ML20217N491

SER Pertaining to Cracking of EDG Lube Oil Piping at Vermont Yankee

ML20217N491
Person / Time
Site:	Vermont Yankee
Issue date:	02/27/1998
From:	NRC (Affiliation Not Assigned)
To:
Shared Package
ML20217N453	List: ML20217N448 ML20217N472 ML20217N491
References
NUDOCS 9804090112
Download: ML20217N491 (9)

Text

,

. l

'.^ l EVALUATION BY THE OFFICE OF NUCl. EAR REACTOR REGul.ATION PERTAINING TO THE CRACKING OF EDG LUBE OlL PIPING AT VERMONT YANKpp 1 INTRODUCTION

{

l Vermont Yankee (W) recently completed an operability determination for the emergency desel i

generators (EDGs) in order to address a degraded condition in the associated skid mounted 1 support piping. The wolds in the skid mounted piping were found to be less than full-penetration welds as assumed in the original design basis. The EDGs are required to remain operable during a design basis earthquake.

Several plants, including Crystal River, Minstone, and Vermont Yankee, have experienced problems with degradation of these welds during normal operation. Subsequent evalusten showed significant lack of penetration and generallack of quality in the welds. The Owners' Group is pursuing this matter and the three licensees have taken differing conective actions. In W's case, the licensee has replaced a section of the lube oil piping, added some addibonal supports, and performed metallurgical examination and destructive load testing of sample welds removed from the system. Based on the examination and testing of the piping, the licensee ,

completed an operability determination which concluded that the EDGs, although degraded,  !

remain operable for both normal operationalloads and seismic loads that would be experienced 1 in a design basis earthquake. In a TIA dated November 10,1997. Region I requested NRR's

> sosistance to review and evaluate the adequacy of the licensee's operability determination.

BACKGROUND At the present time, the identified deficiency involves a degraded condition related to welds on vendor-supplied skid-mounted carbon steel piping associated with EDGs, DG 1-1 A and DG 18 at W. Of specific concem are the welded joints in the lube oil and jacket cooling lines which potentialty have less than full penetration weld configurations. This condition is contrary to the vendor's fabrication drawings which indicate full penetration welds. The owners' Group disseminated information regarding a lube oil piping welded joint failure that occurred on one of the EDGs at Millstone Unit 2. The failure mode, at Millstone, was determined to be due to vibration-induced fatigus during diesel operation. Although the failure was not cata.L@, it resulted in a through wall crack in a weld joint resulting in a lube oil leak.

The main piping systems on the EDG skid consist of lube oli, water cooling and after cooling.

The lube oil discharge piping between the lube oil pump discharge and the oil filter is of special concem since this portion of the piping experiences the most vibration. Oillsaks occurred in a cracked weld in this piping section at Millstone and W. It consists of a 4.5 inch diameter,0.25-inch wall thickness carbon steel (ASTM A513) material. It has two elbows and a 5-inch diameter coupling which fits over the 4.5-inch diameter piping via rubber ferrule-type gaskets on both ends to allow thermal growth and vibration isolation. This section of piping and connecting welds have been replaced both at Milistone and W and all welds in this section are in compliance with ASME Code requirements.

9804090122 980310 PDR ADOCK 05000271 P PDR i

2 DISCUSSION The licensee performed a root cause investigation with assistance from the vendor, Fairbanks Morse (FM) in an effort to more clearly understand the original seismic qualification design basis for the skid mounted piping and the condition of welds on this piping (References 1 and 2). The original Ebasco EDG procurement specmcations, WNP-VI-IV G 1, Revision 2 fted March 27, 1969, required the vendor to deliver a seismically qualified package which ms' # specific requirements. The specifications required that the equipment be capable of performing its intended safety function during a seismic event. The seismic loads, discussed in the specifications, were static acceleration values which were consistent with the technology employed in the ig60s. -

