ML20248M263

Insp Rept 99901298/98-01 on 980114-16 & 0302-06.Violation & Nonconformance Noted.Major Areas Inspected:Implementation of NLI QA Program

ML20248M263
Person / Time
Issue date:	06/10/1998
From:	NRC (Affiliation Not Assigned)
To:
Shared Package
ML20248M243	List: ML20248M242 ML20248M256 ML20248M259 ML20248M263
References
REF-QA-99901298 99901298-98-01, 99901298-98-1, NUDOCS 9806150216
Download: ML20248M263 (22)

Text

_ _ _ _ _ _ _

U.S. NUCLEAR REGULATORY COMMISSION OFFICE OF NUCLEAR REACTOR REGULATION Report No.: 99901298/98-01 Organization: Nuclear Logistics, Inc.

7461 Airport Freeway Fort Worth, Texas 76118

Contact:

Archie C. Bell Quality Assurance Maneger (817) 284-0077 Nuclear Industry: Servicing and refurbishing low- and medium-voltage switchgear, third party dedication of commercial-grade procured components.

Dates: January 14.-16,1998 March 2-6,1998 Inspectors: Kamalakar R. Naidu, HQMB, NRR Stephen D. Alexander, HQMB, NRR David Skeen, PECB, NRR Joseph Petrosino, HQMB, NRR l

. Saba N. Saba, EELB, NRR Yeuh-Li C. Li, EMEB, NRR Approved by: Richard P. Correia, Chief Reliability and Maintenance Section Quality Assurance, Vendor inspection and Maintenance Branch Division of Reactor Controls and Human Factors Office of Nuclear Reactor Regulation l

l Enclosure 3 l

9906150216 990610  ;

PDR GA999 EMVNUCt_O l 99901298 PDR I

l  !

1. INSPECTION

SUMMARY

)

On January 14-16,1998, and March 2-6,1998, the NRC conducted an inspection at Nuclear Logistics, Incorporated (NLI), to review the implementation of the NLI qualty assurance (QA)

Program. Specifically, the inspectors (1) assessed the implementation of the NLI QA program for the manufacture of safety-related batteries at the Fort Smith, Arkansas, faRty of GNB Technologies, Inc. (formerly Gould National Battery), (2) reviewed the criteria that NLl used to extend the life of certain GNB safety-related batteries from 10 to 20 years, (3) reviewed the recent refurbishment by NLI of Westinghouse Type DB-50 low-voltage power circuit breakers for Consolidated Edison's (Coned's) Indian Point Station, Unit 2 (IP-2), and (4) reviewed the dedication by NLl of Westinghouse Type DS 416 breakers (procured as commercial-grade items from the current manufacturer, Eaton/ Cutler-Hammer (ECH)) for use in safety-related applications at the New York Power Authority's (NYPA's) Indian Point Station, Unit 3 (IP3). The inspectors also verified the implementation of the actions taken by NLI to correct two nonconformances identified in NRC Inspection Report 99901298/96-01.

The inspection bases were:

Appendix B, " Quality Assurance Criteria for Nuclear Power Plants and Fuel ,

Reprocessing Plants," to Part 50 of Title 10 of the Code of Federal Regulations (10 CFR l Part 50, Appendix B).

10 CFR Part 21, " Reporting of defects and noncompliance" (Part 21)

The team identified one violation involving failure to develop an appropriate procedure to ,

implement the reporting requirements of 10 CFR Part 21 (Section 3.1). In addition, the i inspectors identified several nonconformances involving: use of a battery quality verification procedure which did not have adequate acceptance criteria (Section 3.2.b.1), failure to properly document and correct conditions adverse to quality (Section 3.2.b.2), permitting the use of an unqualified procedure for double burn process (Section 3.2, b.3), inadequate battery seismic qualification life extension analysis (Section 3.4), inadequate design verification of commercial-grade circuit breaker pole shafts (Section 3.5), and one inspector followup item (IFI) regarding the adequacy of training was also identified (Section 3.5).

2. STATUS OF PREVIOUS INSPECTION FINDINGS.

Nonconformance (99901298/96-01-01) Closed NLI verification plans for motor control components did not specify appropriate acceptance criteria to verify that certain critical characteristics had been met. In an attachment to a letter to the NRC, dated October 4,1996, NLl stated that it had revised SVP-22, " Standard Verification Plan - Circuit Breaker," SVP-1,

" Standard Verification Plan - 600V Starter," and SVP-43, " Standard Verification Plan -Functional Testing Assemblies." The revised SVPs specified appropriate acceptance criteria. According to records, NLI technicians were trained on the revised procedures in March 1997 and the l affected equipmeni passed retests using the revised SVPs. The inspectors did not identify any further concems with these procedures.

2

l I

l i

Nonconformance (99901298/96-01-02) Closed. NLI had not maintained traceability on a GE type circuit breaker sold to Entergy Operations, for use at its River Bend Station. NLI had since obtained the necessary traceability documentation, which was on file. The inspectors did not identify any further concerns.

3.  !

INSPECTION FINDINGS AND OTHER COMMENTS.

3.1 10 CFR Part 21 Implementation

a. I Insoection Scoce t The NRC inspectors evaluated NLI Quality Procedure NLl-QUAL-08, "Nonconformance and 10CFR21 Reporting," Revision 4, dated June 1997, which NLI developed to implement the requirements of 10 CFR Part 21. The NRC inspectors reviewed the procedure, conducted discussions with NLi's QA Manager, evaluated associated procedures and memorandums, and reviewed 10 CFR Part 21 program concerns that were identified during the previous NRC inspection at NLI in 1996 and documented in Inspection Report 99901208/96-01.

b. Observations and Findinas The inspectors observed that NLl-QUAL-08 was posted in a conspicuous location at NLl's Fort Worth, Texas, facility in accordance with the NRC requirements. However, the stated purpose of the procedure y as to provide the process and mechanism for the reporting and disposition of )

"nonconformances"; whereas 10 CFR 21.21(a) requires that procedures adopted pursuant to the regulation provide for the evaluation of deviations and failures to comply to identify defects and failures to comply associated with substantial safety hazards. NLI-QUAL-08 delineated the requirements both for performing a 10 CFR Part 21 deportability evaluation and issuing stop work orders. Section 2.0, " Definitions," of the procedure did not contain any definitions, but stated that the definitions in 10CFR21, Section 21.3 are applicable. Part 21 v. : included as an Attachment to the procedure. The procedure stated that it applies to items which have been supplied to the client and other significant conditions as specified in Section 4.0, I "Nonconformance Reporting," of the procedure.

The inspectors determined NLI-QUAL-08 did not provide appropriate instructions to effectively 1 implement Section 21.21(a) of Part 21. NLl-QUAL-08 focussed on dispositioning production nonconformances in NLl's QA program, but not specifically on implementing the provisions of 10 CFR Part 21. Fundamentally, the procedure commingled in-process nonconforming conditions with those that could be considered deviations and failures to comply in basic components already shipped to or offered for use at NRC licensed facilities. The inspectors ,

determined that NLI Quality Assurance Manual Supplement NLI-08, Revision 0, dated May l

1997, contained more specific 10 CFR Part 21 information ano requirements than NLl-QUAL-08 l which NLI had designated as its official procedures adopted pursuant to 10 CFR 21.21(a). '

I I

During a 1996 irspection, (NRC Inspection Report No. 99901298/96-01-01), the NRC found l that NLl's Part 21 procedures were inconsistent with Part 21 and characterized the finding as a minor violation of Part 21 in accordance with NRC's Enforcan,ent Policy as described in  ;

l 3

I l

1 i

.)

