fact that area of threaded utud-is less than area with 1/16" undercut. NRC & UE confirmed that some unthreaded studs are manually-welded.

It appears that load' capacities for manually-welded Nelson (or headed) studs with'l/16" undercut are not addressed in Bechtel report.

Are load capacities in table nos. 1 and 3 based on tension, shear-or a combination?

DemonstrateLhow calculations are made to show that reduced fillet weld L legs either affect or don't affect load capacity offplate.-

-m. -

. .S ' .' '

.,s'3__

' Mr..E. Gallagher'-

sApril 24, 1980

Page -Four

1A copy of BLUE-675, dated April 9,-1980 was handed out which

.provides'a listing of manually-welded embeds with loads calculated assuming ~1/8"~ undersize welds. After a short discussion, NRC requested-tc see c21culations of how the reduced load capacities were determined.

'13. Regarding Bechtel recommendations in Enclosure 6:

'On plates deemed acceptable with.an average undersize of 1/8", if undersize is concentrated in one area of plate and load.is applied at~this point, does theory that plate is acceptable remain valid? (localized undersize)

Furthermore, if the average weld undersize of 1/8" resulted in a weld profile of less'than 5/16", are the Bechtel conclusions on load capacity still valid in light of AWS-limit on weld size of 5/16" minimum? The NRC's concern is that~AWS; indicates welds less'than 5/16" do not contribute sufficient strength to be reliable.

14. Cives letter SL:124 dated June 30, 1977:

Table 1: ' Attempts to show 1/16" undersize is acceptable but in-making this point, allowable stress of _

43ksi is used instead of 36. What is rationalle?

15. Bechtel letter to Cives dated June 24, 1977, from Divjak to Ross:

'How do the numbers quoted in this letter (no. of studs inspected, etc.) fit with the numbers in tite probability study?

• 16. Seiken letters to Schnell, SLU:6-41, dated Nov. 1, 1976 &

SLBM-6:514, dated Nov. 5, 1976:

Letters indicate Cives was not in control of production quality based on Bechtel/SNUPPS inspection findings.

What' measures were taken? Did DIC/UE perform the recommended inspections at the site? NRC believes this should have highlighted problems before embeds-were installed.

17.- Enclesure 6c(to UE submittal) contains a Bechtel attempt to analyze.NCR.0831. At' random, NRC picked entry #44

.of' Attachment V, pg.-3 and asked why the corresponding DIC inspection report'was eliminated as a data point.

NRC suggested 1(but did .not -insist on) a reanalysis of teach: manual and machine-welded embed.installe'd prior to

-June.9,-1977. About-225 manually-welded embeds and 255-

-N. -- a -

,o i .. .

Mr.E.-Gallbgher April 24, 1980 Page Five machine. welded embeds are in this group. A program to evaluate each might consist of:

a. Identification of each plate location.

b. Determine where load is applied-on plate.

c. Calculate actual _ load (vs. plate capacity).

d. Assume most heavily-loaded stud fails and determine what happens.

Investigation may 'show that-

a. Some plates may not have been used. Some may have all load applied now.

b. Some may be so lightly loaded that failure is precluded (even with failing stud).

c. Some may not involve safety-related attachme :s.

d. Failure analysis may show that failure does not threaten any system or structure from a safety point of view.

The above analysis may remove most of the embeds in question from further concern. It is recognized, however, that this approach may be totally impractical.

NRC'would like to have answers to foregoing questions within 30 days. Mr. Gallagher is attempting to complete his analysis of the Callaway embed report by the end of June (1980).

Response must have UE/DIC input. l Please call me_if you have any questions or if something of importance was omitted from the foregoing list.

Very truly yours,

/0 Donald F. Schnell h 'ager - Nuclear Engineering DFS/sla