ML20100L401

Suffolk County & State of Ny Proposed Findings of Fact & Conclusions of Law on Emergency Diesel Generator Contentions.Certificate of Svc Encl

ML20100L401
Person / Time
Site:	Shoreham File:Long Island Lighting Company icon.png
Issue date:	04/15/1985
From:	Palomino F, Scheidt D KIRKPATRICK & LOCKHART, NEW YORK, STATE OF, SUFFOLK COUNTY, NY
To:
References
CON-#285-560 OL, NUDOCS 8504160603
Download: ML20100L401 (110)

Text

5 6c April 15, 1985 00 M ETED

)- USNFC UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION /H 16 A11 :33 6FricE :s , r t rit M R '

) Before the Atomic Safety and Licensing BoGC6dEipjG A SEgy ct; enANCH

)

) In the Matter of )

)

LONG ISLAND LIGHTING COMPANY ) Docket Nt.. 50-3 22- OL

)

(Shoreham Nuclear Power Station, )

Unit 1) )

) )

SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT AND CONCLUSIONS OF LAW ON EMERGENCY DIESEL GENERATOR CONTENTIONS

)

Fabian G. Palomino, Esq. KIRKPATRICK & LOCKHART Executive Chamber, Room 229 1900 M Street, N.W., Ste. 800

-Capitol Building Washington, DC 20036

-Albany, New York 12224

)

Special Counsel to the Governor Attorneys for Suffolk County of the State of New York

{$0"*IOo2R$$8$h22 O e

s

SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT AND CONCLUSIONS OF LAW ON EMERGENCY DIESEL GENERATOR CONTENTIONS TABLE OF CONTENTS (a) SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT AND CONCLUSIONS OF LAW ON EMERGENCY DIESEL GENERATOR CONTENTION (CYLINDER BLOCKS).

Page I. INTRODUCTION................................................... 1 A. Scope of Cylinder Block Portion of Decision.................................................. 1 B. Background to Litigation of Cylinder Block Adequacy............................................ 1 II. DECISION....................................................... 6 III. DISCUSSION.................................................... 14 A. EDGs 101 and 102 are Unfit for Nuclear Service Because of the Ligament Cracks................... 14

1. Operating Experience at Shoreham Discloses that Ligament Cracks do not Arrest as LILCO Claims They do............................................. 15

11. LILCO's Finite Element Analysis that Cracks will Arrest at a Depth of 1-1/2 Inches is not Reliable...................-........................ 18 111. The Ligament Cracks Themselves could cause Engine Failure.......................... 23 B. The Ligament Cracks Lead to Stud-to-Stud Cracks which may cause EDG Failures....................................... 25

-i-4 e

s

- , . , w,-p, c '--,-em,-v-- - - - ~ - - -

i. Stud-to-Stud Cracks are likely.to Initiate in EDGs 101 and 102 because they hav e Ligament Cracks . . . . . . . . . . . . . . . . . . . 26

11. The Evidence does not Establish s!

that the Physical Properties of the Block in EDGs 101 and 102 are Superior to EDG 103..............................~30 111. The Evidence is Insufficient to Establish that the Load Excursion caused Additional Damage to EDG 103 or that the Damage to EDG 103 would not have been Disabling....................... 35 iv. FaAA's Cumulative Damage Analysis is not a Reliable Index of Crack Behavior............................................ 37

v. A Stud-to-Stud Crack could be Disabling........................................... 50 C. The Possibility of Circumferential Cracks also Renders EDGs 101 and 102 Unfit for Nuclear Service......................................... 52 D. The Evidence is not Sufficient to Establish that the Replacement Block in EPG 103 is Reliable...............,................................. 61 E. The Blocks are not Suitable at LILCO's Proposed Qualified Load of 3300 kW....................... 63 (b) SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT AND CONCLUSIONS OF LAW ON EMERGENCY DIESEL GENERATOR LOAD CONTENTION Page I. INTRODUCTION................................................... 1 II. DISCUSSION..................................................... 2

A. LILCO and the Staff Have Applied a Novel Method for Complying with GDC 17 Requirements which Departs from Standard Practice and is

[ Inadequate................................................ 2

l. The Load Combination Methodology Used by LILCO to Estimate the MESLs Fails to Account For Inaccuracies in Nameplate Ratings and Load Measurements.................................... 6

2. The Qualified Load does not Encompass

!' Cyclic and Intermittent Loads....................... 12 i

d

3. The Qualified Load does not Encompass the Operator Added Load............................. 15 B. The EDG Qualification Test Run was Inadequate to Justify a Qualified Load of 3300kW.................... 21-III. CONCLUSION.................................................... 22 t

(c) SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT AND CONCLUSIONS OF LAW ON EMERGENCY DIESEL GENERATOR CONTENTION l

(REPLACEMENT CRANKSHAFTS (3300 kW))

I. DISCUSSION..................................................... 1 II. CONCLUSION.................................................... 12

-lii-

I. INI'ROIXETION A. Scope of Cylinder Block Ibrtion of Decision.

'l.- 'Ihis portion of our Decision addresses the Suffolk (bmty and the State of New York contentionl_/ that the Shorehm energency diesel genera-tors ("EDGs") fail to satisfy General Design Criterion ("GDC") 17 because r their cylirrier blocks are not adequate.2_/ We-address other parts of the EDG contention elswehere.

B. ' Backgromd to Litigation of Cylinder Block Adequacy.

2. 'Ibe Shorehm EDGs are Transmerica Delaval, Inc. ("TDI") model

_ DSR-48 diesel engines with 8 cylinders in line, having a 17-inch bore and 21-inch stroke. _ 'Ihe EDGs constitute the onsite electrical pwer systen for 1

the Shorehm plant and are therefore intended to provide reliable onsite emergency power to the Shorehm plant in confonnity with General Design :

Criterion 17,10 C.F.R. Part 50, Appendix A, GDC 17. Hubbard and Bridenbatgh, ff. Tr. 23,826, at 12, 14.

l.

I l_/ Ebr the sake of simplicity, these findings will generally refer only to the Cbmty in discussing the evidence and psitions jointly spon-sored by the (bmty and the State.

i' 2,/ Bf Stipulation dated January 14, 1985, the parties advised that Suffolk (bmty no longer seeks to disqualify the blocks on the basis of the canshaft gallery cracks. LIIf0 Ex. E-67. Accordingly, otr de-cision deals only with other types of crackr fomd in the blocks of the EDGs.

3. . Cracks in the tops of the EDG cylinder blocks were first i . discovered in February 1984. Tr. 24,603-04 (Schuster)3_/ All three blocks had 'so-called lignent cracks,-located between the cylinder tore and sttd i hole in a radial / vertical plane. Id_. at 13-14. EDG 101 had 13 lig nent 1

l cracks, EDG 102 had 18 ligment cracks and EDG '103 had 21 lignent cracks at the time of the inspections. LIICO Exs. B-16, B-17 and B-18.4_/ In ad-

' dition to those ligaent cracks, EDG 103 had a attd-to-sttd crack 1.6 inches deep located between sttd holes at the exhaust side, cylinders 4-5,

. in a radi'il/ vertical plane. Wells, et al., ff. Tr. 24,372, at 14-15; LIICO Ex. 3-18. EDG 103 also had 7 surface " indications" (cracks which were not deep encogh to be measurable), five of which occurred in stud-to-sttd loca-tion:s and two of which were located between a sttd hole and the outer per-imeter of the block top. Id.

4

4. Following discovery and maIping of the cracks, EDG 102 was oper-

cted through 100 starts to loads greater than 50 percent. Wells, et al.,

t 3] 'Ihe blocks were inspected by liquid dye penetrant, eddy current and visaal exmination. Maps of the sizes and locations of the cracks are set forth in LIICO Exs. B-16, B-17 and B-18 for EDGs 101, 102 and 103, respectively. Wells, et al., ff. Tr. 24,372, at 13.

! 4/ 'Ihose ligment cracks varied in depth,. with the ones in EDG 103 being L

most extensive and severe. 'Ihus, EDG 103 had 18 ligment cracks having a depth equal to 1 inch or more, and 15 of those 'were 1-1/2 Inches deep. LIICO Ex. B-18. EDG 102 had 10 lignent cracks with a depth of 1 inch or more, and 4 of then were 1-1/2 inches deep. LILCO

. Ex. B-17. EDG 101 had 6 lignent cracks having a depth of 1 inch or i more, and 1 of then was 1-1/2 inches deep. LIICO Ex. B-16.

(

i.

I

h ff. Tr. 24,372, at 15. Based on subsequent eddy current examination, LIICO concitried that there had been no discernible extension of cracks on the EDG 102 block. Id. 'Ihat conclusion is suspect because it is based upon in-cpection of the sttris at only one cylinder, Cylinder 7. See Tr. 24,411 (Johnsan); LIICO Ex. B-21.5/ EDG 102 was operated for 110 hours0.00127 days <br />0.0306 hours <br />1.818783e-4 weeks <br />4.1855e-5 months <br /> during that testing, with 16.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> at or above full load of 3500 kW. LTTCO Ex.

B-14.

5. Between M rch 11 and April 14, 1984, EDG 103 was subjected to ad-ditional operational testirg, but for a longer time at higher loads (incitriing 38.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> at or above 3500 kW) than the EDG 102 testing. LIICO Ex. B-15. On April 14, the plant experienced an abnonnal load excursion in which the power denand exceeded the EDG capacity for approximately 25 see-onds, causing the engine to slow to 390 rpn. S. C. Ex . 7, at 5 ; Tr .

24,656-57 (Yotrigling) . After that time, the abnormal load was dropped and the engine ran for 10 minutes before being shut down. S.C. Ex. 7, at 5.

After restarting and continued testing at 3900 kW for a short time, the op-erator noticed oil seeping frcm a crack running down the front of the block at Cylinder N:. 1, and testing was stopped. Tr. 24,661-62 (Youngling) .

The crack had not been noticed before the engine was restarted. Tr. 24,663 (Yotrigling) .6_/ .Ihe crack was later measured to be 4.4 inches long at the 5/ Althotgh Cylinder 7 arguably has the worst cracks based upon the EDG 102 crack map, we cannot asstane that other cracks in that block did not propagate during the 100-start test, particularly because the worst cracks in EDG 103 aIpear to be between cylinders 4 and 5. See LIIf0 Exs. B-18, B-25.

6] No one inspected the area before restarting the engine so LIIf0 as-serts that it "might" have been there. Id. Based on the record, we (Ebotnote cont'd next page)

p=

front surface of the block (no one recalled itt, depth at the stud hole) .

Tr. 24,668-69 (Wells, Schuster) . EDG 103 suffered such severe crack datage that LIICO and its consultant, Failure Analysis Associates ("PaAA"), con-citded that the block could not be repaired. Tr. 24,665-66 (Youngling, Wells) .7_/ . 'Ihus, LIICO decided to scrap that cylirder block and replace it with one of a more recent, modified design. Tr. 24,665 (Youngling) .

6. Following the EDG 103 block failure in April 1984, the block was reinspected and the cracks were mapped again. LIILO's original crack map fran that reinspection showe'd the nunber of cracks had increased frm 26 to 40, with 26 liganent cracks,12 sttd-to-stud cracks and 2 cracks running frcin sttd holes to the outer perimeter of the block (in the locations where indications had' earlier been noticed). S.C. Ex. 7, at 5-6, 14. Nearly all of the liganent cracks were measured to be at least 1-1/2 inches deep and 5 measured 2-1/2 inches deep. Id_. 'Ihe deepest sttd-tc>-stud crack measured 5-1/2 inches deep, another measured 3.9 inches deep, and a third was 1-1/2 inches deep. Id,.

(Ebotnote cont'd fran previous page) have no basis for coneltding een it initiated or how fast'it proIn-gated.

i 7/

'~

It appears that the reduced mechanical qualities of EDG 103 did not play a part in the decision -- at least Mr. Yotngling was unable to say that they did. Tr. 24,666.

7. After LIILO filed its original written testimony in this proceed-ing on August 14, 1984, it undertook further analysis of the danage to the scrapped EDG 103 block by destructive exanination. Rau and M chob (Sup.),

ff. Tr. 24,372, at 10. 'Ihat destructive exanination as limited to the deepst sttd-to-sttd crack, then believed to be 5-1/2 inchea d=p, a:rl dis-closed that it was 3 inchen deep instead, prmpting LIIro to concitde that its earlier measurements of the other cracks in EIE 103 were incorrect.

Id., Tr. 24,440-43 (hchob, Rau) .8_/ As a result, LIICO revised its esti-mates of crack depth in EI:G 103 as of September 22, 1984, so that only one liganent crack is now estimated to be 2-1/2 inches deep and the 3.9-inch deep sttd-to-sttd crack is now estimated to be 0.85 inch deep. LIICO Ex.

B-25, as revised.9_/ 'Ihe depth of the 4.4 inch sttd-to-sttd crack at the front of the engine renains the same. Id_.

8. During its destrtctive testing in Septenber 1984, LI1CO found a third type of crack in the original EDG 103 block - ciretznferential cracks located in the corner formed by the cylinder liner landing and the cylinder 8/ LIIro ascribes the cause of the earlier measurement errors to be ex-cessive quantities of Widmanstaetten graphite, a degenerate microstructure sanetimes formcd in cast iron, which distorted its eddy current readings. Tr. 24,442 (Rau) . 'Ihis decision will discuss the significance of LIIro's evidence concerning the existence of Widmanstaetten graphite in EIU 103 in greater detail below.

9] Ebr reasons discussed below, we concitde that these revised crack depth measurenents are not necessarily accurate and that the depths of the cracks in EDG 103 may lie sanewhere (we have no my of determining where) betmen those reflected in the tm crack maps.

ootnterbore. Rau and mchob (Sup.), ff. Tr. 24,372, at 11. Such cracks had been known to exist in other TDI DSR-48 diesel engines, but had not been discovered in any inspection of the Shoreham EDGs prior to Septenber 1984. See S.C. Ex. 7 at 2-5. LIICO estimates the maximtm depth of the circtuferential cracks in EDG 103 to be 3/8 inch. Rau and Mchob (Sup.),

ff. Tr. 24,372, at 11.

II. DECISION

9. GDC 17 requires that the EDGs: -

"Erovide sufficient capcity and capability to assure that (1) specified acceptable fuel design limits and design conditions of the reactor coolant pressure boundary are not exceeded as a result of' anticipated operational occurrences and (2) the core is cooled and containnent integrity and other vital functions are maintained in the event of postulated accidents." 10 C.F.R. Part 50, App. A, GDC 17. (Emphasis added.)

Because GDC 17 provides no express ernpirical standards for evaluating the capeity and capability of the EDGs, the normal and usual application of GDC 17 in past practice is especially pertinent.

10. Normally, the adequacy of energency diesels under GDC 17 is de-termined by reference to their continuous duty and short-time ratings.

Thus, IEEE Standard 387-1977_0_/

1 provides that emergency diesel generators l

l_0/ 'Ihis IEEE Standard is referred to by Regulatory Guide 1.9, published by the NRC's Office of Standards Developnent concerning selection of emergency diesel generators for nuclear powr plants, as delineating (Ebotnote cont'd next page) 6

Celected for nuclear power plants ". . . shall have continuous and short-time ratings which shall' reflect the output capabilities of the die-sel generator unit [s]. . . ." LIIf0 Ex. C-4 at { 5.2.1, 10-11. W e kre-word to IEEE Standard 387-1977, though not officially a part of the Stan-

. dard itself, explains its purpose as follows:

"he IEEE has developed this docunent to provide the pria::ipal design criteria, design features, qualifica-tion considerations, and testing requirenents for indi-

, vidual diesel-generator units . . . used in the standby power supply of a nuclear facility, which canply with the melear Regulatory Otmnission's Code of Federal

Regulations (10 CFR 50)."-

"he principal concerns are the ability of the diesel-generator unit to operate at design loads whenever necessary during the '

life of the plant. . . ." Id. at b reword.

Se Standard defines contintous and short-time ratings as follows:

"3. 7.1 contintous rating. S e electric power output captbility that the diesel-generator unit can maintain in the service envirorinent for 8760 h[ours] of operation per (ocentn) year with only scheduled outages for maintenance.

(Footnote cont'd fran previous page)

. . . principal design criteria and qualification testing require-ments that, if followed, will ensure that selected diesel generator units will meet their performance and reliability requirenents." Reg-ulatory G.11de 1.9, Rev. 2, Decenber 1979, at 1, { B. Although canpli-ance,with regulatory guides is not mandatory, they reflect acceptable methods of canplying with the NRC's regulations. Id. at explanatory footnote, at 1.

L

3.7.2 short time ratirg. 'Ihe electric power output capability that the diesel-generator unit can maintain in the service envirorrnent for 2 h[ours] in any 24-h[our] period, without exceeding the manufacturer's design limits and without reducing the maintenance in-terval established for the continuous rating." Id. at 9.

11. Consistent with the foregoirg practice, the Shoreham FSAR cpecifies that the EDGs must have sufficient load-carrying capability to satisfy a corn.inuous and overload perfonnance ratikg. 'Ihe FSAR for a nu-clear power plant presents the utility's explanation of how it intends to canply with the regulatory requirements for safety. 'Ihus, as of the time the hearings in this proceedits began, Section 8.3.1.1.5 of the FSAR required each EDG to be rated to operate continuously (8760 hours0.101 days <br />2.433 hours <br />0.0145 weeks <br />0.00333 months <br />, or one year) at full load of 3500 kW and for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> per every 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> at over-load of 3900 kW. S.C. Ex. 80 at { 8.3.1.1.5; Hubbard and Bridenbaugh, ff.

Tr. 23,826, at 14-16. " Continuous service" means that the EDGs are capa-ble of rmning continuously except for shut-downs necessary to perform scheduled maintenance. Tr. 26,162-63 (Berlirger) .

12. LIIro presented no evidence to support a finding that the cylin-der blocks are adequate to permit operation of the EDGs continuously for one year at 3500 kW and for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> per every 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> at 3900 kW as cpecified in the FSAR. Indeed, Dr. McCarthy admitted that PaAA cannot ctate that the EDGs are capable of operating at the FSAR-specified levels of perfonnance without experiencity cracking in the blocks to the extent

)

that operation would be impaired. Tr. 24,896-98. LIIr0 presented no other testimony on this subject. Instead, LIICO asserted that the proper stan-dard for evaluating the adequacy of the cylinder blocks is whether they are likely to be able to survive operation through one loss of offsite powr coincident with loca of cooling accident ("IDOP/I4CA") event. Tr.