4 NUREG-1600. In response to the identified violation during the 1996 inspection, NLl agreed to revise its procedures to comply with Part 21 requirements. During this inspection, the inspectors informed NLI that it had not taken adequate corrective action for the 1996 ron-cited violation and did not develop an appropriate procedure to implement 10 CFR Part 21.

c. Conclusion NLI still had not established adequate procedures to effectively implement the provisions of 10 CFR Part 21; although this deficiency was identified previously in 1996 as a non-cited violation in NRC Inspection Report. 99901298/96-01. Therefore, this deficiency constituted a repeated violation of 10 CFR 21.21(a) and accordingly, Violation 99901298/98-01-01 is cited.

3.2 QA Oversiaht of GNB Batterv Desian and Manufacturing Process

a. Insoeci!on Scope The NRC inspectors reviewed selected portions of the NLI C2. program to determine if NLI adequately controlled quality activities at the GNB manufacturing facility and corporate engineering offices relating to the manufacture by GNB for NLI of nuclear grade, NCN-Series and NAN-Series lead-acid storage batteries designed for Class 1E service in nuclear power plants. Among the documents reviewed were those that described and prescribed NLl's QA oversight and direct verification (quality control or QC) activities at the GNB Fort Smith, Arkansas, factory. Principal program description and basic verification requirements were contained in NLI Report #R-017-001, Revision 5, dated October 17,1997, " Integrated Verification of GNB Technologies Flooded Stationary Batteries for Use in Nuclear Power Plants; NLI Standard Verification Plan SVP-31, "GNB Batteries and Hardware," Revision 10, dated i

October 19,1997, and Revision 11, dated December 2,1997. The inspectors also reviewed selected individual NLl verification plans for components, parts and materials purchased by GNB such as VP-BATTERY-1-22, Revision 0, for Columbus Oxide 25% red lead. This phase of the review was for assessment of NLl QA oversight program adequacy. Program implementation, compliance and effectiveness were reviewed separately by examining manufacturing and testing records and direct observation of manufacturing activities and NLI QA inspectors witnessing and verifying those activities.

b. Observations and Findinas b.1 Oversicht Proaram and Verification Plan The inspectors did not identify any deficiencies in the description of NLl's QA oversight activities  !

for GNB manufacturing processes in R-017-001, nor in the selected purchased hardware  !

verification plans reviewed. However, the inspectors found that some of the verifications j specified in SV?-31 Revision 10, did not include appropriate (or in some cases any) i acceptance criteria, either in the body of the procedure or in the individual process verification j

) data sheets. The inspectors identified the following deficiencies which still existed in Revision j 11, except as indicated.

fl l

I 4

Data Sheet 2 specdied the acceptable range of Humbolt penetration for paste, but the instrument was found to be uncalibrated.

The SVP-31 text in Section 3.0, " Sample Size," did not address the critical processes of grid casting, paste mixing, and grid pasting; although Paragraph 4.1.2, " Plate Pasting,"

under Section 4.0, " Cell Manufacturing," addressed this process generally.

Paragraph 4.1.1, "Busbar Casting," required verifying that the lead alloy was " tested upon receipt and daily during use for NLl busbars. However, neither 4.1.1, nor its associated Data Sheet 1 specified what the lead should be tested for, test methods, or acceptance criteria.

According to Paragraph 4.1.3, " Cured Plates," cured (dried / hardened paste) plates were supposed to be weighed and dimensionally checked on a 1/16 (32 minimum) sample basis with 100 % visualinspection for voids or missing pellets. Results were to be recorded on Data Sheet 3, " Cured Plates." The out-of-square and flatness tolerances were specified on Data Sheet 3 but not the range of actual allowable plate thicknesses, widths, and lengths. Here, SVP-31 only referred to GNB specifications for other requirements. It did not provide for writing in the applicable requirements for the size plates being inspected from the GNB specifications for use in the field while inspecting cured plates.

Paragraph 4.1.4, "Busbar inspection," specified recording results on Data Sheet 12, but Data Sheet 12 was for visual and dimensionalinspection of flat washers. No data sheet had been provided for recording the results of the busbar inspection; therefore, there were no acceptance criteria, either in the body of SVP-31 or in the data sheets, except the instruction in 4.1.4 to " perform and record a destructive test to see if busbars bend under slight pressure." No guidance was provided on samp!e size for the destructive test, except in Section 3.0, " Sample Size," which prescribed a general 100% sample size except as noted in that section and a rate less than 100% for busbar inspection was not listed. Paragraph 4.1.4 in Revision 11 stated that Data Sheet 13 should be used, but it was a blank, general purpose data sheet, still without acceptance criteria.

Data Sheet 4, " Cell Assembly," listed terminal post alignment as an attribute to verify, but did not define this alignment or state an acceptance criterion except that the GNB alignment fixture must not bind. The acceptance criterion for this attribute was listed in Paragraph 4.1.5.d " Cell Element Test," of the procedure, but not referred to on the data sheet. The inspectors observed the typical use of the data sheets only rather than the whole procedure in the field.

Plate-to-terminal resistance was listed as an attribute to be verified on Data Sheet 4 and an acceptable range given, but the probe location, which could conceivably affect the reading, was not specified on the Data Sheet or in Paragraph 4.1.5 d where the test was also prescribed. NLI stated that the desired probe positioning was covered in operator

training, but, as with the case of the so-called " double-burn" procedure, NLl's QA inspectors training consisted solely of observing GNB operators under instruction with a

more experienced NLI inspector.

5

I i

The sealed-cellleak check parameters were given on Data Sheet 4, but without a tolerance on the test pressure or duration, and without acceptance or rejection criteria (e.g., an allowable amount of pressure drop during the test, if any).

Data Sheet 4 stated that NLI QA personnel themselves were to torque the seal nut, but this is inconsistent with Criterion X of 10 CFR Part 50, Appendix B, " Inspection," which requires that QA overchecks be nade by someone other than the one who performed the operation.

The SVP-31 Section 3.0 sampling basis for Data Sheet 5, " Formation," stated that cell temperatures are " .. constant during formation.. " However, unless the cells can

' dissipate heat just as rapidly as the current produces it, the temperatures will rise. It was also riot clear how temperatures were to be monitored constantly to verify this condition upon which the sample rate for verification was based. Data Sheet 5 specified that cell temperatures remain within a fairly wide range (given), but consistency among cells was not specified. This same sample basis appeared in Revision 11, but it did not refer to Data Sheet 5 which was eliminated in Revision 11.

In Revision 10, neither Section 4.1.6, " Formation," nor Data Sheet 5 specified all the charging and discharging parameters (e.g., charge duration, charging rate, ampere-hours to be charged or discharged, or. termination criteria such as end voltage, specific gravity, etc.) for the formation process. The only specific requirement was that two "24-hour letdowns" be performed during the process. In Revision 11, Paragraph 4.1.6 simply referred to the GNB formation procedure but did not provide a place to document that the requirements were met since Data Sheet 5 had been eliminated. The inspectors were concerned that the formation process, which is critical to subsequent battery performance and life, was not being adequately specified and monitored by NLI under the provisions of SVP-31.

Paragraph 4.2.1.c, " Specific Gravity," [under Section 4.2, " Testing" (IEEE-450)],

specified measuring and recording the electrolyte specific gravity of each cell before capacity testing, but an acceptable range was not provided here or on the associated Data Sheet 6.