- A walkdown and visual examination of the potentially affected skid mounted piping was performed at W on September 12,1997 by the licensee. Based on the visualinspection of the welds, it appears that the welds were most likely performed by employing a single pass seal bead with two additional passes over the single pass resulting in a double crown. '

The licensee notes that reference to a specmc piping code (i.e., ANSI B31.1) for the design of the skid mounted piping did not exist within specification WNP-VI-IV G-1, Revision 2. This was intentionally deleted in the specifications because FM took exception to the B31.1 code and utilized their intamal procedures / processes for the fabrication of the piping. FM stated that these intamal procedures / processes were equivalent to commercial standards and good industry practice and were deemed acceptable based on many years of successful inservice operation of their equipment. FM further stated that the seismic qualification methods utilized for the engines supplied to W (and also in general during the late 1960s and early ig70s) were to analyze the engine block itself and all other major skid-mounted components, such as heat exchangers, strainers, filters, etc. concentrating on anchorage requirements. The interconnected piping was not specifically analyzed. As years progressed, the development of more specific seismic qualification standards (IEEE 344-71 and -75) became available and a selected generic analysis of bounding piping geometry was performed for subsequent nuclear industry customers. As previously noted, this was not the case for plants of We vintage; therefore, no quantitative analysis exists. However, FM stated that the analysis would apply to W and pipnts of Ws vintage as the same design concepts were employed utilizing the same intemel procedures / processes and equivalent seismic loading. The specifications also required a hydrostatic test of the piping systems to 1.5 times the design pressure.

The diesels ste within the scope of Ws USl A 46 program. As part of the A 46 program resolution, a walkdown of the diesels was conducted in response to Gu.oric Letter 87 02,

" Seismic Qualification of Equipment in Operating Plants," utilizing guidana contained in the Seismic Qualification Utility Group's (SQUG) Generic implementation Procedure (GIP). It is noted that the procedural guidance contained in the GIP was developed from wsf ras of i experience data from the performance of diesel engines similar to We which have actually experienced earthquakes of magnitude in excess of Ws seismic design basis. The walkdowns, qualitative in nature and performed by trained seismic engineers, included consiceration of the skid mounted piping and focused on the existence of poteritial seismic-induced vulnerabilities including interaction effects related to proximity with other equipment, structural failure of non-l l

L'-

3 seismic components (commonly referred to as seismic II/I), and overall flexibility of lines. The licensee acknowledges that these walkdowns could not have foreseen or discovered the issue .

of weld penetration but they do provide supporting evidence that known seismic vulnerabilities related to the performance of the piping under seismic loading are not present.

The licensee's analysis indicated that the section of piping between the lube oil pump discharge and the oil filter was the most highly stressed and most susceptible to the fatigue failure due to vibration. This section of piping was chosen for stress analysis and was considered as a bounding sample of all skid-mounted piping configurations. In addition, the licensee discovered that this section of piping st W experienced a leak on weld no. 2 in 1995. AM,;i,g to plant -

records, the as-found condition was a slag inclusion of approximately 3/16 inches in the face of ws!d no. 2 w!!h several are strikes. The corrective action for this leak at that time was to remove the arc strikes, grind the weld and remove visible defects, and roweld using approved procedures and conduct a post weld liquid penetrant examination on this and other welds on this section of the piping. Although this section of piping was replaced in late 1997, concoms regarding the balance of the skid-mounted piping are the focus of this e'/aluation. Some of the  !

stars concems were partially resolved, others remain to be addressed. These are discussed i

laterin the evaluation.

According to FM personnel, as far as they are aware of, the cracked weld discovered at Millstone was the first reported failure in the skid-mounted piping of a FM engine. Similar ]

engines have successfully operated for many thousands of hours (some applications  !

continuously) at many commercial, military, and maritime organizations. Many of these engines l are routinely subjected to severe environmental conditions including temperature, vibration, and shock. Given this successful operating history, FM maintains it is reasonable to expect W s equipment will continue to support their safety function. In the stafs view, qualitative information from commercial and other facilities are not directly applicable for supporting long-term operability at W. However, the generally low magnitudes of seismic and vibratory load demands on the skid-mourited piping and the successful operating history cited above does provide adequate confidence for continued operation in the near term.