)

24,822-23 (Wells, Yomgling) .l.1/

13. According to LIICO's calculations, the maximun loads the EDGs will experience during a IDOP/LOCA are 3429 kW for EDG 101, 3383 kW for EDG 102 and 3881 kW for EDG 103. LIIr0 Ex. B-51, at 6-8; Tr. 25,325, 25,382 (Yomgling) . LIICO estimates that those loads would be imposed during the first 12 minutes of a IDOP/IDCA and w3uld gradually be reduced thereafter dom to' a level equal to approximately 75 percent of the full load of 3500 kW. Tr. 25,370 (Yomgling) . Total projected EDG operation during a IDOP/LOCA is estimated to last 7 days (168 hours0.00194 days <br />0.0467 hours <br />2.777778e-4 weeks <br />6.3924e-5 months <br />) . Rau, et al. , ff. Tr.

24,372, at 54.

f 14. LIICO asserts that the EDGs are capable of surviving a ILOP/IDCA and that they are, therefore, suitable for nuclear service under GDC-17.

Indeed, LIICO asserts that the EDGs are capable of surviving 50 IcoP/IOCAs

\

l 11/

1 'Ihere is a serious inconsistency in LIIf0's position that the proposed

ILOP/LOCA standard should be used because LIIr0 also contends that the EDGs should be approved for the entire projected life of the Shoreham plant, not for use through one IOOP/I4CA. We do not find it necessary '

to address that inconsistency in view of our other findings.

t 4

I e

~ - .-. , _ _4

)

based. on a so-called "cunulative danage analysis" of the EDG 103 block j failure, which purports to be an analysis of crack growth rates under varying load conditions. Tr. 24,694, 25,313-14 (Rau) . Ebr the reasons discussed at pages 37-49 below, we reject LIIf0's cunulative danage analy-sis as inherently unreliable because it is based on inconplete data con-cernirg crack behavior.

15. Like PaAA, the NRC Staff acknowledgal that it could not state

)

that the EDGs are qualified to provide adequate service at the 3500 and 3900 kW power levels as contemplated by the FSAR. Tr. 26,238 (Berlirger) .

The Staff's testimcny is otherwise confusing and internally inconsistent.

t On the one hand, the Staff shares LIIf0's view that the EDGs meet the re-quirenents of GDC 17. Tr. 26,236-37 (Berlinger) . On the other hand, the Staff did not anbrace LIICO's position that the test is whether the EDGs can survive one ID3P/I4CA. See Tr. 26,234-36; 28,131-49; 28,284-85 (Berlinger). Instead, the Staff took the position that the adequacy of the EDGs must be evaluated on the basis of whether they can withstand repeated IDOP/LOCA events throughout the life of the plant. Tr.' 28,139; 28,141-42; 28,148; 28,284-85 (Berlirger) . Yet no Staff witness testified that the EDGs could survive an indefinite nunber of ILOP/I4CAs, or even more than one IDOP/LOCA, at any time during the course of the hearings.

16. 'Ihe Osunty contends that the appropriate test for canpliance with GDC 17 is unbodied in the performance rating requirenents contained in the 1

. , - - - - . - - - , . -- .. . , . , . , - - - - , , - - , , , - , , , , , . - . - -. , a . - .- --

[

FSAR. Hubbard ard Bridenbatgh, ff. Tr. 25,564, at 150. We Cbunty also contends that the evidence is insufficient to establish that the EDGs will l Curvive even a single ICOP/IOCA.

i

17. We agree with the Cbtnty and reject the IDOP/LOCA standard pro-posed by LIIf0 and the Staff. he proper trethod of evaluating EEG adequacy under GDC 17 is by reference to the requirements of the FSAR. We continu-ous and short-time duty ratirgs there set forth are consistent with the

)

cuatmary practice followed by utilities and the Staff in selecting and ap-proving diesel generators as standby mergmcy powr sources for nuclear power plants. Wose ratirgs provide the necessary conservatism and confi-

) '

dence to assure that the actual power needs are met in the event of emer-gency, consistent with the letter and spirit of GDC 17 and its historical c @lication. LIIf0 and the Staff in effect advocate that we apply a lower ctandard to these troubled Shoreham EIGs than is nonnally applied to mer-gency diesels with no histcry of failures. Such an approach is unjustified and muld not meet regulatory requirments.

18. In rejecting the ICOP/LOCA standard prorosed by LIIf0 and cndorned by the Staff, several additional observations are appropriate.

First, althotgh the Staff took the position that GDC 11 does not require mergency diesels to have contintous and short-time duty ratings, Staff witnesses did admit that ordinarily it requires such dtal ratings and also looks to the requirments of the FSAR in makirg detenninations under GDC W

17. Tr. 26,071-73; 27,743; 27,751; 27, %1 (Berlinger, Knox) . Second, the Staff was unable to say whether a If0P/wCA standard had ever been used be-fore in evaluating the adequacy of emergency diesels under GDC 17.

~

l Tr.

l- 26,236 (Berlinger) . Finally, the Staff never proffered an explanation of why it c1 roses to depart from the normally accepted method of eveluating emergency diesels (i.e. , reliance on the continuous and short-time ratings es reflected in ANSI /IEEE Standard 387-1977 and the FSAR) and select the I4OP/LOCA standard for this one particular plant. Under these circunstano-es, we cannot accept such a radical departure frcm the custcmary way of evaluating the adequacy of the diesels under GDC 17.

\

19. It is also pertinent to note that LIIro has proffered no explana-tion as to Wy it initially chose to adopt the FSAR standard of 3500/3900 kW for sizing the engines, maintained that standard for nearly ten years, but now believes it appropriate to depart so drastically frcm it. We can-not accept the notion that LIICO unnecessarily established an endurance rating more than 50 thnes greater (the difference between being capable of 1-reliable operation for a week, mostly at lower powr levels, as in a LOOP /LOCA, instead of a year at 3500/3900 kW, as in the FSAR) than required for ccmpliance with GDC 17.

20. Because the record contains no evidence whatsoever to support the conclusion that the EDGs meet the performance standards set by the FSAR and traditionally required to meet GDC 17, and because the cracking experienced

h 1

in.the blocks is strong circunstantial evidence that they cannot, we l

conc 1tde that LIICO has failed to carry its burden of proof on the issue

)

and the Cbtnty's contention is upheld.

21. 'Ihe foregoing disposes of all the issues concerning adequacy of the blocks at 3500/3900 kW. The balance of these findings are incitded be-cause LIICO has proEosed accepting the EDGs for use at a lcwer qualified load of 3300 kW and in the event that the Board decides to adopt LIIco's

)

proposed IDOP/IOCA standard. Based upon careful evaluation of all of the evidence, even if the approIriate standard were the ability of the cylinder blocks to survive one ICOP/LOCA or to operate only at 3300 kW, LIICO has failed to establish by a preponderance of the evidencel2,/ that the cylinder blocks meet that standard.

i 12/ Once a prima facie case is established (as clearly has been done in this instance), LIICO must establish by a preponderance of the evi-dence that the cylinder blocks are adequate. See Tennessee Valley Authority (Hartsville Nuclear Plant, thits IA, 2A, 1B, and 28),

AIAB-463, 7 NRC 341, 360 (1978); Iouisiana Ibwer & Light (b.

(Waterford Steam Electric Station, thit 3), AIAB-732,17 NRC 1076, 1093 (1983).

i III. DISCIESION

'A. EDGs 101 a'nd 102 are thfit for Nuclear Service I  : Because of the Liganent Cracks.

22. As described in Findings 3 and 6 supra, a substantial nunber of liganent cracks were first discovered in the cylinder blocks of all three

}

EDGs in February and March 1984, and the cracks in EDG 103 propagated sig-nificantly as a result of additional operating experience. 'Ihe cracks ap-

)

pear to be service-induced; i.e., they are caused by operation of the en-gines. Wells ff. Tr. 24,372, at 22-23; Andcrson, et al., ff. Tr. 25,564 at 180-81.13/

23. LIICO and the Staff assert that liganent cracks are benign be-cause they are unlikely to proIngate deeper than 1-1/2 inches and, even if they propagated deeper, they would at most cause minor cooling water leak-age that would not affect continued operation of the engine. 'Ihe (bunty esserts that there is sane risk that the liganent cracks could thenselves propagate to the point where they cause engine shut-down and that such risk is unacceptable for the critical safety function that the EDGs are required 13/ LIICO believes that the cracks result fran the interaction of stresses imposed on the cylinder blocks by a nunber of forces including: (i) pre-load forces derived fran clanping of the cylinder heads to the block tops by means of the cylinder head attri nuts; (ii) thermal loads derived fran tenperature differences in the cylinder liner, cylinder block, cylinder head and cylinder head studs; and (iii) firing Ires-sure loads derival fran gas pressure in the conbustion chamber.

Wells, ff. Tr. 24,372, at 23-26. 'Ihe interaction of those loads is very conplex. Id,. at 23.

t 4

1 I

to perfonn. We are not persuaded that the risk of liganent cracks propagating to the point of EDG failure during a ICOP/LOCA is so small that it is acceptable.

l

i. Operating Experience at Shoreham Discloses

'Ihat Liganent Cracks do not Arrest as LIIf0 Claims 'Ihey do.

24. LIICO asserts that strain gage testing, finite elenent analysis and field experience show that ligament cracks are not likely to extend below the cylinder liner landing, a distance of 1-1/2 inches below the block top, and therefore will not affect operation of the engines.

McCarthy, et al. , ff. Tr. 24,372, at 7, 58.14/

25. However, at least onel5/ of the liganent cracks in EDG 103 14/ '1he Staff agrees with LIICO to the extent that it does not believe

~

that ligament cracks alone render the EDGs unsuitable. Bush, ff. Tr.

23,126, at 28; Tr. 25,930 (Berlinger) . Ebever, the Staff's testimcny on the issue of crack arrest is confused and contradictory. Thus, Dr.

Berlinger is not concerned that liganent cracks would propagate deeper than 1-1/2 inches. Tr. 25,930. He asserted that the operating histo-ry on the three EDGs at Shoreham indicated that the cracks arrested at the cylinder liner landing (aIparently ignoring the deeper ' cracks in EDG 103 ) . Tr. 25,819. On the other hand, Dr. Bush testified that he did not have an adequate basis for concitriing that the ligament cracks in EDG 103 had arrestal and opined that they may have been growing at a decreasing rate. Bush, ff. Tr. 23,126, at 28; . Tr. 25,806-07. Dr.

Bush also asserted that the stress in the block decreases to the point of beccming ccmpressive but he did not know, and had no analysis of, l how deep that point would be. Tr. 26,091, 26,093-94 (Bush).

15/ bbreover, LIIf0's original measurenents of those cracks indicated that 4 other liganent cracks extended to 2-1/2 inches and one extended to 2 inches deep. See LIICO Ex. B-25, as originally fila 1. LIICO re/ised those measurenents in the belief that they were inaccurate due to the (Ebotnote cont'd next page)

Y prolmgated to a depth of 2-1/2 inches, another liganent-type crack propa-gated 4.4 inches down the front of the engine at Cylinder No.1, and a sttd-to-stid crack extended 3 inches deep into the cylinder block. See LIICO Ex. B-25, as revised. 'Ihe experience with EIE 103 therefore contra-dicts LIIf0's assertion because, according to LIICO's theory of crack be-havior, the crack growth dancnstrated in that engine could not happen.

26. LIICO's reliance on " field experience" is also misplacei. 'Ihat field experience concerns TDI R-4 engines at four nuclear power stations --

Catawba, River Bend, Grand Gulf and Ctmnanche Peak.l.6/ Wells, ff. Tr.

24,372, at 20-22. However, the engines at three of those four nuclear i

power stations had no cracks whatever and the exception, Commanche Peak, had " casting defects" (not service-induced liganent cracks) .l.2/ Tr.

(Ebotnote cont'd fran previous page) presence of Widmanstaetten graphite, which can give false indications of crack depth according to LILCO. Ebwever, for reasons discussed later in these findings, those crack depth changes are not necessarily accurate and the cracks may in fact be as deep as originally repre-sented-hy LIIf0.

-16/ 'Ihe same testimony refers to operating experience in marine applica-tions. However, the LIIf0 witnesses were unable to state at what load levels the marine diesels were operated. Tr. 24,689-90 (Wells).

! Thus, the marine experience has'no probative value because we have no way of canparing it to Shoreham. We note also that the marine experi-

~ence involves steady-state operation and the Staff believes that ener-gency use is more stressful. Bush, ff. Tr. 23,126, at 27; Tr. 25,804 (Bush).

1_7/ LIICO's pre-filed testimony also refers to liganent and sttd-to-sttd cracks discovered in the St. Clotd block and asserts that it is known to have poor material qualities. Rau and Wells, ff. Tr. 24,372, at (Ebotnote cont'd next page) 24,685-86 (Wells) . Far from providirx3 ccmfort about the Shoreham EDGs, the unexplained circtmstance that those other TDI diesels do not have cracks

)

like the ones at Shoreham accentuates our concerns about the EDGs. Asked how he accomted for the fact that those engines do not have cracks but the j ones at Shoreham do, Dr. Wells answered that it is difficult to assess and that FaAA believes that lignent cracks are "probably quite widespread throughout the industry." Tr. 24,708-09. Aside frcm the fact that Dr.

Wells' testimony to that effect contradicts his own earlier testimony that

)

the four ntx: lear plant engines do not have any cracks, we cannot accept such speculation. We therefore find that the difference in engine behavior

, stands unresolval and that the operating experience at other nuclear power stations is circtmstantial evidence that the Shoreham EDGs are not suitable for their purpse.

27. In short, the inferences to be drawn frcm operating experience are that the lignent cracks in EEGs 101 and 102 are unusual phencmena in nuclear applications and could propagate below 1-1/2 inches, contrary to t

the assertions of LIIro and the Staff. .

(Ebotnote cont'd frcm Irevious page)

47. However, we are not sure what the point of that testimony is. It cannot be asserted that cracks do not develop in blocks other than

! those with poor material qualities, because EEGs 101 and 102, which definitely have lignent cracks, are asserted by LIIro to be made of normal Class 40 gray cast iron.

)

l l

ii. LIICO's Finite Elment Analysis that Cracks will Arrest at a Depth of 1-1/2 Inches is not Reliable.

28. LIICO relies heavily on its finite elment analysis of stress in the blW< top for its argment that the ligament cracks are benign.

( According to LIICO, that finite elment analysis is based on strain gage tests of stress in the block top and supports the conclusion that ligament cracks will arrest at a depth of 1-1/2 inches. Howver, we cannot accept

)

that analysis as being a sufficiently reliable indicator of crack behavior because the strain gages were not positioned in locations which provide di-rect evidence of the stresses below the block top surface and because other 4

strain gage data contradict the analysis.

29. FaAA took strain gage readings of preload and operating strains in the original EDG 103 cylinder block at various load levels. Wells, ff.

Tr. 24,372, at 27. 'Ihe gages were positioned at three locations: ( a) be -

tween the adjoining sttriboles on the c.xhaust side of the engine at Cylin-ders 4 and 5, (b) at the midpoint of the centerline between cylirriers 4 and 5, and (c) between the sttr1Mle and front perimeter of the engine tomrd the intake side. Id. at 15-16; LIICO Exs. B-22 and B-23. 'Ihe strain data was used to cmpute the stresses in those locations and to predict, by fi-nite elment analysis, the stresses at other block top locations where no strain gages had been placed. Wells, ff. Tr. 24,372, at 27-28.

)

30. According to PaAA's analysis, the stresses in the block at the first thread of the sttri -- aIproximately 1.8 inches deep into the block -

are about 2-1/2 times lower than the stress at the block top. Tr.

25,499-500 (Rau) . We stress at that level is in a rarge of 3 to 5 ksi.

Id. mis is not entirely consistent with L11CO's argunent that ligament cracks will arrest at 1-1/2 inches, because it indicates tensile stress ex-ists as deep as 1.8 inches fran the surface.

31. We Staff alpaars to be unconfortable with the strain gage readings on W11ch the finite elenent analysis is based. We Staff would nonnally expect the highest stresses to arise during a fast start-up of the EDGs and believed that stresses are nonnally greater in energency diesels because of that quick-start feature, but PaAA's strain gage data for the fast start-up conditicn are not higher than steady state operation. Tr.

26,293-95; Tr. 25,804-05 (Berlinger, Bush). Dr. Bush was openly skeptical of the validity of the data and declined to accept it as evidence that quick starts do not prodtre higher stresses. As he put it:

"Such measuronents are highly sensitive to M1ere you put the strain gages, and if you happen to reach it

[ sic] in other locations it won' t tell me very mtch."

Tr. 25,833.

Later he repeated that strain gage data reliability is a function of the locations of the gages, and the particular placanents by FaAA here in ques-tion do not " . . . tell me very mtch about the position betwen the stul aryl the liner, arxl it doesn' t tell me anything about the counterbore region

)

and the thread region." Tr. 25,837-38 (Bush) . Dr. Bush then m phasized that those strain gages disclose the stress at the two places they were lo-cated, and not elsewhere. Tr . 25, 838-39.

32. Aside fran reservations about the strain gage data, the Staff considered PaAA's finite elenent analysis to be inconplete because it fails to account for thennal stresses arising fran two factors: the thennal gra-dient resultirg fran "cald start-up" and the pulsating thermal gradient re-cultirg fran firing of the cylinders. Tr. 25,843-48; 25,874; 25,879-80 (Dush). Dr. Bush considered the thermal gradient resulting fran starting a cold engine to be the more important of the two. Tr. 25,879. Although FaM appears to have undertaken sane analysis of that stress, Dr. Bush was plainly suspicious of it, and he would expect that stress to play a very pronounced role in crack initiation. Tr. 25,875-76; 26,225-26; S.C. Ex. 7.

3'4. 'Ihe Staff witnesses declined to endorse PaM's conclusion that ctress in the block top at the level of the first sttd thread is 2-1/2 times lower than it is at the surface because they lacked sufficient data to draw that conclusion. Tr. 25,881 (Dush, Berlinger).18/ 'Ihe Staff expressed the belief that stresses in the block top do diminish away fran the surface, but did not have an opinion as to where the self-limiting 18/ Even if thM's strain gage data were acceptai as wholly reliable, con-ventional strain gages disclose the surface strains anr1 not conditions beloa the surface. Tr. 26,574 (Dush).

point would be and as aware of no definitive evidence that the cracks had arrested. Tr. 25,853-54; 26,093-94; 25,806-07 (Berlinger, Bush) . 'Ihe Staff's analysis in this area is sumarized by Dr. Bush in response to a question frm the Ibard concerning the effects of operating at different loads for shorter and longer periods of time:

"I simply don' t have enough information to draw too many conclusions about the final propagation aspects, the depths, in other words, whether they [the cracks]

stabilize . . . . " Tr. 26,247-48.