Paragraph 4.2.1.e, " Test Current," directed calculating the required test current based upon the battery "mean average temperature," but neither this instruction, nor the data sheet (Data Sheet 6) provided for specifying the " hour rate" desired for the test (e.g., the 2-hour rate,4-hour rate, 8-hour rate, etc.) or the actual nominal or uncorrected discharge current corresponding to the hour rate to be used, of which, cell / battery capacity is a non-linear function.

Paragraph 4.2.1.f, " Shunt Voltage," specified calculating the required shunt voltage based on the test current using the formula on Data Sheet 6. However, it was not possible to get a meaningful result using the formula as written on the Data Sheet  !

because the shunt voltage formula on Data Sheet 6 did not make mathematical or I electrical sense. The formula as expressed on the data sheet was:

6

~

i l

\

"(Shunt mv ratingishunt current rating) + (x/ test current) = "[ sic) l Where, presumably, "x" was supposed to be the shunt voltage to be measured. First, the equation was not expressed in terms of shunt voltage, i.e., the parameter being

)

calculated, but rather as one ratio divided by another involving the desired parameter. '

Second, having the first ratio divided by the second (which contains the dependant i variable) and the quotient equal to some undefined parameter was not mathematically l l or electrically correct. In effect, it was one equation with two unknowns, which, of j course, is unsolvable. Even if expressed as a proportion, in order to follow Ohm's law I and to be mathematically rigorous, the equation should have been written as follows:

(Shunt mv ratingishunt current rating) = (x/ test current)  ;

where "x" is the shunt voltage to be measured (and maintained in order to ensure constant test discharge current). However, because the shunt mv rating divided by its current rating is the shunt's rated resistance, solving for x or shunt voltage, the formula, more usefully expressed in the form of Ohm's Law would be: 1 Shunt Voltage = (Shunt mv rating) + (shunt current rating) X(test current) = .

The inspectors determined through a review of selected completed data packages that NLI QA inspectors and technicians had worked around the erroneous formula and figured out how to calculate shunt voltage correctly, but had not taken action to correct the pre-printed formula.

. As discussed in Section b.3 below, both revisions of SVP-31 reviewed lacked the '75%

depth-of-burn" acceptance value for the busbar-to-plate fusion operation. SVP-31 only referenced GNB Specifications PD-14-00-FST and PR-14-05-FST.

The lack of acceptance criteria, or appropriate acceptance criteria (and adequate specificity in prescribing test parameters), instructions, procedures and drawings prescribing activities affecting quality (i.e., SVP-31) constituted a nonconformance with respect to Criterion V,

" Instructions, Procedures, and Drawings," of 10 CFR Part 50, Appendix B. Accordingly, Nonconformance 99901298/98-01-08 was identifcod.

c.1 Conclusions The inspectors concluded on the basis of this programmatic review that the overall NLI approach to QA oversight of GNB battery manufacturing processes supplemented with some direct and independent verification activities in selected areas was a generally sound approach, consistent with the intent of 10 CFR Part 50, Appendix B. However, many programmatic elements had significant weaknesses as cited in this report; and implementation was not always consistent or effective, suffering principally from inadequate detailed guidance for, or training of, NLI QA inspectors.

i 7

l

. b.2 QA Audits and Corrective Action ,

- Two NLI internal memoranda written by the NLI Quality Assurance (QA) Manager discussed

" Informational Meetings / Limited Scope Assessments" of GNB's Lombard, Illirois, engineering office, responsible for battery design on September 26,1997, and of GNB's Fort Smith,

{

Arkansas, battery factory on September 30,1997. The NLI President and Vice-President had received copies of the memoranda, dated October 19,1997, and October 6,1997, respectively. .

Both memoranda stated that " based on the informational Meeting / Limited Scope Assessment, I potential areas of concern were identified." As a result of the meetings, NLI generated an action item list. The inspectors noted that many of the issues delineated on the "NLI Action i Item List" identified discrepancies in implementation or establishment of the GNB and/or NLI i quality assurance program. For example, (a) GNB purchase orders (POs) referenced the incorrect revision level of the part number being ordered, (b) GNB receiving inspection records  ;

did not contain provisions for documentation review in all cases, (c) NLI advised GNB to revise its Fort Smith battery cell traveler by October 17,1997, to ensure that acceptance criteria for hold points would be identified; and (d) NLl requested GNB to verify that pull tests are being performed on the jar-to-jar cover joint to demonstrate that the adhesive meets the requirements of the applicable material specification.

The October 6,1997, memorandum further stated that "it appears based upon the limited scope assessment performed at GNB Technologies; Fort Smith, Arkansas, on September 30,1997, that there has been a Quality System breakdown in Design Control (Configuration Control),. l Procurement Document Control and Control of Purchased material (Receipt inspection)." The i October 19,1997, memorandum further stated that "it appears based upcn the limited scope assessment performed at GNB Technologies; Lombard, Illinois, on Septembm 26,1997, that there has been a Quality System breakdown in document control and vendor assessment. In both cases, this breakdown is probably traceable to lack of sufficient trained resources to accomplish the required tasks."

Section 4.1, "Nonconformance Identification," of NLI Procedure NLl-QUAL-08, stated that a nonconformance can be identified by any NLl personnel during the course of a project or any other activity and that, upon identification of a nonconformance, the applicable sections of the

- NLl nonconformarce report must be completed. Section 6.2, " Nonconforming Materials and Corrective actions," of the NLI Quality Assurance Manual (QAM), Revision 1, July 1991, required a nonconformance report (NCR) to be issued to document nonconforming conditions.

' Section 6.2 also stated that if the nonconformance is of significant nature and magnitude, the president or Quality Assurance Manager may issue a stop work order. Although the NLI QA

' Manager's memoranda concluded that NLl's limited scope assessment of GNB's Lombard and Fort Smith facilities indicated a " quality system breakdown" in five areas, NLI failed to promptly

document these conditions adverse to quality in a nonconformance report and take adequate action to ensure that the identified deficiencies were corrected and to ensure that any affected, potentially nonconforming material was controlled. As a result, the inspectors were concemed about the impact of the identified deficiencies on the quality of GNB/NLI batteries. Of particular concem was the reliability of potentially affected battery cells to perform their safety functions under all design basis conditions.

8

c.2 Conclusion On the basis of the review of the NLI limited scope assessment of GNB, the inspectors concluded that NLl, in this instance, had not followed NLl-QUAL-08 by failing to document the discrepancies identified at the two GNB facilities in a nonconformance report, and had not taken adequate corrective action. The inspectors concluded that this deficiency constituted, most basically, a nonconformance with respect to the requirements of Criterion XVI, " Corrective Action," of 10 CFR Part 50, Appendix B, and Nonconformance 99901298/98-01-03 was cited.

b.3 CSservation of Strao Burnina of Vented Lead-Acid Stationary Batterv Cells At the GNB Fort Smith, Arkansas facility, the NRC inspectors observed the' process of plate and separator assembly and the process of positive and negative plate lug to-bus bar fusion (strap burning). These processes were crucial steps in the manufacture of four NCN-21 type battery

. cells. The inspectors' objective was to assess the effectiveness of the GNB process controls and the effectiveness of the NLI QA inspection personnel who are supposed to ensure that the process remains within allowable manufacturing parameters. These four safety-related cells were part of two 60-cell batteries intended for the Baltimore Gas and Electric Company's (BG&E) Calvert Cliffs nuclear power plant. The applicable BG&E purchase order (PO), PO 18995, dated January 15,' 1998, was for two 60-cell NCN-21 batteries, and these were the last -

four of the 120 cells to be built.