STRUCTURAL EVALUATION ,

The licensee considers that the W EDGs, although degraded, are still operable. A major aspect of the basis for continued operation was an analysis performed by the licensee to demonstrate the structural integrity of the skid-mounted piping under normal operation and during a seismic event. As stated earlier, the section of the piping between the lube oil pump discharge and the filter was determined to be the most highly stressed and hence it was the primary focus of this analysis. Other vulnerable sections of the skid-mounted piping were also analyzed.

The ADLPIPE computer code was used by the licensee to eva!uste the lube oil pump discharge piping. The loading conditions considered were intomal pressure, deadweight, thermal, vibration and seismic inertia. The piping system stresses were compared to the allowable values of the B31.1 Code and the ASME Standard OM-3 (References 3 and 4). Thrust loading from pressure was calculated manually and combined with the appropriate ADLPIPE stress re.sults and ASVE Code equations. In addition, vibration data monitored during actual engine operation was used as input into the piping model to address vibration e#ects. '

b.

~

4 Two ADLPIPE models were constructed in order to address and bound specific issues pertaining to the welded joints in the tubing sections. One model served to provide the t aseline data. It consisted of as-built piping geometry, and used the nominal wall thickness i,,dd d by the vendor. The second model used a wall thickness of 0.1 inches for the section of tubing from the lube oil pump connection up to the 5-inch diameter coupling. From the !Hnch diameter

' coupling to the filter, a wall thickness of 0.02g inches was used. These wall thicknesses were based on preliminary measurements of two partial penetration weld depths, which the licensee considers to be bounding values.

The licensee performed dynamic response spectrum analysis for seismic loading. The amplified response spectra for the floor above the EDGs in the turbine building, were utilized using the ASME Code Case N4,11 damping. The spectra are considered conservative by the licensee since the diesel and thii associated skid piping (i.e., lube oil discharge tubing) are on grade at the base mat elevation, which will only experience ground acceleratums instead of the amplified response of the building structure. Dynamic response spectrum analysis was also performed for the assessment of stresses due to vibration. Response spectra in the form of spectral accelerations were obtained from actual vibration data monitored at several locations on the i lube oil discharge tubing sections. The worst case response was applied to the ADLPIPE model. I The resulting calculated stress values due to vibration were assessed using the guidelines of ASME Standard OM 3.

The remainder of the skid-mounted EDG piping was reviewed by the licensee for potential structural weak links in the system. The review was accomplished by visual field observation and manual calculations where required. The potential weak links were indicated either by a lack of physical structural support, or at weld joints that could be affected by the behavior of mechanical couplings. As stated earlier, the piping wall thickness was uniformly decreased in the ADLPlPE model to account for the depth of partial penetration weld joints in the system. The 4.5 inch x 0.25 inch piping wall was decreased to 0.1 inches (including the elbow weil thickness) based on preliminary metallurgical measurements of the welded jdnts. No credit was taken for the additional strength contributed by the weld crowns.

A vibration-induced failure mode characterized by fatigue is considered the most likely potential failure mode by the licensee. Such a failure would occur over a long period of time due to propagation of an existing flaw during engine operation resulting in a through wall crack with subsequent leakage. This, in the licensee's view, is not charactertred as an abrupt or catastrophic failure. According to the licensee, given the successful operational history of W's diesels, as well as those at other FM supplied units, it is unlikely that imminent failure resulting in complete severance of the piping will occur. In addition, the licensee argues that should failure occur, it would likely be characterized as a leak which would be detectable through alarm functions or operator surveillance. With the backup diesel as well as the available W tie line, a j safe plant shutdown could be accomplished if required. The staff concurs with the licensee ifit '

can be established that the wold geometry assumed in the analytical modelis indeed bounding.