34. In short, many questions were raised about the cmpleteness and reliability of FaAA's finite elenent analysis and they were not resolved to our satisfaction on the record.l.9/ We do not accept PaAA's finite elenent Cnalysis because of those unresolved questions and also because it ignores the only enpirical data available. 'Ihat data takes two forms. First, at

-least 3 (and possibly 8) cracks propagated deeper than 1-1/2 inches into EDG 103, as discussed in Finding 25, and that simply contradicts PaAA's 19/ In addition to the Staff's reservations about the cmpleteness of FaAA's finite elenent analysis, we have our own concerns that the analysis may not be reliable. 'Ihus, the EbAA witnesses testified that the stresses in the block and the block structure are very emplex, so cmplex that it was impractical to perform a standard fracture mechan-

.ics analysis of crack behavior on a cmputer. Tr.'24,801-03 (ItCarthy, Wells) . As we see it, a fracture mechanics analysis is akin to FaAA's finite elenent analysis because the finite elenent analysis purports to be a detailed analysis of stress distribution through the block top. See Tr. 24,793 (Wells) . We therefore have difficulty visualizing hFthe cmplexity of the block strtx:ture and the stresses in the block mako fracture mechanics analysis imiracti-cal, but do not affect the reliability of finite elenent analysis.

theory of crack arrest - a crack must have sme driving force to propagate. Tr. 26,092 (Bush).20/ Second, TDI undertook its own strain gage testing and the results showed that the stresses do not drop off quickly belcw the block top as reflected in LIICO's original prefiled tes-timcny of August 14, 1984. Wells, ff. Tr. 24,372, at 28. 'Ihat data in-citried measursmez, fem tw gages located in the cylinder counterbore area, belcw the block top. Tr. 24,852 (Wells) . It indicated that preload stress alone us near 10 Ksi tw inches below the bicek top. Tr. 25,500 (Rau) . It also iMicat d that firing stress was in the range of 3 Ksi.

Wells, ff. Tr. 24,372, at 28. By em);nrison, the largest stress measured by FaAA at any location on the block top was approximately 14 Ksi at 3500 kW and 16 Ksi at 3850 kW. See LIICO Ex. B-30.

35. LIICO has since elected to abandon that testimcny and to ignore TDI's block top strain gage data, allegedly because it was not able to con-firm the calibration of instrtmentation, the accuracy of recording, and the redtx: tion frcm strain to stress data. Tr. 24,851 (Wells) . Dr. Wells did not recall any specific error that suggested the measuranents should be discotnted. Tr. 24,851-52. He vaguely recalled scme problen with a shift of the mean value of strains. Tr. 24,852.2l_/

20/ We dean it highly pertinent in this connection that this inconsistency was highlighted by the Ibard long before the hearings closed. Tr.

25,830-31. Ebwever, LIIco never introducM any testimcny reconciling its position with that fact.

21/ Dr. Wells describd scrne other prob 1cus with entirely different aM distinct TDI strain gage data in the can gallery area. Apparently

~

(Ebotnote cont' d next page)

)-

36. We cannot disniss the TDI strain gage data so easily, at least in the absence of a mcre cogent explanation of its deficiencies. FaAA itself was disposed initially to rely on the data, as evidenced by its prefiled testimony and its June 1984 Block Peport. See SC Ex. 7, at 16-17. Indeed, at the time that Block Reprt was published, FaAA put more faith in the TDI strain gage results than in its finite elenent analysis:

"[I]n such a canplicated case, with interacting effects of clearance gaps, 3-D geometry and loading, friction, conponent-to-congnent distortion interactions, and relatively uncertain material properties the experimen-tal [ strain gage] results are judged to be more reli-able than the models." I_d.d at 16.

LIIfO has presented us with no cogent explanation why that statenent is not as true today as it was in June 1984.22/

iii. '1he Liganent Cracks 'Ihanselves (buld Cause Engine Pailure.

37. Both the Staff and LIIf0 agreed that liganent cracks could cause leakage of coolant fran the engine if they proggated 2-1/2 inches deep (Ebotnote cont'd fran previous page) that data had errors and inconsistencies. Tr. 24,375-76. TDI's can gallery strain gage testing was undertaken seprate and apart fran its block top strain gage testing (Tr. 24,855-56 (Rau, Seaman)) and prob-lens relating to the fonner have not been shown to be relevant to the latter.

22/ LILCO witMrew all testimony by TDI personnel originally filed by LIIro. Accordingly, there was no opportunity to explore the issue of the TDI strain gage data, or other pertinent matters, with TDI wit-nesses, none of whan testified at the hearing.

into the engine. Tr. 26,055-56 (Henriksen); Tr. 24,458-59 (Wells). As noted, at least one liganent crack in EDG 103 did extend 2-1/2 inches deep,

)

so the possibility of such leakage is very real. Ibwever, both LIICO arx1 the Staff discomt the significance of such a leak on engine operation.

Both assert that seeInge would be small and that loss of 20 gallons of coolant will trigger an alann. Tr. 25,232 (Yot]ngling); Tr. 26,055-56, 26,187-88 (Henriksen) . 'Ihe Staff conceded that whether the loss of coolant would be a problen depends on the rate of leakage and was apparently asstan-ing leakage frcm only a single crack. Tr. 26,300 (Henriksen) .

38. No party provided any definitive analysis of how much leakage could be expected fran a liganent crack. As the (btnty points out, the rate of leakage would depend on the nunber of cracks and their widths.

Christensen and Eley, ff. Tr. 25,564, at 153. 'Ihe Cbtnty asserts that coolant could leak rapidly because the coolant water is under pressure.

Id. The nonnal Iressure is 25 pounds per square inch. Tr. 25,490 (Johnson). Althotgh the coolant water is connected to a 1-1/2-inch water supply line that can add 70 gallons per minute to the cooling systen, that make-up capacity is not autanatic and requires mantal activation by an op-erator after being alerted by the low level alann. Tr. 25,272 (FbChrthy);

Tr. 26,300-01 (Henriksen) . During a IOOP/IDCA, the operator would be in the control roan or elsewhere. He would have to be dispatched to the en-gine roon to open the valves if the low level alann ms activated. Tr.

25,491-92 (Yotngling) .

i-

39. In short, the gist of the positions of LIICO and the Staff is that engine overheating caused by loss of coolant frcm liganent cracks is unlikely. However, no one testified that such overheating could not hap-1 pen, nor has anyone undertaken a definitive engineering analysis of the 1

risk. We cannot approve the EDGs on the basis of such speculative opin-

)

ions.

.B . 'Ihe Ligament Cracks Isad to Stirl-to-Sttrl Cracks Which bhy Cause EDG Failures

40. 'Ihe Cbtnty also contends that EDGs 101 and 102 are not acceptable for nuclear service because the ligament cracks in then raake it likely that sttd-to-sttrl cracks will initiate. Anderson, ff. Tr. 25,564, at 159-60, 163-64. 'Ihe Cbtnty asserts that the risk of such cracks initiatire is unacceptable because there is insufficient evidence on Milch to predict the behavior of a sttrl-to-sttrl crack and because such cracks could cause loos-ening of the sttris to the pint of engine failure. _Id. at 165-170. In ad-dition, the (btnty is concerned that sttrl-to-sttri cracks could result in leakage of cooling water, causing overheating of the engines.

41. LIIf0 agrees that sttrl-to-sttri cracks are possible in EDGs 101 and 102, but asserts that they will not necessarily initiate and, in any

- event, they muld not be disabling. McQtrthy, et al., ff. Tr. 24,372, at 59-61. LIIf0 argues that the physical properties of EDGs 101 and 102 are cuperior to that of EDG 103 because of Widmanstaetten graphite in the j ^

)

latter, that EDG 103 was subjected to greater operating stresses than the other tm blocks are likely to see (due to the load excursion which immedi-2 l Otely preceded that block failure), and that even the 3-inch stuS-to-sttd crack in EDG 103 did not impair operation of the engine. Id. Accordingly, LIICO asserts that a sttd-to-stid crack would not initiate and grow during e IDOP/LOCA .to the point that it would cause engine failure. Id. at 61.

LIICO argues that its cunulative danage analysis danonstrates that no crack p is likely to propagate sufficiently to cause failure during a single LOOP /LOCA. Rau, et al., ff. Tr. 24,372, at 52-54.

42. 'Ihe Staff believes that, so long as no sttd-to-sttd cracks exist i

at the time a IDOP/LOCA begins, the EDG is likely to survive operation through a IDOP/LOCA without a forced shutdwn. Tr. 26,088-90 (Bush). 'Ihe Staff therefore endorses use of the EIXis provided regular inspections re-i veal no sttd-to-sttd cracks in the block tops. Bush and Berlinger, ff. Tr.

23,126, at 27-29a. However, the Staff has undertaken no analysis to deter-mine the rate of sttd-to-stid crack propagation. Tr. 26,089 (Bush). 'Ihe i

Staff's opinion is based on its belief that the stresses which cause propa-gation decrease down into the block top. Tr. 26,090-93 (Bush). However, that belief is based solely on engineering " feel" or guesswrk, not on en-pirical analysis. Tr. 26,091-94 (Bush) .

i. Sttd-To-Sttd Cracks Are Likely 'Ib Initiate In EDGs 101 and 102 Because 'Ihey thve Liganent Cracks.

l l

E

u - ,. - _ _ _ _

)

43. 'Ihe only definitive analysis en the subject of sttd-to-sttd crack initiation ms sponsored by LIICO. It predicts that such cracks are likely to initiate in the presence of liganent cracks. LIIf0 undertook a finite elanent analysis to aralyze the stresses in the blocks during operation.

Wells ard Rau, ff. Tr. 24,372, at 42-45. Aside fran the conclusion that liganent cracks will arrest at a depth of 1-1/2 inches,23,/ the results of that analysis are shown in Goodman-Smith diagrams, IllfO Exs. B-49 and B-50, which represent the analyses for loa cycle and high frequency fa-tigue, respectively. Id,. at 45. According to LIIf0, both of those dia-grams disciose that the initiation of sttd-to-sttd cracks is "possible" at load levels of 90 percent (3150 kW) and above. Tr. 24,639 (Wells); Tr.

24,648, 24,704 (Rau) .24/ 'Ihe LIICO witnesses s ated that they could not identify the lomst levels at which crack initiation might occur because 23/ We did not accept this conclusion as valid based in part on the fact

~

that the cracks in EIX3103 propagated deeper than 1-1/2 inches and in part on the fact that the strain gage readings were so limited in scope. However, the analysis of sttd-to-sttd crack initiation does not suffer fran those infinnities because several sttd-to-sttd cracks did initiate in Em 103 and because the strain gage data incitded readings at a betwen-sttd location. See LIIf0 Ex. B-22. Indeed, as Dr. Bush pointed out, one strain gage ms located exactly where it should be for evaluating sttd-to-sttd cracks. Tr. 25,839.

-24/ After a later inquiry as to whether those diagrams predicted the pos-sibility of cracking at 80 percent of full load, Dr. McGrthy asserted that the analysis was so conservative that it did not permit interpo-lation "at lower operating loads." Tr. 24,705 (Rau) , 24,707-08 (dterthy) . We asstrne that Dr. McGrthy did not intend his renarks to apply to Dr. Rau's testimony at 90 percent power because no LIIf0 wit-ness ever expressly retracted that testimony. Indeed, Dr. Rau repeat-ed the testimony later in the hearings. Tr . 24, 925-2 6.

I

.l k  !

the Goodman-Smith diagrams were not appropriate for that purpose. Tr. .i 24,725-26 (Rau) .

44. . LIIf0 repeatedly stressed at the hearity that the restats of the finite elenent analysis reflected in the Goodman-Smith diagrams indicated only the. possibility, not the probability, of sttd-to-sttd cracks

~ initiating in EDGs 101 and 102. See, e .g . , Tr . 24,639-40 (Wells); Tr.

,. - 24,648, 24,922-23 (Rau) . However, it is clear that the diagrams do predict

)

that sttd-to-stid cracks will occur. 'Ihe Staff certainly interprets then that way. Tr. 26,062.(Bush). Indeed, the Staff interpreted then to mean that the stress impsed exceeds the ultimate tensile strength of the mate-

)

rial; that is, the first time the loads were impsed the engines would fail.- Id. As Dr. Bush put it: "I would interIret this diagram [LIIf0 Ex.

B-50] as indicating that in all instances the cracks would initiate." Tr.

26,063. Moreover, even LIIfo interpreted then that way before the hearings ccmnenced. Thus, LIIf0 Exhibits B-49 and B-50 are essentially the same as earlier Goodman-Smith diagrams prepared by FaAA and -inc1tded as Figures 3-13 and 3-14 in its' June 1984 Block Report. See S.C. Ex. 7 at 21-22. 'Ibe Bloc,k Report describes the significance of those two diagrams as follows:

"'Ihe implication is that initiation of ligament cracks in minimtm strength material is predicted, and given a ligament crack, initiation of sttd-to-stui cracks is also predicted." Id. at 15-16. (Dnphasis added.)

.LIIf0 was given the opportunity to reconcile that language with its testi-many that the diagrams show only the possibility of crack initiation and lamely suggested that it may be a matter of sanantics. Tr. 24,922-23

-(Rau) .25/

h

45. LILCO asserts that its finite elenent analysis is conservative.

See Tr. 24,369-40; 24,650-51 (Wells, Rau) . 'Ihe Staff agrees with that as-ressment because no sttd-to-sttd cracks have yet initiated in EDGs 101 and 102. Tr. 26,292 (Bush). However, no effort has been made to determine the degree of conservatism inherent in the analysis, and such a quantification may not be possible. LIIro acknowledged that there is no quantitative my to predict fran its finite elenent analysis when a crack might begin, nor how fast a crack might proptgate. Tr. 24,649-50, 24,687 (Rau) . Under the i

circunstances, and given the general unpredictability the cracks in these engines have already denonstrated,26/ we must assune that sttd-to-sttd cracks are likely to initiate in EDGs 101 and 10'.

-25/ It is pertinent that LIICO did not suggest that the difference in ter-minology is accounted for by the differences betwen the diagrams in the Block Report and the revised diagrams presented at the hearing.

We are left with but one conclusion; that LIICO now seeks to tone down the seriousness of the sittation revealed by its own finite elenent analysis.

26/ Althotgh block top cracks were known to exist in other TDI R-4 en-gines, apparently no one believed they existed in the Shoreham EDGs until early 1984, when each had been operated approxianately 1,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br />. In addition, FaAA's changes in crack depth measurements, the recent discovery of circunferential cracks where none had been de-tected by earlier testing, and nunerous more minor c1mnges that have evolved since LIICO first filed its testimony in August 1984 all add to cur uncertainty about the state of our current knowledge about these cracks.

l L

y li. 'Ihe Evidence does not Establish that the Physical Properties of the Blocks in EDGs 101 and 102 are Superior to EDG 103.

46. LILCO contends that the failure of the original EDG 103 block is not indicative of how the existing blocks in EDGs 101 and 102 rmy behave, because the EDG 103 block had degraded physical properties caused by the presence of Widmanstaetten graphite. 'Ihis is one of LIIf0's most important r2xgunents because of the extensive danage.experiencal in the failed EDG 103 block. After carefully weighing the evidence, we must conclude that LIICO has failed to carry its burden of proof on this issue.

i 47. 'Ihere is no disagreenent among the parties that the failed EDG 103 block contained Widmanstaetten graphite and that it therefore lacked the tensile strength of normal Class 40 gray cast iron. Rau and Mchab, ff. Tr. 24,372, at 30-31; Bush, ff. Tr. 23,126, at 25; Tr. 25,552-53

( Ander<on) .27,/ However, there is a disagreenent about the properties of EDGs 101 and 102. LILCO and the Staff contend that those blocks consist of normal Class 40 gray cast iron and that they are, therefore, superior to the old EDG 103. Rau and Mchob, ff. Tr. 24,372, at 41-42; Bush, ff. Tr.

23,126, at 26-27. 'Ihe Cbunty conterris that there is insufficient evidence i of the properties of EDGs 101 and 102 to make that ju3gment. Anderson, ff.

Tr. 25,564, at 171-72.

1 27/ 'Ihe reduction in strength resultirx3 fran Widmanstaetten graphite can-not be quantified based upn any formula, although LIIf0 purports to do so by reference to samples renoved fran EDG 103. Tr. 24,776-78 (Rau); McCarthy et al., ff. Tr. 24,372, at 35-36.

_ ~ _ . _ _

r

)~

t.

I l 48. FaAA performed tensile tests on pieces renoved frcm EDG 103 and l

found tha't their ultimate tensile strength ("Urs") rangal frcm 14.5 to 21.9

).

Ksi ccmpared to the 25 Ksi that would be expected for normal Class 40 gray cast iron. Rau and Mchob, ff. Tr. 24,372, at 35-36; Tr. 25,283 (Rau);

LIICO Ex. B-40. LIIro could not test the Urs of the metal in the blocks of EDGs 101 and 102 because to do so muld require destructive testing' of a sample and removing the sample frcm the blocks would ccmprcmise their structural integrity. Tr. 24,769-70 (Wachob). Homver, FaAA examined the

[

microstructure in various parts of EDGs 101 and 102. Based on that exani-nation, FaAA concitried that the microstructure of those blocks is different

frcm that of EDG 103 and concluded that the properties of those blocks are "ocmparable" to typical Class 40 gray cast iron. Rau and Mchob, ff. Tr.