The inspectors observed that the burners were not performing the task exactly as described in the GNB Fort Smith Standard Operating Procedure (FSOP)-06-02, " Strap Buming of Flooded Stationary Cells,". Revision AA, dated March 12,1996. The FSOP specified first melting the strap " fingers" and plate lugs with the torch, then melting lead sticks to fill to the top of the mold, stirring the mixture with a stainless steel rod, and finally repeating the last step. ' However, contrary to the FSOP, the bumer was observed first to add some molten lead to the bus bar -

'snd plate lugs, melt the mixture together and stir it with the stir stick, then make another torch -

l

' pass over the mixture to smooth the top surface and to blow off any residual slag. i Steps 13 and 14, "Elernent Buming," of GNB's Flooded Assembly Operator Training and i

' Certification Manual, Revision AA, (no date indicated) described the strap-to-lug " double burn"

. process. The training manual called for the bumer to make two passes with the torch and stirring rod to ensure adequate depth of fusion between the bus bar and the plate lugs.

However neither of the bumers interviewed, nor their first line supervisor, were aware that a double bum required two passes with the stirring rod. It had been their understanding that one i pass with the stirring rod and another pass to blow off slag and mix the upper portion of the bus  ;

bar and plate lugs together was what was meant by the term " double bum." When asked why '

lead was first added to the bus bar and plate lugs before melting them together, the burners  ;

stated that because of the loose fit of the molds, it was necessary on the smaller size cell straps i to first add lead to prevent burn-through during the melting process that would result in molten  !

lead running out of the bottom of the mold. The first line supervisor independently confirmed I l- that it was an accepted practice (although, admittedly a deviation from verbatim compliance {

with the language of the procedure), and only when necessary, to add a small amount of molten l lead obtained from a piece of pure lead called a lead bum stick before melting the strap fingers and plate lugs on the smaller size cells, such as the NCN-21, 9

The double burn described in the FSOP was originally developed by a corporate GNB engineer specifically for the larger H type cells in order to achieve a deeper penetration of the bus bars.

Stirring the mixture twice was supposed to ensure adequate mixing and fusing. The process worked so well on the H-series cells that it was decided to post the FSOP at all of the assembly floor burn work stations, including the stations where the NCN type cells were assembled.

However, the procedure was arbitrarily applied across the board without determining if it was really necessary or workable for all types and sizes of cells, such as by experimentation with test straps in the smaller sizes and consultation with burners and first line supervisors. NLI believed that the double burn was a superior practice and specified its use on all NLl cells.

The inspectors interviewed the NLI QA personnel at the GNB Fort Smith facility because they are responsible for witnessing the burning process for nuclear grade NCN-series batteries to assure that the GNB burners were implementing Procedure FSOP-06-02. Even though, the NLI QA personnel were generally knowledgeable of the manufacturing process and understood the NLI requirements for oversight of the strap burning process, none of them fully understood the double burn process and all of them stated that they believed that the double burn process consisted of a single pass to add melted lead and the second pass was to stir the mixture together with the stainless steel rod. The NLI QA personnel stated that their training for the different process controls consisted of on-the-job training from more senior or previous NLI personnel. The NLI QA inspectors did not, nor had they been instructed to, familiarize themselves with the GNB engineering requirements, specifications or individual procedures governing the burning process, nor did the NLI QA Manual or other procedures require them to have done so prior to witnessing strap burns. (Nonconformance 99901298/98-01-06)

The NRC inspectors also observed two other types / sizes of battery cell bus bar/ plate lugs being fused, the M, and H type of cells. The M cell was smaller than the NCN-21 and the H was larger than the NCN-21. Both the H and M cells are commercial GNB battery cells that are not used in safety-related applications. The inspectors observed that the burner who was assigned to an H type cell used the proper double burn method which was delineated in the FSOP posted at the work station. However, the burners that fused the "M" cells also used the same single burn process that was used on the NCN cells.

The controlling GNB specification for the FSOP is PR-14-05-FST, " Assembly Process for Flooded Stationary Cells-Fort Smith," Revision OD, October 22,1997. GNB's Specification PD-14-00-FST, " General Engineering Requirements for Flooded Stationary Cells," Revision OC, October 22,1997, and GNB Flooded Assembly Operator Training and Certification manual, Revision AA, (no date indicated) are also applicable. NLI had also established inspection requirements in NLI SVP-31, " Standard Verification Plan GNB Batteries and Hardware,"

Revision 11, December 1997. However, when all of the different documents were compared with each other, the inspectors noted inconsistencies, specifically:

Step 13, in the " Process Observation-Element Burning," of the GNB Flooded Assembly l Operator Training and Certification Manual did not contain adequate information to ensure consistency in the strap buming process. It was unclear as to the particular point in time when external lead is supposed to be added to the lug-bus bar junction.

The manual narrative is also different from the FSOP; therefore, the quality and i adequacy of the burn may be indeterminate.

10

~

\

l PR-14-05-FST, Revision OD, requires: the maximum acceptable depth of burn of the plate lug to strap shall be 75% of the lug; conversely, PD-14-00-FST, Revision OC, requires: the minimum acceptable depth of burn of plate lug to strap shall be 75%.

GNB engineering specified a "75% depth-of-burn" value, but neither GNB's Operator Training and Certification manual, FSOP, nor NLl's SVP-31 addressed the 75%

acceptance value for process control or verification. After observing the process, the NRC inspectors concluded that it would be difficult for either the GNB burner or NLI QA inspector to determine whether the actual depth-of-burn is close to 75%.

The NRC inspectors informed NLl and GNB management of their observations, including the fact that the narrative sequence and order of the burn events differ between the FSOP and the Operator Training and Certification Manual.

Shortly after being informed of the inspectors' concerns, GNB's Vice President of Technology and Quality Assurance, informed the NRC inspectors that GNB had decided to immediately remove the certification of the burner whom the NRC inspectors had observed and interviewed.

The Vice President also intimated that GNB was considering termination of this employee because he had apparently violated the FSOP. The NRC inspectors explained that this action was not necessary to satisfy the NRC, as the inspectors were primarily interested in GNB and NLl establishing the most appropriate practices, documenting them clearly and ensuring l consistent compliance through adequate training and supentision, as opposed to a show of precipitous reaction to the problem. The inspectors further cautioned GNB and NLl management that terminating the burner after he had told NRC inspectors that he thought he could not follow the GNB procedure in all cases might be viewed, particularly by other employees, as retaliation against an employee for his candor with the NRC. The inspectors were concerned that this action could have a " chilling effect" on GNB employees and affect the NRC inspectors' ability to obtain information.

As an attemative, the inspectors proposed, and GNB and NLI management agreed, that the inspectors would conduct a series of private interviews with all personnel concerned (with union shop stewards and others requested by the interviewee present) and conclusively establish the facts and circumstances surrounding the procedural issues before proceeding further. In parallel, GNB and NLI agreed to dissect one of the NCN-21 straps burned using the practices in question and determine if the fusion was adequate. Accordingly, three of the five NRC inspectors interrupted their respective inspection activities and interviewed the burner of concern, other strap buming operators, the shop supervisor for the applicable area, the Plant Operations Manager, Manager of Process Engineering and NLI QA inspectors.