l However, in the staffs view, this has not been fully established yet. I i'

The licensee has also performed load tests on the removed section of the piping from the W EDG which was replaced. Weld numbers 4,6 and 8 on tube oil pump discharge piping were selected for testing. Both axial and bending loads were applied. The results of the load tests i

4 ,

4 5 -

demonstrated that the as-installed wolds will withstand loads groter than those which cause stress in the piping to exceed the yield strength of the piping '.sterial. Also, the tests demonstrate that these wold geometries have significant stiength with respect to the design loads and the B31.1 Power Piping Code allowable stresses. However, these results are I applicable to the three specimens tested and there is no assurance that these are boundmg geometries. It is the staffs understanding that additienal tests are being performed by the +

Owners' Group on piping replaced in FM diesels at other nuclear plants.

Based on a review of the licensee's analyses, the staff noted that the piping stresses are -

generally low for the applied loads and assumed weld geometries with a significant margin to B31.1 Code allowables. The notable exceptions are the elbow locations where thrust loadings produce increased bonding stresses. At these locations, while the stresses are high, they are within allowable limits. The maximum stresses occur at the weld joint No. 2 on the elbow located just outboard of the lube oil pump discharge.

The staff noted that all loadings satisfy Code requirements at the maximum operating pressure of 36 psi with the reduced wall of 0.1 inches. At design pressure, the reduced tubing wall of 0.1 inches does not meet the allowable stress values of the ASME Oode Section Ill, Subsection NB Equation 11 (stress due to sustained loads). However, when seismic loading is considered, the tubing stresses do satisfy the higher allowable stress value of Code Equation 12 (stress due to occasionalloads). Since both Equations 11 and 12 of the Code need to be satisfied, the staff identified this as a concom.

In response to staff concem regarding the noncompliance with the ASME Code requirements, additional calculations for this load case we,re performed by using as-found data at the crihcal location. Wold joint No.2, which is considered a criticallocation, was shown to conelate to a mean value of 0.15 inches for the "B" diesel lube oil piping. This is considered less conservative but more reprszentative of actual condit!ons by the licensee. The stress results for this load case were shown to meet ASME Code Section lli Equation 11 allowables considering a maximum upper bound pressure equivalent to the relief valve setting at 80 pol. The staff finds the assumed value of 0.15 inches in this calculation questionable, particularly in light of concems raised in the metallurgical evaluation. Hence, non conformance with the ASME Code requirements remains a staff conoom.

METALLURGICAL EVALUATION The replaced piping is made of carbon steel (ASTWI A513). The licensee performed a '

metallurgical examination of six weld joints (#2, #3, #4, #6, #7 and #8) located in the piping section removed from the emergency diesel generator *B' (EDG-1-B). The licensee's stress analysis has shown that the replaced piping section is most susceptible to cracking because the engine operating stress in this piping section is higher than in any other piping section. The l highest stresses are located at weld joint #2. Florescent 1.iquid Penetrant examination was performed on the OD surface of the subject piping section. With the exception of an OD surface crack about 5/8 inch in length that was found at v eld joint #2, no other OD surface indication was detected. Each of the six weld joints was quarter sectioned for a total of 48 cross sections.

Each cross section was polished and etched for metallurgical examination and the measurements of weld penetration. All six weldjoints were shown to be partial penetration

6 welds, each weld geometry was characterized by poorjoint fit up and without any weld end preparati'on. The reported minimum weld penetration in six weld joints varied from 17.4% to 58% of the pipe wall thickness. During the metallurgical examination, a through wall crack was identified in weld joint #2, however, its length on the ID surface was not measured. The through wall crack did not leak during opemtion. Small cracks initiated from the weld root ID surface with a depth of 0.057 inches and 0.000 inches were found in weld joints #3 and #6, respectively. The licenses performed a failure analysis for weld joints #2, #3 and #8 by examining the fracture surfaces with Scanning Electron Microscopy (SEM) and other techniques. Heavy oxides were observed on the surface of the short cracks at weld joints #2 and #3 and the initial portion of the -

through well crack at weld joint #2. The licensee reported those fatigue striations associated with the cracks were found in weld joints #2 and #3. Each observed crack was reported to be connected to a slag inclusion at the weld root ID surface. The licenses stated that a weld repair was performed at weld joint #2 during a 1996 outage because leakage was found at this weld joint.