24,372, at 35, 41-42; Tr. 24,643 (Rau) .28/

t

~~~28/At several points, FaAA noted that the B-bars cast with the blocks in EDGs 101 and 102 exceeded the minimtm tensile strength for Class 40 gray cast iron and asserted that suggests those block tops also exceed the minimtm UTS for that material. See, e.g. , Tr. 24,642 (Rau); Tr.

24,770 (M chob). However, those B-bars are believed to have been cast in separate molds frcm those blocks, and that would cause thera to have different tensile strengths because of differences in cooling rates.

Tr. 24,642-43 (Rau); In any event, it is clear that the B-bar results are not reliable indicia of the mechanical properties of the blocks because the B-bars cast with the failed EDG 103 block indicate that it was normal' Class 40 gray cast iron too. Tr. 24,733 (M ehob),

~

25,024-26 (Seaman), 25,783 (Dush) . 'Ihe presence of Widmanstaetten graIhite would invalidate any extraplation frcm B-bar to block properties. Tr. 25,066 (Pau). Accordingly, absent indeperxlent proof

~ that the blocks of EDGs 101 and 102 do not contain that degenerate

( structure, the B-bar results are irrelevant. It should be noted that all B-bar test results apparently ccme frcxn TDI because none of the bars for EDGs 101,102 and 103 are in existence any longer. Tr.

25,018 (Rau).

t t

i 1

49. FaAA's metallurgical analysis of the blocks in EDGs 101 and 102 is based on examination of samples fran four different sites in each block.

b Tr. 24,738-39 (Rau) . 'Ihose samples consisted of three small pieces of metal and approximately 10 replicas taken off the block top at Cylinder 1.

_Id; Tr. 24,M 6-47 (m chob). The Cbunty asserts that FaAA does not have an adequate basis for concitriire that those blocks are normal and do not contain Widmanstaetten graphite because the areas sampled are too small, representirg about 100 grams of material fran a block weighing 24,000 pounds. Andermn (Reb.) , ff. Tr. 26,326, at 1.29/ At first, Dr. Bush disagreed with the Cbunty - he believed that the samples were "statisti-cally relevant." Tr. 26,032. However, he acknowledged that it is possible that the microstructure is different below the surface (tut believed the probability of that is low) . Tr. 26,033. He also acknowledged that a larger sample would make him feel more confident of his opinion. Tr.

26,044. Later in his testimony, Dr. Bush testified that he continued to believe that the samples he did see fran EDGs 101 and 102 were very differ-ent fran EDG 103 but the samples were "less significant statistically" than he believed earlier. Tr. 26,287-88. Dr. Rau asserted that the seight of the samples is irrelevant, that the distribution is what is important, and

-29/ LIILO asserts that the location of the samples is more important than their weight and the Cbunty agrees. LIICO Proposed Finding B-10; Tr.

26,649-51 (Anderson) . Nevertheless, as discussed below, the samples were taken fran only 4 locations and the Cbunty never acknowledged that those locations represented an adequate nunber of locations.

l 1

that he would not expect the microstructure to vary markedly frm place to place because the cooling rates would be similar throughout blocks of this I'

size. Tr. 26,651-53,

50. We agree with the Catnty that FaAA's sampling was inadequate to I

satisfy us that ED3s 101 and 102 contain no Widmanstaetten gra@ite. Al-though Dr. Bush believed the samples were sufficient, he obviously would have been more cmfortable with more samples. In addition, he acknowledged I

scue possibility that they were not regesentative of the entire blocks.

On a matter as grave as this, we require more certainty than that. We are particularly unconfortable with the lack of thoroughness of PaAA's exanina-1 tion in view of the fact that Dr. Wachob could not affirm that FaAA found no evidence of Widmanstaetten graphite in EDG 102. Tr. 24,754. As he put it: "There are very local regions that have minor characteristics similar i

to it . . ..

_Id,. Dr. Rau attenpted to minimize the significance of those regions by stating that PaAA was not sure that they did have Widmanstaetten gra@ite and they were only isolated locations. Tr. 24,755.

Dr. Wachob later echoed Dr. Pau's couments and asserted that the .failcd EDG 103 block had Widmanstaetten graphite everywhere, and in extensive amounts'. Tr. 26,654-55. Ebwever, the fact renains that even the small voltane of sampling fran EDG 102 disclosed what might be Widmanstaetten graphite, and we concitrie that further sampling and analysis was required under the circe.astances.30/

30/ Dr. Bush had not seen photonicrographs of samples fran EDG 102 that-had the characteristics of Widmanstaetten graphite. Tr. 26,037. The (Ebotnote cont' d next page) t

51. Aside frcm the lack of persuasive evidence establishing that the blocks in ED3s 101 and 102 are superior to the old EDG 103 block, there is substantial evidence that they are not. 'Ihus, the ntmber and characteris-tics of the cracks in EDGs 101 and 102 (particularly 102) are not so markedly different than those in ED3103 prior to the qualification testing that led to that block failure. See Finding 3, supra. bbreover, EDGs 101 and 102 had less operating time than EDG 103 at the time LILCO Exhibits B-16 throtgh B-18 were prepared. See LIIfO Exs. B-13 through B-15. As discussM in Finding 66, supra, the crack behavior in EDG 103 during the qualification testing suggests that crack initiation and propagation accel-g crates as operating time builds up. 'Iherefore, the slightly more serious crack conditions in ED3103 conpared to EDG 101 and 102 (as reflected in LIIf0 Exhibits B-15 through B-18) may be a function of the differences in

) operating time on the engines. In any event, the cracking in each of the ergines as reflected in those exhibits is sufficiently similar to rebut LILCO's claims that EDGs 101 and 102 possess superior metallurgical properties.

(Ebotnote cont' d fran previous page) l existence of such photanicrographs w]uld indicate that the ones he did review were not necessarily representative of all the block material.

Tr. 26,040-41. 'Ihe existence of areas having the characteristics of Widmanstaetten graphite would cause Dr. Bush to be suspicious unless it were an isolated case. Tr . 26,04 2.

52. In stnumary, we do not have sufficient evidence to justify a find-I-

. ing that the blocks in EDGs 101 and 102 are sufficiently different fran EDG kl '

[ 103 to concitde that the failure of that block will not be repeated in then, and there is substantial evidence to the contrary. It is pertinent to note, however, as discussed below in connection with PaAA's cunulative danage analysis, that even if we were to conc 1tde that the blocks in EDGs

-101 and 102 are superior to EDG 103, we do not have sufficient evidence to

( 'concitde that they are safe because we do not know enotgh about their properties or crack behavior.

iii. 'Ihe Evidence is Insufficient to Establish

} that the Icad Excursion caused Additional Damage to EDG 103 or that the Damage to EDG 103 would not have been Disabling.

'53. We can dispose of two of LIICO's argtsnents quickly. 'Ihey are J'

that (a) the load excursion of April 14, 1984 imposed greater stresses on EDG 103 than either of the other tm EDGs are likely to see, and (b) the 3-inch deep sttd-to-sttd crack did not impair operation of the engine.

McCarthy, et al., ff. Tr. 24,372 at 60-61.

54. Dr. Wells asserted that the load excursion was the major contrib-utor to the crack grcuth in ED3103. Wells, ff. Tr. 24,372, at 20. Tr.

29,039-40. However, he acknowledged that he has no quantitative informa-

, tion to indicate whether the majority of the EDG 103 crack growth occurred before, during, or after that event. Tr. 24,676; 29,039-40. Dr. Bush also 4

h

- , - . - .~ -- --, . . , . - ~ - ----..- . , ~ , _ , . -- - . - - , _ . - - - , - ,

c asserted that the " severe overloads because of the load transient are believed to have caused rapid crack growth." l'ush, ff. Tr. 23,126, at 28.

{.

However, the engine was not " overloaded" during the incident -- the load stayed at 3500 kW. Tr. 24,655-56 (Yotngling), 24,658 (Seaman) . In any g event, under cross-exanination Dr. Bush acknowledged that he would need more details to form an opinion about whether 'and to what extent the load Gxcursion affected crack growth. Tr. 25,882-83. . In short, we simply have

) no evidence on which to evaluate the effects of the load excursion, and neither did the experts who professed to have an opinion on the issue.

t-

55. Similarly, we cannot gatrJe the impact on engine operation of the 3-inch deep sttd-to-sttd crack in EDG 103. LIIf0 asserts that EDG 103 was functioning normally at the time the engine was shut-down on April 14.

Yotngling, et al. , ff. Tr. 24,372, at 18. However, we have no assurance that that state of affairs was likely to continue, nor any evidence about how long it would continue. The fact that LIIf0 concitded that the EDG 103

. block was irreparable and scrapped .the block on that account, as discussed Garlier, is dispositive against LILCO's position here.

I.

t 36 -

i J

- -e, w -3 r--w. *- u.-%---7,*, , , , - ,,,,w- -

,-w.- .nw s,.4-y,,,---y ,-,,,m,,-,,,,ew,%,,--.,,-,,,y.,-..y,mw,-wwyy,w,,-.-,, , y,- y,%,,--e.,- -

)

i iv. FaAA's Qmulat Lve Damage Analysis is not a Reliable Inc.ex of Crack Behavior.

l

56. FaAA relies heavily on its so-called cunulative damage analysis in asserting that EDGs 101 and 102 are likely to be able to perform reliably throtgh a single ICOP/LOCA. 'Ihe cunulative damage analysis .ns used by FaAA to " bound" (i.e. , calculate) the amount of sttd-to-sttd fa-tigue crack extension etpected during a LOOP /LOCA based on the cracking in EDG 103 during the qua]ification testing betw en March 11 and April 14, 1984. Rau and Wells, ff. Tr. 24,372, at 49. 'Ihe starting pint of the analysis is a calculation of the relative stresses imposed on EDG 103 by the various loads on the engine based on PaAA's strain gage readings. Tr.

24,694 (Pau) . 'Ihe analysis relates that operating profile to the damage experienced in EDG 103 during that testing. Id. It then conpares the power levels (and time at those levels) pastulated for a IOOP/LOCAh/ to those experienced by 1:DG 103, calculates the relative amounts of stress ex-perienced in a IOOP/LCCA vis-a-vis that EDG 103 qualification testire, and calculates the amcunt of crack progression that is likely in a IOOP/LOCA based en those relat:ionships and on EIG 103's behavior. _Id . As suggested by the prefiled testimony, the calculatien of relative damage accounts for the difference in material properties of cast iron with Widmanstaetten graphite and normal Class 40 gray cast iron. Tr. 24,693-94 (Rau) .

31/ 'Ihose pamr Ic.els wre 3,881 kW for 0.2 hour2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br />, 3,409 kW for 0.8 Mur, and 2, C17 kW far 167 Murs. Pau, et al . , ff. Tr. 24,372, at p. 54.

57. According to PaAA, the "cunulative damage" to be expected in a IDOP/LOCA is about two-thirds of that to which EDG 103 was exposcd during l the qualification testing betwen bhrch 11 and April 14, 1984. Rau, et al., ff. Tr. 24,372, at 52-53. FaAA asserts that the maximtm erack exten-sion during testing of EDG 103 was 1-1/2 inches, the growth reflected in the sttri-to-sttd crack, located on the exhaust side of the engine betwen

- Cylinders 4 ard 5, which allegedly propagated fran 1.6 to 3 inches. _Id. at

53; Tr. 24,781, 24,783 (Rau) . FaAA emphasizes that "maximtm" crack exten-sion took place in "dogenerate" cast iron exposed to more than 150 percent of the stress that would be experienced during a LOOP /LOCA and it caused no j operational problens. Id. Indeed, because of the allegedly superior properties of the cast iron in EDGs 101 and 102, FaAA's analysis predicts that they would survive a IDOP/LOCA with less than 2 percent of the damage experienced by EDG 103 during that qualification testing. _Id.32/

32/ 'Ihe record is not precisely clear how FaAA gets fran the conclusion

""~

( that the stress in a IOOP/LOCA would be two-thirds of what EDG 103 ex-l periencerl to the conclusion that the damage in a IOOP/LOCA would only be 2 mrcent of the EDG 103 damage. Prest nably, it has scmething to do with the allegedly superior qualities of the cast iron in EDGs 101 and 102 conpared io the Widmanstaetten graphite-corrupted EDG 103 Jblock. FaAA calculated that the fatigue crack growth rate in EDG 103 l~ was 10 to 100 times faster than it would be in EIXis 101 and 102. Rau l and mchab, ff. Tr. 24,372, at 40-41. We flatly reject that notion based upon conparison of the cracks in the EDGs discussed below. 'Ihat canparison denonstrates that the extent of cracking in EDG 103 prior to the qualification testing was not so much greater than that in EIXis 101 and 102 (particularly EDG 102), which had less operating time. In view of our lack of knowledge about fracture mechanics in theso

blocks, we cannot have any confidence that EDG 103 had not reached a I critical reint in its load history before that qualification testing, causing crack initiation and propagation to accelerate. Likewise w (Ebotnote cont'd next page)

I i

58. LIIro asserted that its cunulative danage analysis is conserva-tive for a variety of reasons.33/ Ibwaver, in view of our finding that the analysis itself is unreliable, we do not address the points individually here.

59. 'Ihe Staff did not have much caument on PaAA's cunulative danage theory. Er. Bush testified that he " generally" agreed with PaAA's method-ology. Tr . 26, 228. He later elaborated on that stater.ent, how ver, I

cxplaining that he did not mean that he agreed with the analysis itself --

that would require review of the step-by-step details and he made no such review. Tr. 26,313-14. Aside fran the lukewarm nature of that endorse-ment, it app 3ars that Dr. Bush really has no confidence in the analysis; he testified that the existence of any stui-to-stul crack in EDGs 101 and 102 would be cause for him to stop engine operation, because it would indicate (Ebotnote cont' d fran previous page) cannot be sure that EDGs 101 and 102 have not reached the same criti-cal point.

33_/ 'Ihey are that: (a) it assumes the presence of a stui-to-stui crack 1.6 inches deep when no such cracks exist in EDGs 101 and 102; (b) it will be more difficult for cracks to initiate in EDGs 101 and 102 because they allegedly do not contain Widmanstaetten graphite; (c) even if a crack grew to 3 inches deep, it would not affect op3 ration of the en-gine based on ED3103 's performance; and (d) the analysis ignores any accelerated crack growth that may have been causal by the " overload excursion." Tr. 25,316-19 (Ran) . For reasons that should energe fran discussion of those factors in other Findings, we do not agree that all of then are valid bases fo; describing the analysis as conserva-tive.

)

that we do not know enough about such cracks. Bush, ff. Tr. 23,126, at 30.

As he put it:

"A deliberate decision to continue operation without .

repair of such a crack is not justified because the presence of such a crack indicates that current analytic techniques do not accurately model crack initiation and growth." Id. at 30.

'Ihat testimony is simply inconsistent with the conclusions of PaAA's cunu-lative damage analysis that even a 1.5 inch deep stui-to-stul crack would not be a problon in a LOOP /LOCA.

60. 'Ihe (bunty attacks the validity of the cunulative damage analysis on a nunber of grounds.34/ First, it asserts that the analysis purp rts to analyze the worst crack extension in EDG 103 during the qualification testing, but it ignores the 4-1/2-inch crack running down the block front at Cylinder 1. Anderson and Bridenbaugh, ff. Tr. 25,564, at 167. Second,

-34/ LIIf0 asserts that the Cbunty's criticisns are entitled to no weight because the (bunty's witnesses performed no independent cunulative damage analysis and have no experience in performing such analyses.

LIIfO Proposed Fialing B-37. However, both of the Cbunty's witnesses testified that it is not necessary to perform such analyses in order to review and conment upon then, and that they have the requisite edu-cation and experience to review and connent as they did. Tr. 26,428 (Anderson, Bridenbaugh). LIIf0 introduced no evidence to rebut those statenents. In any event, it is elonentary that an expert may be qualified by education and experience to offer testimony on a subject even though he may have never before dealt with the particular issue at hand. Moreover, the (bunty's connents establish the credentials of the County's witnesses far more persuasively than honors and distinc-tions m uld.

i the (bmty asserts that crack dynamics are affected by the sequence of loads as well as load duration, and the analysis fails to account for that fact.3,5/ Id. at 167-68. bbst importantly, the 03unty asserts that PaAA i

l lacked significant information about the behavior of EDG 103 that is vital to valid predictions of behavior of the other blocks. Id. at 169-70. 'Ihat information is:

" men did the liganent cracks first initiate in each of the three EDG blocks, and what was the cunulative dam-age index of each at that point? W en did the original crack between the stui holes in the EDG 103 block first initiate, and what was the additional damage index ac-cunulated between the initiation of the ligament cracks in the same block and that' point? Wen and mder what conditions did the original crack between the stui holes in the EDG 103 block grow to 5-1/2 inches in depth, and what was its rate of growth? W en did the large crack running fran cylinder no.1 down the front of EDG 103 first initiate and at what rate did it prop-agate? " Id. at 169-70.

61. FaAA failed to provide any satisfactory answer to the concerns raised by the (bmty.36/ It acknowledged that the analysis did not use the

, -35/ Dr. Rau acknowledged that the analysis does not account for the rela-tive sequence of loads at different power levels. He assertal that load sequence is not significant to such analysis except where the loads vary enormously. Tr. 24,817-18.

-36/ At various times througlout these hearings, FaAA denonstrated a lack of objectivity which, though not surprising in expert witnesses, l clashed with the image it sought to project. On one occasion, an FAAA witness went so far as to suggest that the engines can be adequately lubricated by water in place of oil. Tr. 25,251-53; 25,515 (becirthy) . FaAA's testimony about the 4-1/2-inch crack and its im-pact on the cunulative danage analysis was another such instance. For exanple, Dr. Bau asserted that it took no less than the full amount of (Ebotnote cont' d next page) crack at the front of the engine at Oylirxler 1, and that it was the deepest. Tr. 24,812-13 (Rau) .37_/ It attenpted to justify that cmission on the basis that its analysis suggested that the position between Cylinders 4 and 5 had the least amount of " margin" (i.e., the most ptential for crack propagation). Id. 'Ihat conclusion appars to have been based on FaAA's strain gage data, which reflect! lower stresses at the Cylinder 1 location than between Cylinders 4 and 5. Tr. 24,806 (Rau) . However, the fact of (Ebotnote cont' d frcm previous page) operating tone between burch 11 and April 14, 1984 for the crack in EDG 103 at Oylinder 1 to propagate 4-1/2 inches deep. Tr. 24,804. He flatly rejected the possibility that, based on the circtustances sur-rounding its discovery, it may have propagated rapidly, in less than an hour and 20 minutes. Tr. 24,804-05. Asked how he es so sure it took no less than the full operating time, he lamely pointed out that the pre-March 11 inspection results showd an indication at that loca-tion, and relied on strain gage tests to the effect that the stresses are lower there than between Cylinders 4 and 5. Tr. 24,806. We fail to understand why Dr. Rau takes ccmfort in the latter point in view of the fact that the crack at Cylinder 1 is the deepest known crack in the block. As noted above, that circunstance only suggests that FaAA's strain gage analysis is faulty. As to the presence of the in-L dication, another PaAA witness confirmed (cnly begrtdgingly) that he made no effort to test its depth because it was so short (0.4 inch) that it was not likely to be deep into the stui hole. Tr. 24,809-10 (Johnson).