The NRC inspectors evaluated the information to determine (1) if there was intimidation by GNB management, (2) what the burning practices in use were, (3) what the operators' understanding of the requirements was, (4) what supervisors expectations wera, (5) what NLI QA personnel routinely observed, and (6) what was the NLI QA inspectors' training on and understanding of the requirements for strap buming of batteries for NLI. After interviewing the GNB and NLI personnel, the inspectors determined that GNB and NLI personnel had not been intimidated by their management and the issue did not cause a chilling effect on the workers.

l l

11

The inspectors informed NLI that a safety-related operation was not performed in accordance with an appropriate procedure because (a) the designated procedure wcs not consistently followed and (b) even had it been followed, there was no objective evidence that the double burn process was qualified for the manufacture of NAN or NCN type cells. Further, neither -

GNB nor NLI had an engineering justification to implement the double burn process to the Class 1E NCN or NAN-type cells. (Nonconformance 99901298/98-01-07) c.3 Conclusions NLI QA inspectors had not been trained on the strap burning procedural requirement. Being unaware of the requirements, NLI QA failed to recognize that the GNB burners were not strictly following the procedures when fusing lead bus bars (i.e.," straps") to the individual positive and negative plate lugs of NCN type cells intended for safety-related applications. NLl concurred with the NRC inspectors that its QA personnel had been inspecting the strap fusion process without proper training, without detecting that the process was being performed contrary to procedures, and that the procedure was not appropriate in all cases.

M Licensee Purchase Orders

a. Insoection Scooe The NRC inspectors reviewed purchase orders (POs) issued by several nuclear utilities (licensees) to NLI to determine if NLI adequately translated the technical and quality requirements into POs passed on to GNB for the manufacture of Class 1E battery cells.

b. Observations and Findinos The licensee POs contained (or referenced) sufficient detailed technical requirements, and imposed upon NLI applicable requirements of 10 CFR Part 50, Appendix B, and 10 CFR Part 21, for the supply of Class 1E battery purchase files. For example, Duke Power Company (DPC) PO MN 21954, for its McGuire Station, dated December 11,1996, imposed 10 CFR Part 21 and stated the PO was for components used in nuclear safety-related applications. The PO ordered four NCN-27 batteries, Amperage: 1945AH, Duke MSDS: 10799,60 cell flooded, wet charged,100% capacity,125 VDC. The PO stated that the batteries shall meet all requirements of DPC Specification MCS-1356.01-00-0001, Revision 001, shall be supplied under the supplier's Quality Assurance program that has been approved by DPC, and stated that if lower tier procurement is required, then applicable QA requirements must be invoked on lower tier subcontractors / suppliers. The PO also invoked the requirements of DPC Specification MCS 1356.01-00-001, Revision 001, dated September 27,1989, " Vital Instrumentation and Control Batteries and Racks Procurement Specifications" on NLl. The inspectors observed that the DPC specification addressed seismic and environmental qualification as well as test, inspection, schedules and report requirements. I NLl job order package 017-074 to GNB for the design and manufacture of the batteries for

. DPC, McGuire Nuclear Station Units 1 & 2, indicated that the DPC PO ordered four strings of 60 NCN-27 battery cells. The NLI project synopsis for the DPC order stated that it was "to 1

12 i

l supply dedicated and qualified GNB NCN-27 battery racks and cells strings (60/ string) to McGuire Nuclear Station Units 1 & 2, per DPC MN21954. The cells are to be formed to 100%

capacity and discharge tested in accordance with IEEE 450-1987 requirements - one minute and one hour discharge tests... Seismic and environmental qualification on the batteries and racks will be performed by analysis."

c. Conclusion The inspectors concluded that NLl translation of licensee requirements into its pnhase orders to GNB was generally adequate. ,

3.4 Extension of NCX-1950 Batterv Life

a. Insoection Scom NRC Region 11 Inspection Report 50-269/95-14 identified that the qualified life of GNB Type NCX-1950 batteries that had been originally qualified for 10 years at the Keowee Hydro Station (Class 1E standby power source for the Oconee Nuclear Station), was being extended by NLl for Duke Power Company to 20 years, ostensibly under the provisions Institute of Electrical and Electronic Engineers (IEEE) Standard 535-1986. Note that NCX series batteries are commercial GNB batteries that have been in Class 1E service for some time. The inspectors reviewed the report that NLl produced for Duke in support of the battery aeismic qualification life extension, NLI Report No. C-017002-11, " Battery Seismic Qualification and DLsign Life," dated February 28,1998. The inspectors also reviewed supporting documents including NLl's .

Calculation C-017050-1, based on its Report C-017002, Revision 1, dated Auguet 12,1996, l and Wyle Laboratories' Seismic Simulation Test Report Nos. 44681-2, and 42261-1 dated October 27,1981, and November 12.1991 respectively. In addition, the inspectors reviewed the NLl technical procedures NLl-TECH-04, " Equipment Qualification," and NLl-TECH-05,

• Materials Engineering," related to equipment qualification.

b. Observations and findings The purpose of NLI Report No. C-017002-11 was to demonstrate that the GNB NCX Series j (NCX-7 through NCX-35) and GNB nuclear safety-grade NCN Series batteries (NCN-7 through j NCN-35) were seismically qualified for the 20-year design life. Note that the NCX-1950 cells in j use at Keowee belongs in the NCX-7 through -35 series, but bears the earlier style of type i designation based on rated capacity rather than the number of plates. The report stated that  ;

qualification of the GNB NCN and NCX Series batteries was performed in accordance with ,

NLl-TECH-04 and NLI-TECH-05, but NLl-TECH-04, and NLl-TECH-05 did not provide sufficient i guidance to assure consistency and quality of the seismic qualification report and did not include appropriate quantitative and/or qualitative acceptance criteria for determining that ,

important technical requirements concerning similarity had been satisfied. The report also  !

referenced the two Wyle Reports cited above that documented numerous seismic qualification

, tests previously performed for various NCX battery cells at Wyle Laboratories. In NLI report i l C-017002-11, the basis for seismic qualification was by combination of test and evaluation and  !

l was supposed to show the applicability of the Wyle test results to the batteries being qualified. ,

13 l l

{

However, the NLI report was not a stand-alone document and required extensive supplemental explanations to support conclusions concerning battery similarity, and the applicability of the test data referenced. Specifically, Report C-017002-11 did not a

provide adequate technicaljustification to demonstrate the applicability of the previous Wyle test results to the batteries being qualified, a

include a detailed comparison of the batteries to demonstrate the similarity between the specimen batteries and the batteries that successfully withstood qualification tests, descuce what test data included in the Wyle test report were applicable and specifically which data in the seismic qualification reports is used in C-017002-11,

= adequately explain why it was acceptable to ignore some failures that were noted as anomalies in the Wyle report.

In view of these deficiencies, the inspectors determined that NLI had performed an inadequate design verification calculation, in part because of the lack of an adequate procedure to help ensure that such a complicated analysis is done correctly. In response to the inspectors' concerns, NLI prepared Supplement 1 to C-017002-11, dated March 6,1998. In this supplement, NLI included a summary of the applicable test data and an evaluation of all the anomalies including justification for ignoring some test failures identified in the Wyle reports.

There was also a detailed description of the similarity evaluation which included a comparison of battery design features to demonstrate that the identified design differences would not impact the qualification of the batteries. (Nonconformance 99901298/98-01-05).

c. Conclusions The inspectors concluded that Supplement 1 to NLI Report No. C-017002-11 adequately addressed concerns raised by the inspectors and provided a reasonable technical basis for the seismic qualification for the batteries and the extension of the qualified life from 10 to 20 years.