The licensee suggested that the observed small cracks at weld joints #3 and #8 and the initial portions (about 0.06 inches) of the through wall crack at weld joint #2 are fabrication defects, initiated by shrinkage stresses resulting from cooldown of the fabricated welds. From the metallurgical examinations it appears that the small cracks are not active since it is filled with heavy oxides. The licenses stated that the measumd vibrational piping stresses are small (below the threshold stress) and that the vibrational stresses alone will not be able to cause cyclic crack growth. Therefore, the licensee attributed the major drivmg force for the crack propagation at weld joint #2 to tw the residual stresses resulting from the 1996 weld repair. The  !

staff agress with the licensee's assessment that the residual stresses play an important role in  !

promoting the crack growth.

l The licensee performed an evaluation using Linear Elastic Fracture Mechanics (LEFM) similar to l

the method described in GL 90-05 to determine the limiting flaw size at which the structural integrity of the cracked piping wold can be maintained. *PC-CRACK," a computer code developed by Gtructuralintegrity Associstes (SIA) was used for the evaluation. The limiting flaw size was determined using two separate LEFM crack models. One model assumes a through-wall circumferential crack in a cylinder under tension and bending and the other model assumes l part through-wall circumferential inner diameter (10) surface crack in a cylinder under tension.

The results of the evaluation have shown that the limiting flaw sizes in both models are significantly larger than the 0.625 inch circumferential crack found in weld joint #2. The licensee's evaluation is reasonable in that it confirms the structural integrtty of the degraded weld joint #2.

i EVALUATION FINDINGS Based on its review of the licensee's analysis, the staff has ideW the following findings I

. related to both structural and metallurgical evaluations:

[.q 7

(1) Mmimum Wald Penetration The metallurgical examination was performed on six weld joints in the removed section. The reported minimum weld penetration in those six welds varied from 17A% to 58% of the pipe well thickness. Due to the lack of proper weld end preparation and joint fit up, there is no assurance that the minimum weld penetration in the remaining lube oil piping will fall within that range.

(2) Wald Shrinkape Crack ,'

Shrinkage cracks were observed in 50% of the weldjoints examined (three out of six). The depth of the observed shrinkage cracks varied from 0.000 inches to 0.000 inches. The initiation of the shrinkage crack is knotyn to depend on the welding heat input, weld geometry, joint fit up and the weld cooling rate. Since these conditions tend to very from i weld to weld, it is difficult to make a reasonable estimation regarding the limiting shrinkage crack size that will be initiated in the remaining skid-mounted piping. i

. I (3) Wald Residual Strannes l l

For the through wall crack found in wold joint #2, the licensee attributed the main driving  !

force for the crack growth to the presence of the residual stresses resulting from the weld repair performed in 1996. The licensee considered this weld to be an atypical weld because this is the only weld having a record of weld repair. The licensee did not discuss the root cause for the observed leakage at weld joint #2 because failure analysis was not performed on this weld prior to repair. Based on the fact that the weld joint #2 was not repaired when the leakage was observed in 1996, it is apparent that the driving' forces for the through-wall cracking are the as-fabricated weld residual stresses and the stresses generated unoer j normal operating conditions. Therefore, the possibility exists that similar cracking may occur -

at other weld joints on the skid-mounted piping. l (4) Crack Growth  !