3_7/

7 LIICO reads the same testimony as proof that FaAA did " consider" this crack in its ctmulative damage analysis and denonstrated in the analy-sis that propagation of the crack in a ICOP/LOCA kould be less than that of a sttd-to-sttri crack. LIICO Propsed Finding B-38. LIICO's argtraent begs the question because the Cbunty's point is that the analysis was not based on the deepest crack. bbreover, LIICO's asser-tion that this crack would not grow more deeply than a sttd-to-sttri crack is mystifying since the crack did in fact grow deeper than any sttrl-to-sttrl crack during the qualification testing.

t i

the 4-1/2-inch crack being the deepest is strong circunstantial evidence that the analysis is not consistent with reality. In any event, the cunu-lative danage analysis fails to account for the deepest crack and that in itself is a serious weakness in it.

62. Clearly, the underlying prenise of Pa7A's cunulative danage anal-ysis is that the crack behavior observed in EDG 103 is a reliable indicator of how the cracks in EDG 101 and 102 will behave; otherwise, the analysis is meaningless. LIIf0 has the burden of proving that prenise is valid, and it has failed. Far fran establishing th..t the behavior of the cracks is predictable, the record establishes that they are decidedly unpredictable and that we do not possess enough information to draw conclusions about how fast and under what conditions the cracks in EDGs 101 and 102 will grow.

We now focus on our greatest concern -- namely, that no one knows when the cracks started and how fast they grew.38/ It follms, therefore, that no one can reliably predict how they will behave in the future.

-38/ FaPA did not perform a normal fracture mechanics analysis, which is a detailed stress distribution through the block top and calculates crack growth rates. Tr. 24,793 (Wells) . However, Dr. Pau contended that its cunulative danage analysis is a form of fracture mechanics analysis. Tr. 24,793-94. FaAA did perfonn fracture mechanics calcu-lations early in its work. Tr. 24,794 (Wells) . 'Ibe objective was to predict the renaining life on EDGs 101 and 102 based on the existing cracks in those blocks. Tr. 24,798 (Wells) . 'Ihe analysis predicted that liganent cracks would propagate rapidly, with no arrest, so PaAA decided to abandon it because it was not Consistent with observed crack behavior. Tr. 24,800-01 (Walls) . Apparently, the analysis is so conplicated that it would not fit on any existing canputer avail-able anywhere in the world. Tr. 24,801 (StCirthy) .

63. Dr. Wells acknowledged that PaAA does not know precisely when the stud-to-sttd crack between' Cylinders 4 and 5 cecurred. Tr. 24,710-11. - As he put it: "And even now having broken it open, it is not possible to trace its precise origin through time." Tr. 24,711. Similarly, Dr. Pau admitted that he could not answer specifically how long it took for the

)

4-1/2-inch crack at Oylinder 1 to propagate to that extent. Tr. 24,804.

Yet the rate of that crack's growth clearly would affect the cunulative danage analysis. Dr. Rau was asked whether, assuring the 4-1/2-inch crack

)

propagated to that length in 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> and 20 minutes (the period between en-gine start-up and its discovery on April 14, 1984), he would concitde that l the EDGs could survive a IDOP/LOCA. Tr. 25,452. After much vascilation,

)

he stated that it was "too close to call" and argued that the assunption was unrealistic. Tr. 25,452-54. Undoubtedly in Dr. Rau's mind it is, but y there is no credible evidence in the record to support that opinion.

64. FaAA attenpted to resist the notion that the crack behavior in EEG 103 was unpredictable. Thus, for exanple, the PaAA witnesses were eva-

)

cive in~ responding to the following question fran the 03unty at Tr. 24,819:

"If you had done your cunulative danage analysis using the state of the block top in this case as sMwn on Ex-

) hibit B-18 for the period of operation of EDG-lO3 fran the time _the engine wnt into service until March 8th, 1984, that analysis would not have predicted the crack initiation and propagation that occurred fran March lith through the time the block failed, would it?"

Y -M-

)

Dr. Wells answered that PaAA made no attenpt to do that and Dr. Pau responded that he could not answer it (adding new facts to the question and then suggesting that the new facts made the question nonsensical) . Tr.

24,820. In a follow-up question, the FaAA witnesses were asked whether it was true that the danage accunulated on EDG 103 during the qualification testing was greater than the danage that had accunulated during its entire oprating history before that time. Tr. 24,821. Dr. Rau equivocated and ultimately respnded that: "The inspection records speak for thanselves."

Tr. 24,821-22. We agree, and they disclose the basic fallacy of FaAA's cunulative danage analysis -- they reveal that EDG 103 suffered far more i danage relative to operating time during the period March 11 through April 14, 1984 than it had during its prior operating history.

65. Analysis of L11CO Exhibit B-15 discloses that, prior to March 11, 1

1984, EDG 103 had operated a total of 1,270.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />, of which 350.5 were at full load and 77.5 were at loads in excess of 3500 kW. By canparison, the qualification testing fran March 11 consumoi a total of 108.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />, of which 24.5 were at full load and 14 exceedei 3500 kW. Thus, if total operating time were the key, the qualification testing represented less than 10 percent of the prior load history; if loads at 3500 kW were the key, that testing represented less than 10 percent of the prior load histo-ry; and even if we ignore those statistics and focus solely on loads above j

3500 kW, that testing represents only 18 percent of the prior history.39/

3_9/ If we adopt that view, we must assune that the warst load profile is 3900 kW. As reflected in LIICO Exhibit B-15, qualification testing at I

(Ebotnote cont' d next page)

I

-66. Based on any of those measures, it appears that the rate of dete-rioration of the original EDG 103 block was accelerating at an alanning rate and reached critical proportions during the qualification testing.40/

Tnus, the amomt of additional danage sustained as totally disproportion-i ~

ate to the load profile of that testing. We sheer nunter of cracks in-creased abnost 50 percent -- fran 27 (comting 8 indications) to 40 (comting 7 indications) . he nunber of sttrl-to-sttri cracks increased fran 6 to 12 (again comtirg indications), an increase of 100 percent.41./ Fi-nally, the depths of the cracks also increased significantly, with sone -

increasing alanningly. hus, the 5 sttrl-to-sttri indications having no mea-curable depth before the test propagated to depths ranging fran 0.85 inch to 1.5 inches, with most being to 1.5 inches.42/ We other sttrl-to-sttri crack propagated fran 1.6 inches to 3 inches deep. Finally, the total (Ebotnote cont'd fran Irevious page) 3900 kW constrned only 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> or slightly more than 3 percent of the 29

hours at 3900 kW that precedal it.

40/ Dr. Bush confirmal that the rate of growth can accelerate as a crack becones larger. Tr. 29,058.

i-g/ Ignoring the indications presents an even wrse picture. Were ws one sttrl-to-sttri crack (and 5 irxlications) before the testirg. After l it, there were 6 such cracks (and 6 nere indications) .

42_/ It is pertinent that all of the crack growth data is based on LIIf0's l measurenents. If there are errors in those measurenents they tend to

! overestimate the crack depth before the qualification testing and to I underestimate the crack depth following the test. Werefore, the l crack growth picture that energes fran conparing LIIfO Exhibits B-18 and B-25 in highly conservative.

^

l_

depth of all 29 cracks and indications before the test was 29.1 inches (comting indications as mere surface defects with no measurable depth) for l

l an average depth of just over 1 inch. By contrast, the total depth of those same cracks and indications, ignoring new ones, is 44.15 inches for f an average of just over 1-1/2 inches. In short, under any view of the sit-uation, the damage sustained by EDG 103 during the qualification testing

! which forms the basis for PaM's emulative dmage analysis as at least 50 parcent of the damage sustained by the engine througlout its prior operating history, even though the total time on the engine during that testing was less than 10 percent of its prior operating history (and less j than 20 percent of its operating history even if we totally ignore all time on the engine at 3500 kW and below) .

i i

67. Were is yet another reason why FaM's ctmulative damage analysis

is not reliable. We data on which it depends is unreliable because the crack depth measurments in EDG 103 are uncertain. As noted earlier, LIT O decided that its original crack map of EEG 103 (depicting the condition of the block after the block failure 3,/,on 4 April 14, 1984) was incorrect when it sectioned the stui-to-stui crack between cylinders p and 5 of EDG 103, i

l 43/ 'IWo PaM witnesses attsptai to argue that the events of April 14,

~

1984 did not constitute a " failure." Tr. 24,434 (Johnson, W:Carthy) .

I However, the effort merely dmonstrates how totally biased and unobjective PaM's conduct during the hearings was. The words " block I failure" are FaM's own, extracted frcm LIIf0's own exhibit and there used to describe the events of April 14. See LIIf0 Ex. B-15; Tr.

24,435-36 (R:Carthy) .

then believed to be 5-1/2 inches deep, and fomd it was 3 inches deep instead. Rau and W1chob (Sup.) , ff. Tr. 24,372, at 10. Only that crack was acttally sectioned by FaM. Tr. 24,440 (Mclub). Ebwever, that dis-covery prmpted EhM to recalibrate its eddy current probes to account for the very noisy signals produced by the Widmanstaetten graphite in EDG 103.

Tr. 24,442 (Rau) .

68. 'Ihe change in the eddy current test involved raising the noise level above the level which would ordinarily be used to define the end of a crack. Tr. 24,475-76 (Rau) . FaM did not measure the extent of Widmanstaetten graphite in the areas where it revised its earlier crack depth measuranents. Tr. 24,475 (Johnson) . Cmmcn sense suggests that such an approach could ignore a signal provoked by additional crack length in favor of an asstmption that the signal is caused by the presence of graph-ite.44_/ FaM itself implicitly acknwledged that the revised measurenents could err in favor of representing a crack to be too shallow by saying that they were " conservative" in the sense that the cracks are at least as deep as represented by those revised measuronents. Tr. 24,479 (Johnson) . In 44/ It is pertinent that FhM did not determine the extent of Widmanstaetten graphite that may have been in each area where an eddy current readirs was taken and acknowledgal that there is more in sme places than in others. Tr. 24,475-76 (Johnson, Pau) . Indeed, FaM acknowle3ged that Widmanstaetten graphite does not alwnys cause unreliable cady current readings. Tr. 24,592 (Johnmn) . 'Ihe eddy current standard now depends on the ju3 9 ment of the inspector. Tr.

24,594-95 (Johnmn) .

l

)

l summary, we cannot be sure that the shallower crack depths reported in LIIf0 Ex. B-25, as revised, are correct and that the cunulative danage analysis uses the correct crack depths for conparison.

69. 'Ihe data used in PaAA's cunulative danage analysis may understate the crack growth in EDG 103 during the qualification testing for a related reason. 'Ihat is, those estimates of crack growth involve conparire the post-test crack depths with those recorded prior to the test, before PaAA was aware that Widmanstaetten graItite could be causing it to overestimate the depth of the cracks. Dr. Johnmn himself describal the eddy current tests conducted on EDG 103 prior to Septenber 1984 as unreliable because they may overestimate the depth of the cracks. Tr. 25,335.45/ If the cracks in EDG 103 were actually shallower than represented in LIIf0 Ex.

B-15, then it is clear that LIIf0 is underestimating the amount of crack growth in that block between March 11 and April 14, 1984.5/

-45/ Dr. Johnmn rechecked the eddy current test results on the 1.6-inch deep sttrl-to-sttri crack (as measured before the qualification testing) and asserted that the signals were unmistakably cracks dan to at least 1.4 inches. Tr. 28,823. Ebwever, that does not eliminate the questions raised about the reported depths of other cracks in that block as listed in LIIf0 Ex. B-18.

46/ Dr. Rau testified that this would have only a small effect on the cu-mulative danage analysis. Tr. 25,315. 'Ihat may be but it does noth-ing to assure us that the analysis itself has any validity.

v. ' A Sttd-To-Stud Crack could be Disabling

70. LIlCO asserts that a sttd-to-sttd crack would not be disabling to the EDG even if it proggated more deeply than LIICO predicts. 'Ihus, the worst consequence LIICO could envision fran a sttd-to-sttd crack would be loosening of a cylirder head .sttd, and even that would not be a problen be-cause there are seven other sttds to hold the cylinder head down. Tr.

25,234-37 (Wells) . According to LILCO, a sttd-to-stid crack (mlike a lig-ament crack) cannot realistically get to the eter coolant area. Tr.

25,236 (Wils) , 25,238 (ftCarthy) . It is pertinent, however, that LIILO admitted that it had not undertaken an analysis of the effects of extensive ettd-to-sttd cracking. Tr. 25,234 (Wells) .

71. To support its position, LIICO enphasizes that a 3-inch deep cttd-to-sttd crack did propagate in EDG 103 and caused no problens with en-gine operation. Tr. 25,233 (Yomgling) . Howver, no one kncws when that crack becane that deep or what would happen to it if the engine had not been shut down. Indeed, that crack was considered by LIICO to be so seri-ous that it could not be repaired reliably, pranpting LILCO to scrap that c.ntire block. Tr. 24,665-66 (Youngling, Rau) . Based on this record, we cannot concitde that the 3 inch deep sttd-to-sttd crack was benign, or that it would not have propgated rapidly and perhaps contributed to the initiation of additional cracks if the EDG had not been shut down.

i.

s

_1

72. 'Ihe Staff minimized the consequences of cooling mter leakage fran a sttd-to-sttd crack with the same reasoning it aIplied to ligament

[

. cracks. Tr. 26,189 (Henriksen) . Dr. Bush testified that a sttd-to-sttd crack would not cause the engine to fall apart, but he ms amre of no real analysis of the subject. Tr. 26,189 (Bush). Dr. Berlinger could vismlim loosening of the sttd tolts as the result of a throtgh crack, but consid-cred it to be unlikely. Tr. 26,190 (Berlinger).

73. The Cbtnty contends that a deep sttd-to-sttd crack could cause loosening of the cylinder head sttds, causing loss of pomr and overloading of the renaining cylinders. Bridenbatgh, et al., ff. Tr. 25,564, at 165.

It asserts that could cause engine failure. Id. 7.he (btnty is also concerned that a sttd-to-sttd crack could cause coolant leakage as acknowl-cdged by Mr. Henriksen.

74. We cannot disniss the consequences of a sttd-to-stid crack as LIICO and the Staff have done. It is obvious fran their testimony that they have performed no in-depth analysis on the point. bbreover, the sttd-to-sttd crack in EDG 103 pronpted LIICO to scrap that block, and that- be-havior is simply inconsistent with the idea that such cracks are not poten-tially disabling. Finally, the Staff's only metallurgical expert testified that:

"'lhe appearance of a sttd-to-stid crack in nonnal qual-ity cast iron muld indicate that too little is known concerning the stresses and stress distributions lead-ing to such a crack. A deliberate decision to continue l

l

)

operation without repair of such a crack is not justified because the presence of such a crack indi-cates that the current analytic techniques do not accu-I rately model crack initiation and growth." Eush, ff.

Tr. 23,126, at 30.

'Ihat Staff testimony is simply incmpatible with the notion that sttd-to-sttd cracking is not potentially disabling.

C. 'Ihe Ibssibility of Ciremferential Cracks also Renders EDGs 101 and 102 Unfit for Nuclear Service.

75. Smetime after August 14, 1984, when the sttd-to-stu3 crack in EDG 103 was sectioned, LIIf0 discovered circmferential cracks in the cyl-inder counterbores of that engine. Rau and Michob (Sup.) , ff. Tr. 24,372, at 11. According to LIICO, the cracks were "very shalloa," extending a maximtm of 3/8 inch into the block top. Id. A magnetic particle inspec-tion reprt dated Septmber 19, 1984 discloses that the cracks extended en-tirely around the circtmference of all 8 cylinders of EDG 103. Hubbard and Anderson (Sup.) , ff. Tr. 25,565, at 11; S.C. Ex. S-10. Circtmferential cracks had been known to occur in TDI R-4 engines but, prior to that time, LIIfO believed that the Shoreham EDGs were free of stch cracks. S.C. Ex.

7, at 4-5.

76. All three of the EIX3s had been inspected for circtmferential cracks using liquid penetrant in February and lurch 1984. Tr. 24,866-67 (Schuster). Indeed, the inspection on EIX3103 was regated in April 1994.

l

h

_I_d.

d at'24,867. 'Ihere wre no reparted irdications of circunferential y cracks prior to the sectioning of EDG 103. Tr. 24,444 (Johnson) . It is therefore obvious that those prior inspections did not reveal any circun-ferential cracks. LIICO explained that it is difficult to inspect for such

, cracks because they fonn in the corner created by the cylinder counterbore and cylinder liner landing, and it is hard to clean'that area. Rau, et al.

(Sup.) , ff. Tr. 24,372, at 12.

77. In its supplanental testimony filed after the circunferential cracks were discovered, LIICO stated that, because of the difficulties in testing for circunferential cracks, FaAA assuned that such cracks exist in EDGs 101 and 102. Id. Ibwever, in their oral testimony, Dr. Rau and Mr.

Schuster stated that EDGs 101 and 102 have no circunferential cracks, based on ultrasonic testing perfonnal in February and Itrch 1984 which allegedly revealei no such cracks. Tr. 26,692-93 (Rau); '28,870-72 (Schuster) . Dr.

Rau plainly was testifying on the basis of hearsay and did not participate in the alleged test. See Tr. 26,868-70. Moreover, Mr. Schuster had no satisfactory explanation as to why LIICO did not credit that testing in preparing its original testimcny. Tr. 28,872-73. We do not believe LIICO's non-destructive exaninations of EDGs 101 and 102 should be consid-ered to be reliable. Non-destructive exanination failed to discover the circunferential cracks in EDG 103. 'Ihe LIIfO testimony contradicting its own prefiled testimony is plainly unreliable and can be given no weight.

The Staff expressed the opinion that it would not be surprised to see

)

circunferential cracks in EDGs 101 and 102 because the geanetry of the area where they have been observed is condtx:ive to crack formation. Tr. 26,023 (Bush). We concitde that EIX3s 101 and 102 should be assuned to have cir-conferential cracks.