3.5 Dedication and Refurbishment of Circuit Breakers

a. Insoection Scoce The team r'eviewed NLi's purchase order documentation files for the refurbishment of Westinghouse Type DB-50 circuit breakers for Consolidated Edison's (Coned) Indian Point Station, Unit 2 (IP2), the purchase (from the current manufacturer, Eaton/ Cutler-Hammer) and dedication of Westinghouse Type DS-416 circuit breakers for New York Power Authority's l Indian Point Station, Unit 3 (IP3), and the refurbishment of three Westinghouse Type DHP circuit breakers for Illinois Power Company's (IPC's) Clinton Power Station (Clinton). The inspectors also reviewed the training requirements for NLI personnel performing safety-related 1 activities.

I  !

14

b. Observations and Findinas

. b.1 ' NLl Breaker Technician Trainina The NLl training manual contained no specific category for a circuit breaker technician.

Personnel files of the individuals identified as breaker technicians and interviews with them revealed that most of the individuals did not have much experience at maintaining or refurbishing circuit breakers and what experience they did have was not with nuclear safety-related equipment. The technicians learned about specific circuit breaker types mainly by -

reviewing original equipment manufacturer (OEM) instruction booklets, the applicable NLl breaker refurbishment procedures and any plant-specific information that may have been supplied to them by a particular customer.  !

L NLl had recently instituted a new training program specifically for circuit breaker technicians, engineering personnel, test technicians, and QA/QC personnel. Circuit breaker technician

training includes reading all applicable generic nuclear industry communications (including l' vendor service advisory letters, technical bulletins, etc.) concerning each model circuit breaker and identifying any discrepancies between the NLI procedure and the industry communications to NLI management. However, because this is a new program, the inspectors will review the effectiveness of this program during a future inspection to observe if it has been effectively implemented. (Inspector Followup item 99901298/98-01-04).

During the review of the implementation of the QA program for the manufacture of batteries at GNB, Fort Smith, Arkansas, the inspectors determined that NLI QA inspectors assigned to oversee the quality assurance program at GNB had not been trained to the procedural '

requirements of the strap burning process. This issue is discussed in detail in Section 3.2.b.3

< of this report.

- b.2 Breaker Dedication and Refurbishment Procedums The inspectors found that NLI develops its breaker refurbishment procedures and practices primarily through trial-and-error experience gained during refurbishment or dedicating new

. commercial grade breakers for licensees as evidenced by the history of three breaker issues involving Salem, IP2, IP3, and Clinton. The inspectors observed that the licensees found several discrepancies in the work thet the NLI technicians performed on their breakers. NLl ,

used the licensees findings as a basis for reworking the breakers and revising its procedures, j Refurbishment of GE Maane-Blast Breakers In'1996, NLI used this approach when it

< . refurbished a large number of General Electric (GE) Magne-Blast circuit breakers for Public  !

Service and Gas Company's Salem power station (Salem). Salem raised several issues during  !

! the refurbishment work and the licensee sent representatives to observe the work being performed to ensure the breakers met their requirements. The NLl Magne-Blast breaker _

refurbishment procedure is now on revision 15 and is quite detailed. By contrast, the NLI procedure for the Westinghouse breakers is not as detailed and has less than 5 revisions.

l.

i 15 l l

l

l Westinghouse Tvoe DB-50 Breakers for IP2 On July 9,1997, IP 2 contracted NLI to perform refurbishment of 6 Westinghouse Type DB-50 safety-related circuit breakers. NLI refurbished the breakers in accordance with its procedure NLi-TECH-P103, " Remanufacturing of Westinghouse 480 V DB-50 Circuit Breakers," which Coned approved before NLI began the work. The licensee performed their maintenance procedure on the first three refurbished breakers delivered to IP2, and found several discrepancies. As a result, Coned issued a "stop work order" to NLI on October 16,1997, until the discrepancies were addressed by NLI. NLI wrote Nonconformance Report (NCR) #53 to address the issues.

Coned issued a 10 CFR Part 21 notification on January 14,1998 (Event Notification 33535),

followed by a detailed letter on February 17,1998, describing three defects found in the NLl-refurbished breakers: (1) a disconnected jumper wire that prevented the breaker alarm switch from being reset from the control room, (2) intermittent binding of the manual closing mechanism roller bearing caused by wear, and (3)"G"-gaps, which affect contact pressure,  ;

were out of adjustment. The resolution to the findings in the NCR stated that the alarm switch jumper was disconnected during testing and there had previously been no explicit step in the procedure to reconnect it, but that this would be corrected. NLI reportedly learned from the OEM that there had been previous failures of bearings involving collapse of the inner race. The NLI procedure covered checking and adjusting the "G" gaps, but when tests are performed after the adjustment, there is a possibility for the "G" gap to change. NLI resolved these issues by revising the refurbishment procedure to add steps to reconnect the alarm switch jumper and check the G gaps after all the other adjustments are made. NLI stated that the procedure will also be revised to include replacing the roller bearing with a bushing per the OEM recommendation.

The licensee had determined that the defects identified would not prevent normal breaker operation. The disconnected jumper would not allow remote reset of the alarm from the control room, but it could be reset at the breaker. The roller bearing problem does not affect the l electrical closing operation and is only used during manual operation. The G-gaps, while out of  ;

tolerance, would not alone have prevented the breaker from closing. The breaker reportedly  ;

operated 20 times at NLl and met all of the timing requirements. NLl personnel informed the inspectors that they have not performed any other DB-50 breaker refurbishment and that there is no current contract to refurbish any other DB-50 breakers. Thus, the specific defects identified by Coned do not appear to be a generic concern.

Westinghouse Tyoe DS-416 Breakers for IP3 Documentation of the purchase, testing, dedication, and delivery of 11 new Westinghouse Type DS-416 circuit breakers by NLI for the New York Power Authority's (NYPA's) IP3 plant indicated that NLI procured the breakers from the local Westinghouse Electric Supply Company (WESCO) as commercial grade components and then dedicated them for nuclear safety applications. WESCO bought the breakers directly from Eaton/ Cutler-Hammer (ECH) which has been manufacturing the Type DS breakers since 1994. When the first two NLI-dedicated breakers were delivered to IP3, NYPA performed its preventive maintenance procedure as part of the receipt inspection. The NLI breakers did not meet the acceptance criteria in the NYPA procedure in several areas. As a result, the breakers were sent back to NLI and a NYPA representative visited the NLI facility to observe the l performance of the NYPA maintenance procedure on the breakers. Only two of the l discrepancies were determined to be nonconforming conditions by NLl: (1) the levering-in

)

l 16

l mechanism screw was inadvertently tightened during the final check , and (2) the levering-in hinge was wedged behind the face plate because the face plate was not properly installed. NLI agreed to add specific steps to its dedication procedure to address these two issues.

l Westinghouse Tvoe DHP Circuit Breakers for Clinton Power Station Illinois Power Comoany (IPC), issued a purchase order to NLI for the refurbishment of three Westinghouse Type DHP 4-kV circuit breakers for its Clinton Power Station (Clinton). When the licensee received the refurbished breakers, plant personnel performed procedure B 68-97-6, Revision 5, "Clinton ,

Power Station Westinghouse DHP Circuit Breaker Receipt inspection Checklist," and found I several discrepancies. The licensee also issued an interim 10 CFR 21 report on February,13, . )

1998, to inform the NRC that IPC and NLI were investigating the discrepancies for deportability, i and would complete the evaluation by April 15,1998. The inspectors discussed this report with l NLI and were informed that NLl Nonconformance Report # 60 addressed the concerns raised i by the licensee. NLl evaluated the discrepancies and determined that most of them were due '

to a difference in methodology for measuring tolerances and gaps. NLI determined its method of measuring gaps was correct.