Fatigue striations were found on the fracture surface at weld joints #2 and #3. This i observation tends to support the contention that the cyclic crack growth is the major mode of j crack growth in the lut:s oil piping system. Although the vibratbnal stresses have shown to  !

be small on the affected piping, the cyclic loading resulting from the on-off t ansient when 4 performing the testing of the diesel generators may provide an appreciable driving force for fatigue crack growth. In any event, the leakage found at weld joint #2 clearly demonstrates that the initial shrinkage crack can become active and grow through wall with the existing as- 4 fabricated weld residual stresses under normal operating conditions. The observed crack growth rate at weld joint #2 does not appear to be very high since the through wall portion of the crack is only about 5/8 inch in length after a period of plant operation close to one fuel cycle. However, it is still necessary to evaluate the potential crack growth in the remaining  ;

skid-mounted piping to ensure its long term structural integrity. '

l

r g (5) Piping structuralintegrity in the piping structural analysis discussed earlier, the assumed value of 0.15 inches for the pipe wall thickness appears to be highly uncertain in view of the concems related to the minimum weld penetration, weld residual stresses and crack growth. The results of the structural analysis do not meet the requirements of the ASMbi Code Section 111 at the design pressure when a reduced wall Mckness of 0.1 inches was used as a bounding well thickness.

(6) Fatigue evaluation '

The fatigue evaluation is based on comparison of stresses due to steady state vitration and selsmic inertia with the requirements of ASME Standard OM-3. The analysis does not take into consideration the cumulative effect of the vibrationalloading resulting frorn transients during the monthly testing of the DGs. In the absence of quantitative data on the cumulative fatigue damage the degraded welds may have sustained to date, they remain vulnerable to failure during a seismic event.

(7) Hydrostatic test of the piping system The licensee's reliance on pressure testing of the replaced section of the piping as well as {

the initial qualification hydrostatic test of the piping system to 1.5 times the design pressure does not provide assurance against fatigue failure due to cyclic loadings experienced by the affected piping.

CONCLUSION Based on its review of the licensee's operability evaluation, the staff concludes that there is reasonable assurance regarding short term operability of t% BDGs contingent upon compensatory measures that the licensee should establish. The NRR staff recommends that the skid-mounted piping should be inspected with visual examinations during and after each EDG testing. This conclusion regarding the adequacy of the short term operability is based on the following factors:

The replacement of the most highly-stressed portion of skid-mounted piping by the licenses such that all weld joints are in compliance with the ASME Code requirements; The demonstration by the licensee via structural analysis, me'.anurgical examinations, data monitoring and load testing that the stresses and vibration isvols in the remainder of the akid-mounted piping are generally low; The likelihood of a failure modo characterized by a leak and the ava!!sbility of a backup diesel and W tio line, Augmented inspection during EDG testing, and I

l

e, , . -

9

\

1 The observed crack growth at the most highly-stressed wold joint which indicated that accelerated crack growth is not likely to occur in the near term.

In the evaluation findings, the staff has identified a number of concems in the structural and metallurgical analyses performed by the licensee. A credible crack growth evaluation cannot be  ;

performed without a quantitative knowledge of weld penetration and the shrinkage crack size.

Therefore, the long-term structural integrity of the remaining skid-mounted piping cannot be '

assured based on the information currently available. The NRR staff recommends that Region I

• review the conclusions of this evaluation and the identified concems regarding the ie,4 ' ai.

operability of the EDGs with the licensee to ascertain the licensee's intentions regarding its corrective action.

1 REFERENCES

1. Letter, S. Goodwin, Vermont Yankee, to K. Jabbour, NRC, dated November 4,19g7, s@ porting ooerability of V'l 5DGs.  ;

2. Letter, Coltec Industries-F&onnks Morse to USNRC, dated December 18,19g7, related to weldments on the FM EDGs.

i

3. ANSI B31.1-1977 Edition " Power Piping."

4. ASME O&M Standard OM 3-1982," Requirements for Prooperational and initial Start-up Vibration Testing of Nuclear Power Plant Piping Systems." i Principal Contributors:

J. Rajan W. Koo 4

9 i

l l