78. It is clear that the circunferential cracks are operationally indtred. LIIf0 testified that the primary mechanisms which cause circun-ferential cracks are stresses imposed by operation of the engine, primarily firing pressures and thermal expansion of the cylinder liner against the counterbore. Tr. 25,090-91 (Wells) .

79. LIIEO asserts that the presence of circunferential cracks in EDGs 101 and 102 poses no threat to their proper functioning. McCarthy, et al.

(Sup.) , ff. Tr. 24,372, at 12. LIIf0's argunent proceeds along two major lines as follows: (a) the circunferential cracks in EDG 103 did not prog-gate' to the point of impairing function despite 1200 hours0.0139 days <br />0.333 hours <br />0.00198 weeks <br />4.566e-4 months <br /> of operation ard the allegally inferior physical properties of that block (hence EDGs 101 and 102 with allegally superior properties will see even less crack growth); and (b) analytical predictions are that the cracks proggate into a decreasing stress field so that they will arrest before they becone deep enough to cause operational problons. _Id. at 12-14.

80. 'Ihe Staff agrees with LIIf0's analysis that the cracks will ar-rest, based on its ju3gment that tensile stresses decrease with distance fran the corner Q1ere the cracks fonn. Bush (Sup.) , ff. Tr. 25,775, at

7-8. We Staff's opinion is not based on cupirical analysis but, instead, on a review of the factors believed to be causing the cracks and applica-tion of " engineering jtdgment" to the sun of those factors (to which no particular values are assigned) . _Id_. at 8.

81. We munty asserts that the circunferential cracks could cause engine failure. Anderson, et al. (Sup.) , ff. Tr. 25,565, at 2.  % e W un-ty's basic argunent is that there is insufficient evidence on which to pre-dict the behavior of circunferential cracks. _Id . at 12 ; Andermn, et al . ,

ff. Tr. 25,554, at 180. We munty also challenged LIIro's assertion that the deepest crack discovered in EDG 103 was 3/8 inch deerdl/ based on Dr.

Anderson's observation of smaller, branch-like cracks radiating below the tip of that crack to a depth of 1 inch. Andermn, et al. (Sup.) , ff. Tr 25,565, at 11-12. Dr. Andermn believed the branch-like cracks indicated that the circunferential crack was propagating. _Id_. at 13.

82. FaAA disputes the existence of the brancklike cracks observed by Dr. Anderson. Drs. Pau and W1chab did not observe than and suggested that Dr. Anderson probably was confusing artifact of Widmanstaetten grarinite with cracks. Tr. 25,0%-97 (Rau, Wachob) . Both LIICO witnesses noted the.t Dr. Anderson had exanined a sample that was not metallographically polished (Tr. 25,098)48/ and Dr. Pau asserted that reliable conclusions could not te 47/ It is also pertinent that LIIro visually exanined only 2 of the 8 cyl-irriors of EDG 103. Id . at 11. It therefore cannot generalize about the depth of any craEs in the other 6 cylinders.

48_/ Dr. Andermn acknowicdged that fact. Tr. 26,354.

)

drawn in the absence of such polishing. Tr. 25,139.

83. Dr. Anderson responded that he viewd the sample under a microscope with power varying fran 20x to 60x and did not confuse graphite artifact for cracks. Anderson (Reb.) , ff. Tr. 26,326, at 2. Dr. Pau contended that Dr. Andermn's examination ws not reliable even using a microscope. Tr. 26,666-68. A photograph of the sample after liquid penetrant was applied to it was introduced as LIIfO Ex. B-64. 'Ihe witness-es disagreed about what it portrays. Tr. 26,684-88 (Rau, Anderson) . Dr.

Bush expressed the view that no firm coiclusions could be drawn fran review of the photograph. Tr. 26,688-89. We note that the exhibit is a macrograph, not a micrograph taken under any significant degree of magnifi-cation. Tr. 26,691 (Rau, Bush) . 'Ihe photograph does not assist us in resolving the controversy.49/

84. On balance, w cannot accept LIIf0's p3sition that Dr. Anderson's observations were not reliable. We consider it to be highly pertinent that Dr. Bush did not join the PaAA experts, expressing the view that he wuld have to examine the sample to form an opinion. Tr. 26,688-89. bbre 49/ In a sense, resniving this evidentiary conflict is not necessary in view of our conclusion below that even a 3/8-inch circunferential crack in EDG 103 is cause for concern and raises serious unremlved questions about the reliability of EIXis 101 and 102. However, the Irssibility that the one crack in ED3103, one of only two (out of the eight cylinders) sectioned, may have other subsurface cracks that could link up with the surface crack to a depth of 1 inch is cause for grave concern.

l l

)

I impartantly, all of the experts agreed that a definitive opinion wuld Tr.

depend on inspection of the surface after metallographic polishing.

LIIr0 has 25,139 (Rau); Tr. 26,664, 26,670 (Anderson); Tr. 26,682 (Ibsh) .

possession of the sample and apparently elected not to parsue that course (w are not aware of any such inspections) .50/ In view of all these factors, we feel conpelled to accept Dr. Andersan's testimony as reliable and to assute that there wre branch-like cracks belcw the surface that could link up and extend the main crack to a depth of 1 inch.

85. FaM testified that a finite elenent analysis of the counterbore area where circunferential cracks fonn indicates that the stresses becono fully conpressive approximately 0.4 inch deep (measure 3 along the 45 degree angte at which the cracks extend fran the counterbore/ cylinder liner land-irx3 corner) . Tr. 25,341, 25,343-44 (Rau) . The details of that finite ele-ment analysis are otherwise sketchy and the analysis appears to be based simply on hypothesis. Dr. Rau testified that the analysis did not use and In addition, the was not confirmed by any strain gage data. Tr. 25,505.

fact of the branch-like cracks below the 3/8-inch crack itself st:3 gests that the finite elonent analysis is wrong.

86. 'Ihc (bunty's position is that the circunferential cracks are dan-Christensen and Eley gerous because they could cause engine failure.

jp/ vb note that LIICO has not been reltx: tant to supplenent the record where it has been to its advantage to do so.

57 -

)

l

---

'-------._._.m_ _ _ _ _ _ _ _ _ _ _

(.

(Sup.) , ff. Tr. 25,565, at 13; Christensen and Elcy (Reb.), ff. Tr. 26,326, at'3. We fact that there were nunerous branches belcw the tip of the cir-cunferential crack fran EDG 103 which had been sectioned is cause for con-cern that the crack was proggating deeper. Anderson (Sup.) , ff. Tr.

25,565, at 13. We (bmty projects tm possible causes of engine failure:

(a) the crack could pennit the cylinder liner landing to flex under the force of the 1600-1700 psi of firing gessure, causing canbustion gases to leak outside the cylinder into the cooling ster and possible overheating of the engine; (b) the crack could gopagate through the counterbore, caus-ing the cylinder liner landing to seprate fran the block and the liner to fall into the crankcase. Christensen and Eley (Sup.), ff. Tr. 25,565, at 13; Christensen and Eley (Reb.), ff. Tr. 26,326, at 3. We (bmty does not assert either event as a probability, but simply a possibility that cannot be characterized as unlikely. Christensen and Eley (Reb.), ff. Tr. 26,326, at-3.

87. We Cbunty's experts had not perfonned any calculations to deter-mine the stresses in the cylinder liner landing area where the circunferen-tial cracks occurred in EDG 103. Tr. 26,355 (Anderson, Christensen, Eley) .

They explained that they did not have reliable input data to make such cal-culations and pointed to the fact that PaAA's reprts of stresses in the block top were contradictory. Tr. 26,355-57 (Christensen, Elcy) . An eval-uation of the stresses is very conplex because of the various forces at work in the area and that conplexity has caused modern engine designers to

+

i

+

ju s l

depart fran the' design reflected in the TDI R-4 engines in this area

- . because of that conplexity. Tr. 26,365-66 (Christensen) .

88. Similarly neither of- the (bmty's diesel experts had perfonned

' calculations to.detennine how deep a circumferential crack must propagate to permit up and down movenent of the cylinder liner. Tr. 26,367-69

(Christensen, Eley) .

i

! 89. According -to the Staff's witness, Mr. Henriksen, the Cbmty's

_ . . concern that conbustion gases might leak into the cooling wster as the re-i-

.. cult of movenent of the cylinder liner is "a little far-fetched."

t Tr.

1 26,216. L He explained that, because of the design of the engine as ,

1 reflected in Staff Ex.10, any leakage of gases would be into the atmo-f cphere and not.into the jacket water. Tr. 26,218-19. He also opined that, i

! even if conbustion gases were to leak into the jacket water synten, they c

) are not likely to cause overheating. Tr. 26,217-18. Mr. Henriksen's ex-planation of where conbustion gases would exit appears to be plausible and ,

. we accept it because the Cbmty did not offer any rebuttal. testimony. How- r ever, we are not confortable with the prospect of conbustion gases (and presunably fire) being snitted into the engine roon during an energency be-

- cause it is likely to cause confusion and panic during a situation Wiich i

denands calm heads. We have no evidence that any of the parties have con-aldered this aspect of the matter, nor the possibility of fire in the en-I gine ro7n as a result of conbustion gas leakage. Absent such evidence, we b cannot approve the possibility of such leakage as an acceptable risk.5,1/

51/,

1 W. Henriksen did not o2nment on the (bmty's concern that a circun-i ferential crack could cause failure of the cylinder liner landing and I' loss of. the cylinder liner itself.

J 4

l' ______ _

I J

90. FaM predicts that a circunferential crack is likely to propagate at rotghly a 45-degree angle fran the corner. Tr. 25,100 (Rau). FaM does not believe that such a crack is likely to propagate to the point of caus-irg engine failure because the structure is reinforced by 8 stul bosses for approximately one-third of the circunference of the cylinder liner landing.

Tr. 25,100 (Wells) . Hence, accordirg to FaM, the crack would have to prop 1 gate 4 to 5 inches to sheer throtgh those stul bosses and cause struc-tural failure. Id. at 25,100-01.52,/ Howver, FaM acknowledged that, stul boss areas aside, a circunferential crack propagating at a 45-degree angle will reach through to the jacket water within a distance of 1-1/2 to 2 inches. Tr. 25,511 (Rau) .

91. Dr. Wells disagreed with the Cbtnty's concern that a circunferen-tial crack could cause sane up-and-ckun revonent of the cylinder liner, asserting that the liner is actually supported by the cylinder stuis. Tr.

25,246. He asserted that the cracks would have to be several inches deep

-52/ Dr. Bush also beliexed that the stul bosses muld geserve the strtr-tural integrity of the cylinder liner landire in the event of circun-ferential crack propagation. Tr. 26,190. Ikuever, Dr. Bush's opinion that ciretmferential cracks will be self-lhnitirg and do not present a hazard to EIU reliability is not based on any calculations he perfonned. Tr. 26,019-20. In this case, little might can be .

attributed to Dr. Bush's opinion, because it is based solely on engi-neering julgment and he has no expertise in diesel engines. He expressed the opinion that the cracks would arrest at a depth of 1/2 to 3/4 inch, but describal that as a " qualitative julgment . . .a guesstimate." Tr. 26,021-22. He confirmed that his opinion wts based on engineering julgment: "It is simply that, an engineering julgment, nothing more." Tr. 26,151-52.

j l

1

\

for that movenent to occur. _Id . We cannot accept Dr. Wells' opinion here.

He has no exprtise in diesel engines and he did not profess to have undertaken any real analysis of the area. Even a layman can visualize the possibility of some flexire of the cylinder liner landing in the presence of a 360-degree circunferential crack and the forces exerted by 1600-1"00 psi of firing pressure. We need scxnething more than Dr. Wells' tnadorned Gxpression of opinion to igrure the testimony of the Wunty's diesel ex-perts on this point. On balance, we do not believe that either of the County's perceived failure mechanisms is so implausible that they represent acceptable risks, and w therefore find that the pssible existence of cir-cunferential cracks in EDGs 101 and 102 render those EDGs unsuitable for nuclear service.

D. The Evidence is not Sufficient to Establish that the Replacanent Block in EDG 103 is Reliable.

92. We Cbtnty asserts that the replacanent cylinder block in EDG 103 has not been tested adequately, based upon the depsition testimony of TDI's general rnanager to the effect that TDI never tested that new block design. Bridenbatgh, et al., ff. Tr. 25,564, at 182. Instead, TDI relied on the fact that the block top prtion of the replacanent block is the same as the top prtion of TDI's R-5 crgine block, and that block has been test-ed. Id. %e Cbtnty contends that testing of that design in a totally dif-ferent block is not adequate. Id.

i

93. LIIro asserts that the replacenent block is adequate because the block incorporates design changes which should enhance its strength over the blocks in EDGs 101 and 102. Wells, et al., ff. Tr. 24,372, at 68-69.

'Ihose changes are lengthening the stal head bosses, increasing the block top thickness, and increasing the cold clearance gap between the cylinder liner and comterbore. _Id . In addition, the block is cast fran Class 45 gray cast iron, a material possessing a higher tensile strength. Id at 69-70. LIIf0 also asserts that the gecmetry of the block top in the re-placenent block is virtually the same as that in the R-5 engine, that the R-5 engine has been tested extensively at higher power loads, and that the testing of the R-5 engine is applicable to the R-48 replacanent block.

Wells, ff. Tr. 24,372, at 71.

94. We accept that the design changes in the new EDG 103 block en-hance its strength over the blocks in EI:Gs 101 and 102. However, LIIf0 presented no quantitative analysis upon diich we can measure that en-hancenent. Accordingly, we have no basis upon which to conclu1e that it is sufficiently superior to those other blocks to be sufficiently reliable for service in a If0P/LOCA. tereover, LIIf0's assertion that the R-5 testing is aIplicable to the replacanent block is accanpanied by no explanation and is proffered by Dr. Wells, a metallurgist possessing no expertise in diesel engine design and manufacture. thder the circunstances, LIIf0's assertions d.o not meet the concerns expressed by the (bmty's diesel experts.

9

i E. 'Ihe Blocks are not Suitable at LIIf0's Proposed Olalified Ioad of 3300 kW.

95. As an alternative to approving the EDGs at the originally l

cpecified ratings of 3500/3900 kW, LIICO has propsed that we approve the EDGs for use at a new " qualified load" not to execul 3300 kW. We find that the evidence is not sufficient to assure that any of the blocks will oper-cte reliably even at the lowr 3300 kW load.

96. LIICO failed to introdtre any new evidence to supprt the adequa-cy of EDGs 101 and 102 at 3300 kW. It merely asserts that op3 ration at 3300 kW prodtres lomr cyclic stress than at 3500/3900 kW, and therefore redtres the risk of crack initiation and the rate of crack propagation.

Rau and mchob, ff. Tr. 28,799, at 10-11. LIIf0 also asserts that its cu-mulative danage analysis denenstrates that the blocks can withstand a IOOP/LOCA involvirg power requirenents of 3500/3900 kW, and that, there-fore, the engines will withstand a IDOP/LOCA at lower loads with even greater margin. Id. at 11-13. Finally, LIIf0 argues that EEGs 101 and 102 have survived operation for hundreds of hours at loads of 3500 kW or above without sttrl-to-sttti cracks, and that dononstrates their ability to survive o ICOP/ILCA involving brief excursions above 3300 kW. Id. at 12.

97. We find that LIIf0 has not sustained its burden of proof on this issue. LIIf0 failed to introduce any evidence W1atrever abcut ligament and circtmferential crack behavior at 3300 kW, apparently content with its h

analysis at 3500/3900 kW. Si ce ve do not accept LIICO's analysis at those loads for the reasons discussu above, its failt:re to address these con-

)

cerns at 3300 kW constitut.es a failure of proof which is fatal to its case.

98. In addition, LIIr0 could not quantitatively state how the reduc-

! tion in load frcm 3500 to 3300 kW would affect : fatigue crack initiation.

Tr. 28,902-03 (Rau) . LIICO's own analysis predicts that sttd-to-sttd cracks will initiate at loads of 3150 kW and above. See Finding 43, supra.

Moreover, no one knows enough about crack behavior to predict what will happen with the EDGs in the future.53/ LIIf0 relies primarily on its etzt>-

lative danage analysis for its prediction that 3300 kW would present no problen if a attd-to-sttd crack initiate 3. Set Tr. 28,909 (Rau) . However, we have already rejected that analysis as unreliable (see Findings 37-49, aupra) and LIICO failed to present any new evidence to support the reliability of that analysis at 3300 kW.

127. 'Ihe Staff agrees with LIICO that the blocks in EDGs 101 and 102 cre sttitable for the qualified load of 3300 kW because no sttd-to-sttd crack developed frcm operations at 3500 kW an3 above. Bush, ff. Tr.

53/ Even at the restmo3 hearings in bhrch 1585, FaM could not say how

~

long it takes for a liganent crack to gmw 1-1/2 inches deep, nor how long it would take for a sttd-to-sttd crack to initiate in EDG 101 nor how long it took for the deepest attd-to-sttd crack to grow to 3 inches deep in EDG 103. Tr. 28,894-95; 28,896-97 (Rau) . Dr. Rau ad-mittei that no one knows the specific c::ack propagation rate of any crack. Tr. 28,897. It is therefore c1mr that LIICO lacks the data essential to permit it to pratict crack behavior.

I I

I 28,503, at 31. Howver, Dr. Bush did not know how mtch more likely it is for cracks to develop at 3500 kW instead of 3300 kW. Tr. 29,024. Although his prefiled testimeny asserted that he muld not expect sttd-to-stui crack grwth to be high enough to cause a loss of EDG function during a LOOP /LOCA boundal by 3300 kW, he perfonnai no specific analysis in the area. Bush, ff. Tr. 28,503, at 32; Tr. 29,030; 29,032. Indeal, he lacked the data needal to calculate hov long it would take for such a crack to propagate to a depth of 3 inches and did not know how deep a sttd-to-sttd crack must propagate to cause engine failure. Tr. 29,033; 29,057-58. His opinion ap-pears to be based primarily on his belief that the so-called load excursion caused most of the damage in EIr,103 during its qualification testing, a belief which has no scientific foundation. Tr. 29,038-40; 29,013-44.

128. LILCO an3ues that its endurance testing of EDG 103 describal in Crankshaft Findings 3-6, infra, establishes that the replacanent block is cuitable for loads up to 3300 kW. Johnson et al., ff. Tr. 28,799, at 3, 8.54/ lbwaver, as discussed at Crankshaft Findings 3-6, infra, the bulk of 54/ LIIro admits that the endurance run does not disclose anything about the endurance limit of the blocks on EEGa 101 and 102, because that test was parformed on EIG 103 with a replaconent block that has a sig-nificantly different design and a tensile strength substantially higher than that of EDGs 101 and 102. Tr. 28,848-49, 28,866-67 (Rau) .

'Ihe Staff alno agrees that the use of IIG 103 for the endurance test did not contribute to the resalution of issues concerning liganent cracks in EEGs 101 and 102 blocks, the Fotential for developing stui-to-stui or sttd-to-end cracks in those blocks, or the circunfer-ential cracks in the original block in EDG 103. Dush, ff. Tr. 28,503, at 24, 31.

t that testing ms condtx:ted at loads that conservatively can be viewd as no higher than 3230 kW due to instrtment accuracy. See also Tr. 29,053 (alsh) .5,5/ Accordingly, that endurance test only establishes that EDG 103 l