NLI determined that there were 6 nonconforming conditions on the three refurbished breakers.

Three items (2 loose nuts and 1 loose terminal lug) are not explicitly contained in the NLl procedure and may have been inadvertently overlooked. Two cases of measurements being out of specification (anti-close interlock and the puffer dimension) were identified. The last nonconformance was the use of a micro-ohm meter instead of a Ductor@ to measure contact '

resistance, which gave an unacceptable reading. The contact resistances were subsequently measured with a Ductor and found to be within specifications. NLI agreed with the inspectors that there was a lack of detail in its refurbishment procedure in some specific areas and agreed to revise the procedure to provide detailed instructions in those affected areas.

b.3 1lse of inadeauatelv Dedicated Commercial-Grade Comoonents On January 15,1998, NRC inspectors identified two issues with the NLI DS-416 breaker dedication process that appeared to differ from the dedication process performed by Westinghouse Nuclear Services Division's (WNSD's) Repair and Replacement Services in Cheswick, Pennsylvania: (1) The pole shaft in the breakers being dedicated by NLI were of the commercial design in which the pole levers are welded to the shaft using two separate welds of approximately 120* each rather than a continuous weld of 100 as is used on the special nuclear-grade pole shafts used and supplied by WNSD; and (2) the direct trip actuator (DTA) used with Amptector programmable overcurrent trip systems installed on the NLI breakers was recognized to be of the commercial-grade type that only has glued joints (internally) to hold its permanent magnet and its pole pieces together instead of bolting the pieces together.

DS-416 Commercial Pole Shafts After finding the commercial grade pole shaft in a DS-416 breaker that had been supplied to (and retumed from) IP3, the inspectors informed NLI that NRC Bulletin 88-01, " Defects in Westinghouse Circuit Breakers," and W NSID (now WNSD)

Technical Bulletin 87-11 " Westinghouse Circuit Breakers, Type DS/DSL, Welds on Breaker Pole Shafts," (which was also attached to the NRC bulletin) directed licensees to inspect the pole shaft welds to determine if they were acceptable for operating plants to justify continued operation until such time that the pole shafts with questionable welds could be replaced.

17

The NRC bulletin and NSID-TB-87-11 were issued in response to a failure of a DS type reactor trip breaker to open in July 1987, during a manual trip demand from the control room. The RTB had bound mechanically in the closed position because the main drive link cam follower roller had become wedged between a raised segment of the close cam and the adjacent side frame plate. Excessive lateral movement of the main drive link and a broken center pole-lever-to-pole-shaft weld permitted the binding to occur. After the TB was issued, WNSD revised the weld procedure specification and weld inspection criteria for the pole shaft welds. W procures replacement pole shafts from a local supplier using the revised criteria and uses them to replace the commercial-grade pole shaft when it dedicates the breakers supplied by ECH.

In response to the inspectors' observation, NLI stated that its technicians had inspected the welds in accordance with NSID-TB-87-11. NSID-TB-87-11 stated that the pole lever weld fillets should be at least 3/16-inch, free of visible cracks or other defects and the weld beads should extend around the shaft continuously for at least 180-degrees of its circumference. The pole shaft welds examined by the inspectors had fillets that appeared to be of acceptable size and quality, but, as stated previously, had two 90- to-120-degree beads instead of a single 180-degree bead. NLI was not aware of the unique features of DS-416 breakers supplied by WNSD for Class 1E service because it had purchased the breakers from ECH as commercial-grade items and had not yet had any experience with WNSD-supplied DS breakers, such as through refurbishment. The NLI QA manager acknowledged the inspectors' concerns and immediately informed IP-3 personnel that they had received DS-416 breakers with commercial-grade DTAs, and with commercial-grade pole shafts.

To address the concerns with the potentially inferior pole shafts, NLI performed a series of tests (including a static load test and a 5000-cycle operational test) designed to confirm the adequacy of the welds on a pole shaft of the type installed in breakers built by ECH that they supplied to IP3. In addition, NLI analyzed the expected static loading and theoretical weld capacity of the pole lever welds and documented their analysis in Calculation C-048010-W,

" Pole Shaft Weld Evaluation." Revision 0 of this evaluation, dated January 16,1998, was attached to NLI NCR #62, dated February 4,1998, in which NLl documented the concerns with the pole shafts and DTAs identified by the NRC inspectors.

However, the inspectors found that the pole shaft weld evaluation contained several errors.

First, the calculation was based on erroneous static values for certain key breaker operating parameters. The force on each moving contact arm was estimated in the calculation based on the value of 95 pounds, reported in a test report, TIR-048010-4, attached to the calculation.

The inspectors noted that the reported value 95 pounds was only about one-fourth of the correct single-pole value for the combined force of all four pole base springs under proper compression of 332 pounds.

NLi's lead breaker technician demonstrated on the DS-416 test breaker the method he used to obtain the 95-pound value. The inspectors observed that the technician measured the contact force with a force gage by pressing the contacts together enough just to engage the arcing contacts and having the mains just touching, but not compressing the pole base springs behind l the stationary main contact fingers. When, at the inspectors request, the technician measured 18 l

l L

f t.

the force on one pole base spring, by compressing the proper amount as shown in the t

Westinghouse technicai manuals for the breaker, with its associated main contacts, the force developed by one of four pole base springs was about 90 pounds, nearly as much as NLI had thought was the amount for one pole.

Second, the inspectors pointed out to NLI that its calculation failed to take into account the force-amplifying effect of the location of the insulating link relative to the location of the main contacts. NLl's calculation had used its 95-pound value directly in the computation of the operating torque on the center pole lever. However, the main contact force for one pole (which should have been about 360 pounds for the test breaker to begin with) is amplified at the insulating link because the main contacts are about 5 inches from the moving contact arm hinge, or 2 inches farther out on the arm than the point at which the insulating link is connected to the arm at only about 3 inches from the hinge.

Third, the inspectors noted that the calculation failed to take into account the force and hence torque-multiplying effects of the dynamic impact loading during the breaker closing operation.

The inspectors recognized that the 5000-operation cycle testing NLl had conducted addressed this factor, but it was not considered in the calculation.

Finally, on the basis of the erroneous test data and flawed static analysis, NLl's pole lever weld evaluation asserted that its calculated theoretical weld capacity and the comparable torque exerted on the center pole lever weld in the static load test would be over four times its calculated pole lever operational torque; whereas, when the expected operational torque was more accurately determined, the weld capacity or test load might exceed the operational torque '

only by a factor of 1 to 2, if any.

Subsequent to the inspection, after the deficiencies had been explained to NLI, the vendor developed an improved calculation, Revision 1 to C-048010-W, dated March 9,1998. In this revised calculation, NLl computed the expected pole lever weld torque using a value of main contact force that was more consistent with the design. The revised calculation also employed an appropriate contact force correction factor based on measurements of the main contact and ,

insulating linkjoint distances from the moving contact arm hinge. Finally, the revised calculation used a value for the center pole lever torque lever arm that was measured with the breaker closed. While rigorous dynamic analysis of the forces and torques during the closing operation suggests that this may not be the worst case position, it should be nearly so.

Therefore, the inspectors determined that the revised calculation more accurately estimated at least the static torque on the center pole lever weld. The resultant design and test margin was closer to a factor of 2 rather than 4 as originally asserted.

In subsequent discussions of the pole shaft issue between the inspectors and WNSD, WNSD explained that they had no reason to dispute NLl's conclusions about the suitability (for nuclear safety-related applications) of the current vintage of commercial pole shaft produced by ECH.

Having previously examined the older commercial grade pole shafts that actually failed in service, the inspectors noted the welds on the new ECH pole shaft pole levers appeared to be of substantially better quality. WNSD further explained that it could only use, or supply, and stand behind its special nuclear-grade pole shaft that WNSD has made to its specifications, and 19

reportedly under strict controls, solely for this application because it is the only currently-produced DS pole shaft for which WNSD has verified the design and fabrication process by design testing and for which WNSD does production unit inspection and testing.

DS-416 Commercial Direct Trio Actuator Westinghouse Amptector@ and Digitrip@ solid-state electronic overcurrent trip systems effect breaker tripping through an electromechanical device called a direct trip actuator (DTA). These trip systems are made to be installed in i Westinghouse low-voltage metal-enclosed power circuit breakers (e.g., Type DS) as well as those of other manufacturers (e.g., General Electric Type "AK"). A DTA contains a permanent magnet, that holds a plunger in place (in the reset condition) against the force of an internal ,

actuating spring. The plunger is linked externally to a trip lever such that a trip signal from the {

Amptector counteracts the permanent magnet and allows the spring to move the plunger and )

actuate the trip lever. The original W DTA design used a single piece of metal bent in a U-shape for the magnet pole piece, with the permanent magnet placed at the bottom of the "U" In 1994, Eaton/ Cutler Hammer (ECH), who bought the DS breaker product line from Westinghouse and is the current manufacturer of commercial W Type DS and DHP switchgear, redesigned the DTA so that two separate pieces of metal replaced the single U-shaped piece and the magnet was placed along the side of one of the pieces. A high temperature glue was used to hold the magnet, the core, and the two metal pieces together.

After several commercial customers reported that DTAs failed in service, ECH modified the DTA design to include bolting the assembly together in addition to being glued. The modified DTAs can be identified by either a blue or yellow round sticker on the black plastic casing. The inspectors observed that the DTAs that had been mounted on the breakers being dedicated at NLI for IP3 were of the commercial grade quality glue-joint type. They had no stickers, and also had electrical leads protruding from their cases instead of terminals on the cases for leads from the Amptector or Digitrip units, which is another sign of a commercial-grade DTA. The  ;

commercial grade DTAs supplied by ECH with tneir DS breakers with the blue or yellow stickers (indicating both screws and glue holding the magnet and pole pieces together) have reportedly only been supplied by ECH to WNSD. Those supplied in breakers or separately through strictly commercial channels would not be expected to have the stickers and would be expected to be of the glue-joint only design.

If the DTA cannot be reset after tripping because some of its internal parts have come loose, it ,

can hold the trip shaft in the trip position and prevent the breaker from closing. NLl addressed these issues and presented their evaluation to the inspectors. One of the pole shafts in question was installed in a Type DS-416 breaker and cycled over 5,000 times without failure.

Inspection showed that there were no cracks in the welds. A DTA of the type in question was artificially aged in an oven and cycled several times, then installed on the test breaker and operated over 5,000 times (during the pole shaft tests) with no separation of the glue joints.

NLI contacted Eaton/ Cutler Hammer via E-mail and received a response that indicated only large-frame GE breakers fitted with Amptector or Digitrip overcurrent trip systems that use the DTA had reportedly experienced problems with the glued joints in the DTAs. No DTAs used in DS-416 breakers had experienced failures due to separated glue joints. Based on their own evaluation and testing and discussions with ECH, NLl concluded that the ECH-supplied pole shafts and DTAs were acceptable for use in DS-416 breakers in Class 1E applications.

Nevertheless the deficiencies identified constituted inadequate review for suitability of 20

application and design verification and hence, a nonconformance with respect to 10 CFR Part 50, Appendix B, Criterion 111," Design Control." (Nonconformance 999012298/98-01-03)

c. Conclusions Training and experience of NLI breaker technicians was weak; although, NLl's new training program, if effectively implemented, should improve technicians level of knowledge. However, NLl's lack of breaker maintenance and rebuilding experience, in conjunction with procedures requiring extensive revision before being satisfactory, has resulted in breakers having to be returned for rework when the discrepancies are identified by the licensees. Although NLI has taken corrective action and revised their procedures when problems have been identified by licensees, the inspectors concluded, and emphatically pointed out to NLI, that this trial-and-error method was unacceptable for a supplier of basic components and related services under what is supposed to be a QA program in accordance with 10 CFR Part 50, Appendix 8. The inspectors also pointed out that such heavy reliance on the licensee to detect errors substantially reduces the margin of safety maintained by all parties performing work as correctly as possible and verifying the work. It has created deviations or failures to comply that were evaluated under Part 21 requirements to determine if they could create substantial safety hazards, and in some case deemed reportable to the NRC.

The inspectors further concluded that NLl's failure to communicate effectively with the licensee when planning and specifying the work to be done was a weakness, as exemplified by NLl's not informing themselves of receipt inspection requirements that the breakers are expected to meet upon delivery to the plant. Had NLI obtained that information before performing the work, there would likely have been fewer breakers needing to be returned for rework.

Finally, the inspectors concluded that in its dedication of the DS-416 breakers for IP3, NLI had performed an inadequate review for suitability of application for the pole shafts. Even though NLI eventually inspected the pole lever welds using NSID-TB-87-11, not all the acceptance criteria were met and, absent the subsequent evaluation prompted by the NRC, IP3 would have had to continue the inspections mandated by WNSD and NRC Bulletin 88-01 until the pole shafts could be replaced with nuclear-grade pole shafts. On the basis of the original calculation with the deficiencies identified by the inspectors, the inspectors concluded that the subsequent evaluation that NLI had performed was an inadequate design verification of the DS-416 pole shaft welds. However, the inspectors ultimately concluded that on the basis of the substantially improved appearance of the ECH pole shaft welds, the satisfactory static and dynamic test results, and the margin determined in the revised pole lever torque load calculation (which should bound its remaining flaws), NLI ultimately provided reasonabla assurance that the welds were acceptable. On the basis of the reported satisfactory cycle test results and the fact that in-service failures in the commercial DTAs had only been reported for large-frame GE AK Type breakers, NLI provided reasonable assurance that the commercial grade DTAs as currently manufactured by ECH would be suitable for Class 1E service, but as thus far demonstrated, only in Type DS-416 breakers.

4. PERSONS CONTACTED A. Seiken President A C. Bell Quality Assurance Manager W. Malik Project Manager 21

6 ITEMS OPENED, CLOSED, AND DISCUSSED ltem Number Iygg Description Opened 99901298/98-01-01 VIO Failure to develop a procedure 99901298/98-01-02 NON (A) Failure to take corrective action for conditions adverse to quality 99901298/98-01-03 NON (B) Inadequate design verification of pole shaft 99901298/98-01-04 IFl Effectiveness of training 99901298/98-01-05 NON (C) Inadequate design procedure to calcu' ate seismic qualification life extension 99901298/98-01-06 NON (D) Inadequate training of QA inspectors 99901298/98-01-07 NON (E) Unqualified procedure for double bum process -

i 99901298/98-01-08 NON (F) Lack of acceptance criteria in procedure Closed 99901298/96-01-01 NON Inadequate dedication 99901298/96-01-02 NON Loss of equipment traceability.

l 22

_ . - - - _ _ _ _ - - _ _ _ _ _ _ - - -