is adeqtate for operation up to 3230 kW. We cannot asstme that the addi-tional load on the EIX3s at 3300 kW would not be the proverbial straw that breaks the canel's back.

55/ Indeed, the fact that LILCO chose to run the endurance test on EDG 103

~

rather than on either EDG 101 or 102 stygests that LIICO itself is -

concerned about the reliability of the blocks on those engines. Al-thatgh the replacanent block on EDG 103 was more suitable than either the EDG 101 or 102 block to obtain data on stresses in the canshaft gallery (Dush, Tr. 28,503, at 24), LIICO admits that the time neces-sary to obtain that data ms approximately 10 to 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br />. Tr. 28,875 (Rau) . 'Dius, there ms little or no necd to perfonn the entire endur-ance test on EDG 103. Indeed, successful ecmpletion of the testing on either EDG 101 or 102 might have rerolved any issue concernity their reliability at the load levels tested. Bridenbatgh, ff. Tr. 28,918, at 4. See also_ Crankshaft Findin) 3 n. 4, infra. Even LIIro admits that testing EIX; 101 or 102 would have provided valtnble data as to whether its analysis of the blocks ma indeed conservative and might have climinated high cycle fatigue as a mechanism for crack initiation. Tr. 28,853-54, 28,878-01 (Rau) .

I. INTRODUCTION

1. In November 1984, LILCO revised the Shoreham licensing basis to reduce the emergency loads that the EDGs would be committed to sup-ply. The revised FSAR now provides that the maximum load imposed on i

the EDGs may not exceed a " qualified load" of 3300kW. LILCO reduced its previously projected maximum coincidental demands on the EDGsl/

to the maximum emergency service load ("MESL") totals by, among other things, ignoring. cyclic and intermittent loads and by adjusting sever-al other loads on the basis of measured rather than nameplate data.

Bridenbaugh and Minor, ff. Tr. 27,500, at 5-6.

2. Prior to LI CO's FSAR revision, the Staff established an in-terim licensing basis for TDI diesel generators which, as applied to the EDGs, required LILCO to run one of the EDGs for 10 to the 7th power ("10E7") load cycles, or approximately 745 hours0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br />, at or above their maximum continuous emergency service load. Dawe, et al., ff.

Tr. 27,153, at 9-10 Tr. 27,343 (Youngling). LILCO chose to qualify the EDGs at a load of 3300kW by operating EDG 103, which already had I

run for approximately 220 hours0.00255 days <br />0.0611 hours <br />3.637566e-4 weeks <br />8.371e-5 months <br /> at nominal loads of 3500kW and above, for an additional 525 hours0.00608 days <br />0.146 hours <br />8.680556e-4 weeks <br />1.997625e-4 months <br /> to achieve 10E7 cycles. LILCO operated

, EDG 103 during this test at a nominal load of 3300 +100kW. Dawe, et j' al., ff. Tr. 27,153, at 38.

~

1/ Previously, the Shoreham FSAR had required each EDG to be rated to operate continuously at full load of 3500kW and for 2 hours2.314815e-5 days <br />5.555556e-4 hours <br />3.306878e-6 weeks <br />7.61e-7 months <br /> per overy 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> at overload of 3900kW based on a projected maximum coincident demand for the highest loaded EDG of 3881.4kW for the first ten minutes of a LOOP /LOCA, after which the projected maximum load was re-

duced to 3409.2kW. Hubbard and Bridenbaugh, Tr. 23,826, at 14-16.

i

_ ____.______m.___._.________________________.__m___. _ . _ _ _ _ . _ _ _ _ _ _ . _ _ _ _ . _ _ . _ _ _ . _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _

II. DISCUSSION A. LILCO and the Staff Have Applied-A Novel Method for Complying with GDC 17 Requirements which Departs from Standard Practice and is Inadequate.

3. The standard practice in the nuclear industry for 4 ,

determining compliance of diesel generators with GDC 17 is to refer to their continuous and short-time ratings. Tr. 27,759-60 (Berlinger).2/

To ensure that a diesel generator has sufficient capacity and capabil-ity to handle the maximum loads imposed on it, the Staff's standard 4

practice is to confirm that the continuous rating encompasses the con-tinuous loads and that the short-time rating encompasses intermittent and cyclic loads and an additional load that may be inadvertently con-nected by operator error. Knox, ff. Tr. 27,735, at 9; Tr. 27,997-98 '

(Knox); Tr. 27,952-53, 27,980 (Berlinger); Bridenbaugh and Minor, ff.

i Tr. 27,500, at 11. The Staff knows of no instance in which the short-time rating of an emergency diesel generator at a licensed nu- l clear plant does not encompass the maximum continuous load plus the highest single load that could be connected erroneously by an operator.3/ Tr. 27,957 (Hodges); Tr. 27,960, 28,036 (Buzy).

1 i

2/ Indeed, until this case, the Staff has always required that diesel generators used as sources of emergency onsite power have continuous and short-time ratings that reflect their output capabilities. Tr. 27,751, 27,758, 27,992 (Knox). The Staff generally accepts these nameplate rat-ings as accurate indicators of the capacities and capabil-ities of the diesel generators. Tr. 27,992 (Knox); Tr.

27,968 (Berlinger).

3/ For simplicity, we will refer to this as the " operator ,

error load".

i 1

4. In this case, however, the Staff and LILCO have abandoned

, this standard practice and the usual method for determining compliance i with GDC 17. The revised FSAR now provides that the qualified load i "will be used for all purposes," thus the EDGs have no continuous or short-time rating, only one " qualified load" rating. Bridenbaugh and l

Minor, ff. Tr. 27,500, at 6; Tr. 27,180-81 (Dawe). In effect, LILCO i proposes to use this qualified load rating as the functional equiva-lent of continuous and short-time ratings. Tr. 27,182-83 (Dawe).

However, the qualified load rating of 3300kW is not high enough to fulfill the function of a short-time rating. Prior to this case, GDC 17 consistently has been applied such that the margin between the highest (usually short-term) rating of an emergency diesel generator and the maximum load is sufficient to encompass intermittent and cyclic loads as well as the operator error load. Knox, ff. Tr.

27,735, at 9. In this case, however, the 1.4 percent margin between i LILCO's highest MESL estimate and the qualified load rating of 3300kW does not accommodate the maximum loads imposed on the EDGs due to cyclic and intermittent loads, the operator error load, and loads due i

to non-conservatisms used by LILCO in estimating the MESLs.

5. The Staff adopted the novel qualified load approach because it concluded that the continuous and short-time ratings of the Shoreham EDGs could not be relied on to assure that the EDGs had suf-ficient capacity and capability to comply with GDC 17. The Staff i

reached this conclusion because (i) TDI had an inadequate quality assurance program, (ii) the EDGs had unsuccessful operating experi-ence, and (iii) the adequacy of the replacement crankshafts could not L

be determined conclusively. Tr. 27,962-65, 27,968-69, 27,981-84 (Berlinger). If the qualified load established by endurance testing were high enough to serve the function of a short-time rating, the Staff's approach, in our view, would not be unacceptable. However, i

for the EDGs, the Staff only required that their qu'alified load exceed the continuous loads. Tr. 27,962-89 (Berlinger). This approach fails to provide reasonable assurance that +he EDGs can handle the loads or-4 dinarily encompassed by the short-time ratings. Indeed, the Staff

admits that endurance testing at the qualified load does not demon-strate the capability to handle such loads. Tr. 27,989 (Berlinger).

The ef fect of the Staff's approach is to apply a lower standard of i compliance with GDC 17 to the Shoreham EDGs than is applied to reli-able emergency diesel generators.4/ We reject the use of this lower standard for the EDGs.

4/ The Staff contends that endurance qualification testing is not a lower standard because the extent of testing required of EDG 103 may be more than required of other emergency diesel generators in nuclear plants. Tr.

27,987-91 (Berlinger). That testing, however, is no more rigorous than the qualification testing used by manufac-i turers of reliable diesel generators to determine their I continuous ratings. Tr. 27,962 (Berlinger); Christensen

and Eley, ff. Tr. 23,826, at 18-19. In any case, the l Staff's argument does not address the fact that it has eliminated the functional requirement of a short-time rat-

i. ing for the Shoreham EDGs.

6.- The need for avoiding'EDG operation above the qualified' load (or, for diesels . reliably manufactured, above the established short-time rating) is implicit in the language of GDC 17, which requires that the EDGs provide sufficient capacity and capability to assure various safety functions. In addition, sufficient margin is required by. Regulatory Guide 1.9, particularly where, as here, the magnitudes of all loads are not precisely known. Bridenbaugh and Minor, ff. Tr.

27,500, at-13-14.5/

'7. As discussed in detail below, the EDGs' highest rating, the proposed. qualified' load of 3300kW, is inadequate to envelope the maxi -

mum loads that may be imposed on the EDGs. The qualified load is exceeded by a combination of MESL with cyclic and intermittent loads, loads due to non-conservatisms used by LILCO in estimating the MESLs, and the operator error load. Accordingly, the qualified load of 3300kW is inadequate to provide the EDGs with " sufficient capacity and capability to assure" that the requirements of clauses (1) and (2) of GDC 17 will be met. See Bridenbaugh and Minor, ff. Tr. 27,500, at 8, 11, 13-15, 17-21.

-5/ LILCO claims that Section 5.23 of IEEE Standard 387-1977 establishes that no margin is required between the design load and the EDG rating. Tr. 28,173 (Berlinger). Section 5.23, however, provides only that the EDG rating must en-compass the actual loads imposed on diesel generators and does not address whether and to what extent margin is required in load estimates. In any event, the qualified load does not encompass the design load (see note 12, infra) which clearly violates Section 5.23.

4

+

, e ..-- a, 3,e- -...mm-= %ef -,--,.m-- - - - - - ,-r,- w_.e- -

-m--y_--... y.-,,,,c - --n., .m,

.l. The Load Combination Methodology Used by LILCO to Estimate the MESLs Fails to Account for Inaccuracies in Nameplate Ratings and Load Measurements.

8. -The. actual values (during a LOOP /LOCA) of the individual loads that make up the MESLs are not precisely known. Bridenbaugh and Minor, ff. Tr. 27,500, at 17. In estimating the MESL totals for each EDG,'LILCO used a load combination methodology that relies on measured loads and nameplate ratings. . Bridenbaugh and Minor ff. Tr. 27,500, at 8, 19-20.6/ Although the accuracy of this methodology cannot be pre-cisely identified, we agree with the County that the MESLs can be con-servatively estimated with some degree of certainty to be as much as 128kW higher than LILCO has estimated. We base this finding on the experience of the County's witnesses that, at best, measured loads may conservatively be assumed to be no more accurate than 12-1/2 percent system error, and that nameplate ratings are no more accurate than 15 percent. Bridenbaugh and Minor ff. Tr. 27,500 at 19-20; Tr. 27,606-08 (Bridenbagh, Minor). Thus, when considering only the potential inac-curacy of LILCO's load combination methodology, the maximum load on the EDGs will exceed the qualified load.

-6/ The four or five loads actually measured and used by LILCO j -in generating the MESL for each EDG account for 34, 35 and 64 percent of the MESLs of EDGs 101, 102 and 103, respec-tively. Bridenbaugh and Minor, ff. 27,500, at 19-20: Tr.

27, 210 (Youngling). As discussed in note 8, infra, LILCO actually measured an additional 22 or 23 loads, but used their nameplate ratings in the MESLs rather than the mea-surements.

9. As explained by the County's witnesses, nameplate ratings are fairly accurate when comparing, for example, the rating of a par-ticular motor with its kilowatt requirement at a particular horsepow-er. However, much more opportunity for error in the nameplate rating is introduced when that particular piece of equipment is operated in conjunction with a complicated system such as an EDG. Tr. 27,606-07 (Bridenbaugh). Variations also exist in the way that different manu-facturers rate their equipment. Nameplate ratings may be assigned to equipment for a particular set of operating conditions (e.g., a par-ticular voltage or flow) that may differ from the condition in which the equipment is applied, such that the kW rating differs from the ac -

tual kW output. Tr. 27,607-08 (Minor).7/

10. LILCO argues that the nameplate ratings are conservative and should be more accurate than to within +5 percent. Tr. 27,201-02 (Dawe); Tr. 27,217 (Dawe). In support of that position, LILCO relied primarily on the assertion that 26 of 27 of the loads that were measured by LILCO were found to be below their nameplate ratings. Tr.

7/ Similarly, the conditions and adjustments used by LILCO during load measurements may differ from the system condi-tions during a LOOP /LOCA. See Bridenbaugh and Minor, ff.

Tr. 27,500 at 18-20; Tr. 27,608 (Minor); Tr. 27,606-07 (Bridenbaugh). Furthermore, LILCO did not conservatively consider possible instrument error in its measured loads in the MESLs which would increase the loads on EDGs 102 and 103 by approximately 22kW each and on EDG 101 by 11kW.

Tr. 27,214 (Youngling); Bridenbaugh and Minor, ff. Tr.

27,500, at 20.

27,214-15 (Youngling). That evidence, however, fails to address the County's concerns about system error, and the load measurements that were made do not necessarily indicate whether the nameplate ratings for the unmeasured loads are similarly conservative. More impor-tantly, LILCO arrived at the MESL estimates in part by reducing the previous estimates (one by as much as 141kW) through the use of measured. loads rather than nameplate ratings. Tr. 27,203-05, 27,207-09 27,461 (Dawe, Youngling). This is contrary to industry practice and further degrades the already negligible margin by eliminat.ing the conservatism present in those nameplate ratings. Tr.

27,575-76 (Bridonbaugh); Bridonbaugh and Minor, ff. Tr. 27,500, at 8, 17.8/

11. LILCO also claims that the MESLs are conservative because they are based on the assumption that all equipment operates simulta-noously at their nameplate or measured loads and further asserts that all equipment does not operato simultaneously at those loads. Dawe, 8/

~

LILCO also asserts that the MESLs are conservativo becauso it did not take credit for the approximately 30kW diffor-once betwoon the nameplato ratings and the measured loads that were not used in the MESL. Tr. 27,213-14 (Youngling). The insignificance of this minor conserva-tism is demonstrated by the fact that the 30kW difference resulted from the measurement of 22 or 23 loads (i.e., the averago difference betwoon the nameplato rating and the measured load was loss than 1.5kW) and nono of the measured loads woro conservatively corrected for possible instrument arror. Tr. 27,204-05 (Dawo); Tr. 27,214 (Youngling). Soo noto 7, supra.

et al., ff. Tr. 27,153, at 19-21: Tr. 27,190-91, 27,201-02 (Dawe). In l support of its position, LILCO relies on the results of the integrated l electrical test ("IET") which allegedly measured the cumulative loads on the EDGs in a simulated LOOP /LOCA condition. The results of the IET purport to show that the actual loads during a LOOP /LOCA are 400 to 500 kW 1ess than the MESLs. For the reasons discussed below, how-ever, we do not accept the IET as an accurate simulation of LOOP /LOCA conditions and therefore cannot accept LILCO's assertions.9/

1

12. First, as discussed below, the MESLs lack conservatism in that they fail to consider cyclic and intermittent loads and the oper-1 ator error loads. Thus, contrary to LILCO's assertion, the MESLs do not assume that all equipment operatos simultaneously at the nameplate 1 or measured values (Dawe, et al., ff. Tr. 27,153, at 19); they assume that only the continuously loaded equipment does.

F i

13. Second, although the results of the IET suggest that not all equipment operates simultaneously at their nameplate or measured val-l ues, we do not have reasonable assuranco that the IET accurately simu-lated LOOP /LOCA conditions. Tr. 27,543-46 (Bridonbaugh); Dridenbaugh and Minor, ff. Tr. 27,500, at 17-19. The primary purpose of the IET

, 9/

We find little comfort in the Staff's view that the MESLs j conservative. Tr. 27,738 (Berlinger and Knox) Tr.

27,756, 27,788-89 (Berlinger). The Staff did little or nothing to verify the accuracy of the loads listed in the MESLs. Tr. 28,128, 28,130, 20,189 (Knox).

was not to determine the actual loading on the EDGs during a LOOP /LOCA, but to demonstrate their independence and redundancy. Tr.

28,152 (Knox); Tr. 27,545-46 (Bridenbaugh); Tr. 27,533, 27,570 (Minor) Dawe, et al., ff. Tr. 27,153, at 19. Moreover, the Staff could not conclude that the IET was an accurato simulation of plant conditions during a LOOP /LOCA and therefore did not consider the IET results in evaluating the adequacy of the qualified load. Of great significance to the Staff was the recommendation of LILCO's engincor-ing and licensing personnel that the IET results should not be used in evaluating or ostimating the actual EDG load requirements during a LOOF /LOCA because the IET could not wholly and accurately model plant conditions during an accident. Tr. 28,273 (Dorlinger) Tr. 28,151-56 (Berlingar, Knox). We find it particularly persuasive that the Staff could not concludo that the IET was an accurate simulation of plant conditions during a LOOP /LOCA. We are reluctant to give any weight to the IET results given the views of LILCO's personnel on their accura-cy.

14. Furthermore, the apparent largo discrepancy (approximately 400kW to 500kW) betwoon the IET results and the MESL totain also sug-goats that the IET did not accurately simulato LOOP /LOCA conditions.

Tr. 27,543-46 (Dridonbaugh): Tr. 27,208, 27,461 (Dawo). Although l

LILCO offorod a number of possible explanations for a part of this largo differenco,12/ LILCO could not provido any basis whatsoever to 10/ Soo Tr. 27,642-83 (Dawo, Youngling). LILCO estimated that a portion of tho difference (approximately 150kW each (Footnoto cont'd next page) account for a significant portion of that difference. Moreover, while LILCO's partial explanations appear to be plausible, we have no rea-sonable assurance that these explanations do in fact hold true, be-cause individual equipment loads and pump flows were not measured dur-ing the IET. Tr. 27,200 (Youngling); Bridenbaugh and Minor, ff. Tr.

27,500, at 17. Thus, because LILCO's explanations for the large dif-ference between the IET results and the MESL estimates neither fully account for that difference nor are verifiable in any event, we do not rely on the IET as an accurate simulation of LOOP /LOCA conditions which establishes the conservatism of the MESL.

(Footnote cont'd from previous page) on EDGs 101 and 102, and 300kW on EDG 103) is attributable to the dynamic response of the four RBSVS and CRAC chillers. Tr. 27,643-44, 27,650-51, 27,668 (Dawe).

LILCO also claimed that the dynamic response of the bat-tory chargers and loop level pumps accounted for differ-onces of approximately 10kW and 6kW, respectively (Tr.

27,644-46 (Dawe)), and that conservatisms in the nameplate ratings of the RBCLCW, RHR and core spray pumps accounted for approximately 40kW, 30-90kW and 110-145kW of the dif-ference, respectively. Tr. 27,648-49, 27,652, 27,657-58, 27,663 (Dawe). With respect to the RHR pumps on EDGs 101 and 102, however, LILCO admitted that the pump flow during the IET was only 90 percent of the flow used in the MESL measurement of the RHR pumps. Tr. 27,652-53 (Dawo).

Initially, LILCO had claimed that a large portion of the 400 to 500kW difference was attributable to the timing of the applicatiou of loads (Tr. 27,412, 27,461-62 (Dawe):

Tr. 27,545 (Bridenbaugh), but after cross-examination LILCO modified this position. Tr. 27,642 (Dawe).

I  ;

i l

l l

2. The Qualified Load does not Encompass Cyclic and Intermittent Loads

15. The MESL for each EDG excluded intermittent and cyclic

-loads. When these loads (diesel generator air compressor, diesel gen-erator fuel oil transfer pumps, motor operated valves ("MOVs"), and non-operating MOVs), are added to the MESL estimates, the maximum load imposed on two of the EDGs exceeds the qualified load of 3300kW (3427.3kW for EDG 101, 3360.7kW for EDG 102, and 3284.6kW for EDG 103). Bridenbaugh and Minor, ff. Tr. 27,500, at 9-11. LILCO's esti-mate of these loads (3331.4kW, 3285.4kW, and 3284.6kW, respectively) do not include non-operating MOVs loads. Dawe, et al., 27,153, at 18.

16. LILCO's exclusion of the cyclic and intermittent loads . from the MESLs is unprecedented. At other nuclear plants, cyclic and in-termittent loads are bounded, if not by the continuous rating, by the short-time rating. Bridenbaugh and Minor, ff. Tr. 27,500, at 11. As previously noted, the Staff's standard practice is to assume that cyclic and intermittent loads occur simultaneously with the MESLs and to confirm that the short-time ratings encompass these loads. Tr.

27,992, 27,998 (Knox). Indeed, LILCO's previous FSAR recognized and was consistent with this standard industry practice, as the EDGs' short-time rating encompassed the cyclic and intermittent loads. Tr.

27,323-28 (Dawe). The qualified load for the Shoreham EDGs, however, does not encompass these loads, and we find no justification for this departure from the standard interpretation of GDC 17 requirements.

t-l l

l

17. LILCO argues that the cyclic and intermittent loads are in-significant because they are not expected to occur simultaneously or for long periods of time (Dawe, et al., ff. Tr. 27,153 at 12-14, 16-18). However, they will be applied automatically to the EDGs in the first minur.es following the initiating event of a LOOP /LOCA when the EDG loads are expected to be the highest, and most of these loads i will continue to be applied throughout the course of the accident.

Bridenbaugh and Minor, ff. Tr. 27,500, at 11-12; Tr. 27,526-27 (Bridenbaugh).ll/

18. The Staff accepts LILCO's exclusion of cyclic and intermit-tent loads from the MESL as appropriate (Tr. 27,742-43 (Berlinger)),

but that position is inconsistent with the Staff's standard practice end its actual review of the adequacy of the 3300kW qualified load.

The Staff's review conservatively assumed that cyclic and intermittent loads occur simultaneously with the maximum continuous load and that the totals of these combined loads exceeded the qualified load.12/

--11/ Depending on the particular configuration of the system at the initiation of a LOOP /LOCA, some of the non-operating MOVs may automatically initiate in the first ten minutes of a LOOP /LOCA. The single largest load from a non-operating MOV is 40 kW. None of these valves are locked j out automatically on a LOCA signal. The majority of'the

non-operating MOVs, however, require manual initiation to operate. Bridenbaugh and Minor, ff. Tr. 27,500, at 11; i Tr. 27,526-27, 27,555-56 (Bridenbaugh). Although LILCO identified one circumstance in which non-operating MOVs also may be manually applied later in a LOOP /LOCA (Tr.

27,306 (Dawe)), there is no evidence demonstrating that non-operating MOVs will not be used at other times, as the County contends.

12/ In establishing the qualified load concept as an interim licensing basis, the Staff did not specifically intend to (Footnote cont'd next page)

Thus, the Staff required- that the EDGs must be qualified to support -

.these loads. Tr. 27,742-43, 28,003-06, 28,177, 28,281 (Berlinger);

Tr. 27,752, 27,758-59, 28,186, 28,194-95 (Knox); Knox, ff. Tr.'27,735,

'at'4-5.13/

3. The Qualified Load does not Encompass

.the Operator Error Load

19. The MESLs also excluded the operator error load that could

.be applied to the EDGs.14/ Tr. 27,627-28 (Minor); Bridenbaugh and (Footnote cont'd from previous page) exclude cyclic and. intermittent loads from the loads that were to be encompassed by the qualified load. However, since LILCO chose 3300kW as the qualified load, the Staff had to determine whether the EDGs were adequate at loads above 3300 kW because the EDGs' design load, which in-

.cludes cyclic and intermittent' loads, exceeds the quali-fled-load. Knox, ff. Tr. 27,735, at 4-6; Tr. 27,745-46, 27,993, 27,995, 28,215 (Knox); Tr. 27,790, 28,003-06 (Berlinger).

13/

If the only issue in this case were whether the qualified load of 3300kW was inadequate only because intermittent and cyclic loads could cause the maximum loads on the EDGs to exceed the qualified load, we might have accepted LILCO's calculations of the maximum cyclic and intermit-tent loads and determined that such loads are de minimus in their impact on the qualified load. However7 we do not dismiss these loads because they combine with the operator error load and non-conservatisms used in estimating the MESLs to cause the maximum EDG loads to further exceed the qualified load.

14/ As previously explained in note 3, supra, by the term "op-erator error-load" we mean the largest single load which an operator could erroneously add to an EDG. .

Minor, ff. Tr. 27,500, at 28. Such loads by themselves could increase the maximum load on one cf the EDGs to over 3740kW, far in excess of the qualified load.15/

20. Although the Staff contends that the highest (short-time) ratings of other emergency diesel generators are not required by GDC 17 to accommodate operator error load (Tr. 28,174 (Knox); Tr. 27,948, 28,149, 28,281 (Berlinger)), that position is inconsistent with its standard practice and with its review in this case. As previously noted, the Staff's standard practice is to confirm that the short-time ratings in fact encompass this load. Knox, ff. 27,735, at 9; Tr.

27,998 (Knox); Tr. 27,952-53, 27,980 (Berlinger).16/ The Staff knows 15/ For the LOOP /LOCA condition, when the maximum single load that could be erroneously applied to each of the EDGs is added to the MESLs, the total loads are increased to at least 3459.4kW, 3414.8kW and 3583.5kW, for EDGs 101, 102 and 103, respectively. For the LOOP-only condition., the total loads are even higher, 3741.8kW, 3575.2kW and 3709.9kW for EDGs 101, 102 and 103, respectively. Dawe, et al., ff. Tr. 27,153, at 32-34; Bridenbaugh and Minor, ff. Tr. 27,500, at 28-29. See also Tr. 28,037-38 (Hodges).

~~16/ Contrary to the assertion in LILCO's Proposed Finding L-55 l

n.5, Dr. Berlinger's testimony is not inconsistent with I

this finding. Dr. Berlinger testified that the Staff pre-l viously has not " determined" the short-time capacity and capability of diesel generators by adding the operator error load to the maximum coincident demand. Tr. 28,277, 28,280 (Berlinger). In fact, the Staff, after confirming that the short-time rating encompasses the operator error load, " determines" through testing whether in fact the diesel generators have the capability and capacity to han-die such loads. Knox, ff. Tr. 27,735, at 4.

of no instance in which the short-time rating of an emergency diesel generator does not encompass the maximum continuous load plus the op-erator error load. Tr. 27,757 (Hodges); Tr. 27,960, 28,036 (Buzy).17/

Indeed, because the EDGs have no short-time rating, the Staff's review in this case conservatively assumed that the qualified load of the EDGs would'be exceeded by loads up to 3900kW for short periods and required that the EDGs be capable of supporting these loads. Tr.

28,176 (Berlinger); Bush, ff. Tr. 28,503, at 21-22. We believe that the Staff,'s practice is better evidence of what GDC 17 requires than its ad hoc testimony here.

21. - To assure that the EDGs have sufficient capacity and capa-bility, the qualified load must envelope the operator error load.

Bridenbaugh and Minor, ff. Tr. 27,500, at 29. Such an adequate quali-fled load is important for a number of reasons. Human error cannot be precluded in the operation of any equipment. No procedures and training can ensure that an operator will rot erroneously add loads.

At Shoreham, the operation of'EDG equipment in emergency situations is relatively complex and offers many opportunities for error.

17/- The fact that EDG 103's former short-time rating did not encompass loads due to operator error does not support LILCO's view that short-time ratings are not intended to encompass such loads; rather, it demonstrates the inade-quacy of EDG 103's short-time rating. Hubbard and Bridenbaugh, ff. Tr. 23,826, at 17-18; LILCO's Proposed Finding L-55, at 21.

Furthermore, operator error is likely to be more prevalent under the stress and strain of the rapid responses that would be required in the event of a LOOP /LOCA. Bridenbaugh and Minor, ff.'Tr. 27,500, at 24-28; Tr. 28,121-23, 28,165 (Buzy, Eckenrode,-Clifford); Tr.

27,885-86 (Berlinger). In addition, operators may give lower priority

-to maintaining the EDG load below the qualified load limit of 3300kW, particularly since LILCO expects the operators to continue to have confidence in the higher nameplate ratings of th'e EDGs. Tr. 27,842 (Clifford).18/ Indeed, the operators may violate the procedures and exceed the 3300kW limit if in their judgment it is necessary for safe operation of the plant. Tr. 27,916, 28,074, 28,306-07 (Clifford).

22. Even if LILCO's procedures and training could reduce the chance of exceeding 3300kW to an acceptable risk (and we believe this is not the case), that would inherently entail less flexibility for the operator to respond to an emergency. Tr. 27,542, 27,566 (Bridenbaugh); Tr. 28,092 (Clifford, Eckenrode). We do not believe that maintaining the EDGs at a maximum load level (which was 18/ Nearly all of the training of operators for response to accident situations is to assure that adequate or surplus cooling is being delivered to the reactor core. Operators will tend to maintain full pumping of all cooling systems until there is positive assurance that some systems are not necessary for the situation. As the TMI accident dem-onstrated, plant conditions may not be understood for hours or even days. Bridenbaugh and Minor, ff. Tr.

27s500, at 24; Tr. 28,073-74 (Clifford, Hodges).

v.

k' established by LILCO unilaterally) justified reducing an operator's

[

ability to deal with an accident. [ NOTE: The material that follows in this section is relevant only if the Board concludes that proce-dures and training could, conceptually, eliminate the problem of oper--

ator error load]. In any event, LILCO has failed to carry its burden of proof on the issue of the adequacy of the procedures and training.

Suffolk County's concerns over LILCO's procedures were not adequately addressed by LILCO, and the County concurred in many of the Staff's concerns. Bridenbaugh and Minor, ff. Tr. 27,500, at 25-30; Tr.

27,542, 27,566 (Bridenbaugh).

23. Significantly, as of the end of the hearing, the Staff could not conclude that LILCO's procedures and training provide reasonable assurance that the qualified load will not be exceeded.19/ In partic-ular, the Staff could not find that there was reasonable assurance that (i) procedures and training would not lead the operators to load the EDGs over 3300kW, (ii) procedures and training provide the neces-sary guidance to reduce the EDG load to less than 3300kW within one hour, and (iii) the training program adequately addressed technical 19/ The Staff's analysis was based on the assumption that the qualified load adequately enveloped the MESLs. Tr. 27,738 (Berlinger); Tr. 27,882 (Clifford). As we have noted above, there is no reasonable assurance that the MESLs im-posed on the EDGs will not exceed the qualified load.

Thus, the qualified load may be exceeded even without an operator error load.

concerns associated with the EDG load limit of 3300kW. Clifford, et al., ff. Tr. 27,732, at 9-10; Tr. 28,107-08 (Clifford).20/ Although LILCO has committed to perform a job task analysis that should address these concerns, that analysis must be conducted and completed before an' objective determination can be made that will provide a basis for determining whether or not the procedures and training are adequate.

Tr. 28,290, 28,292 (Clifford); Tr. 28,329-30, 28,336-37 (Buzy).

24. Delegation of this issue to the Staff for informal resolu-tion is not appropriate.21/ In Commonwealth Edison Co. (Byron Nuclear Power Station, Units 1 and 2), ALAB-770, 19 NRC 1163, 1175 (1984), the Appeals Board held that the issue of the adequacy of certification 20/ The Staff's view is based on numerous concerns regarding, among other things, whether the operators are physically able to manage the procedures, whether the procedures ade-quately prioritize between potentially' conflicting proce-dures, whether the wording of cautions in procedures is adequate, and whether operators have sufficient flexibili-ty for manual load transfers with the 3300kW load limita-tion. See, e.g., Tr. 27,824-29, 27,840-42, 28,052-53, 28,056-57, 28,082-83, 28,089, 28,091-97 (Clifford); Tr.

28,100-101, 28,337 (Buzy); Tr. 28,081, 28,092-93, 28,100 (Eckenrode); Clifford, et al., ff. Tr. 27,732, at 7-9.

~~21/ The Commission has long held that "[a]s a general proposi-tion, issues should be dealt with in the hearings and not left over for later (and possibly more informal) resolu-tion...[T]he ' post-hearing' approach should be employed sparingly and only in clear cases [e.g., where " minor pro-cedural deficiencies," not controversial questions, are involved]." Consolidated Edison Co. of New York (Indian Point Station, Unit No. 2), 7 AEC 947, 951 & n.8, 952 (1974). See Commonwealth Edison Co. (Byron Nuclear Power Station, Units 1 and 2), 19 NRC 1163, 1175 n.45 (1984).

1 procedures for quality assurance personnel could not be delegated to the ~ Staff because the efficacy and outcome of the remedial programs j undertaken by the utility (which had not been completed by the close l

, of the record) were central to a finding of reasonable assurance of proper facility construction.22/ Here, we are dealing with critical I safety procedure issues which are contested and are analogous to the procedures at issue in Byron. Because LILCO relies on its procedures and training to ensure that operators will not load the EDGs above 3300kW, we reject LILCO's assertion that the resolution of the issues of the adequacy of those procedures and training is not central to a finding of reasonable assurance.23/

~~22/ In Byron, as here, the utility undertook an analysis to

-address significant concerns raised by the evidence, but the results of that analysis.could not be explored before the record had closed and the evidentiary record did not demonstrate the requisite assurance. Id. at 1175.

23/

~~

LILCO asserts, without any explanation, that the resolu-tion of these issues would not be aided by further pro-ceedings. LILCO's Proposed Finding L-82 at 36 n.7. To the contrary, further proceedings would likely aid the resolution of these issues because they are quite complex, involve large elements of judgment and expertise, and con-cern each of the procedures and their inter-relationship with each other. Such issues are precisely the types of issues for which cross-examination is required "for a full and true disclosure of the facts." Southern California Edison Co., (San Onofre Nuclear Generating Station, Units 2 and 3), LBP-82-46, 15 NRC 1163, 1217 (1982).

Delegating the resolution of these issues to the Staff would deny the opportunity for an on-the-record hearing in violation of Section 189a of the Atomic Energy Act.

Southern California Edison Co., supra, 15 NRC at 1217. At the very least, this Board must await completion of the (Footnote cont'd next page)

! l l

B. The EDG Qualification Test Run was ,

Inadequate to Justify a Qualified I Load of 3300kW. l l

25. The; endurance testing of EDG'103 for 525 hours0.00608 days <br />0.146 hours <br />8.680556e-4 weeks <br />1.997625e-4 months <br /> was not i performed at a true load.value of 3300kW, as discussed in our findings on the adequacy of the replacement crankshafts at 3300kW. To estab-l~ lish an appropriate qualified load consistent with industry practice i

and with our views stated above, the endurance run should have been

conducted at a higher load level which encompassed not only the MESLs but also.an-adequate margin to account for cyclic and intermittent loads, the - operator error load, EDG load meter instrument accuracy, and uncertainties

in the measurements and nameplate ratings used by LILCO in estimating the MESLs. Tr. 27,542-43, 27,575-76 (Bridenbaugh);.Tr. 27,520-21, 27,540-41, 27,577-81, 27,584-85 (Minor);

_Bridenbaugh and Minor, ff. Tr. 27,500, at 8-9, 12-15, 19-22, 29.

An ondurance test which accounted for all of these loads would have pro-vided reasonable assurance that the EDGs have sufficient capacity and (Footnote. cont'd from previous page)

Staff's review of the job task analysis to determine whether, or to what extent, the parties must be accorded

an opportunity to address the results of that review.

Metropolitan Edison Co., (Three Mile Island Nuclear Sta-tion, No.-1), ALAB-729, 17 NRC 814, 887-88 (1983). If the Staff review discloses the need for corrective action, the determination of the adequacy of that action is a matter for the Board. Id. at 887; Long Island Lighting Co.

(Shoreham Nuclear Power Station, Unit 1), LBP-83-57, 18 NRC 445, 520 (1983).

i L l l

~. _ _ _ _ _ - - . _ - . . _ _ . . ~

I r-capability to handle the maximum loads that may be imposed on the EDGs. See Tr. 27,539-41,-27,608-09 (Minor).

III. CONCLUSION

26. Based upon all of the foregoing, we conclude that LILCO has failed to establish by a preponderance of the evidence that the EDGs with the proposed qualified load of 3300kW provide sufficient capacity and capability to assure that the requirements of clauses (1) and (2) of the first paragraph of GDC 17 will be met. We hold, therefore, that the EDGs do not comply with the requirements of GDC 17.

4 J

d h

I. DISCUSSION -

_l.

The County and State have stipulated that they do not chal-1

. lenge the adequacy of the replacemer.t crankshafts for operation at

~

- loads up to'3300..kW.

Tr. 28,417-18-(Dynner). The purely analytical evidence introduced by the parties does not demonstrate the adequacy

_ of the crankshafts above 3300kW.1/

i 4

2. The.only other evidence in the record on reopening that re-lates-to'the adequacy of the crankshafts at or above 3300kW is the 745 hour0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> endurance-run on EDG 103. For the reasons discussed below, we .;

~

find that the endurance testing, when viewed conservatively, p establishes the adequacy.of the replacement crankshafts for continuous ^

- operation at' loads no higher than 3230kW.2/

1/ .The replacement crankshafts do not comply with the DEMA  ;

recommendations for continuous operation at loads at or

, Labove 3300'kW. Tr. 23,377-79, 23,382 (Sarsten);.Tr.

22,809 (Pischinger). Nor do calculations' performed under.

the-Kritzer-Stahl criteria. demonstrate the adequacy of the crankshafts-at'3300kW~or above. The calculated safety  ;

factor for the replacement crankshafts at 3300 kW is 1.074

under the Kritzer-Stahl. criteria (Pischinger, ff. Tr.

L 28,416, at 3) which is less than the lowest acceptable. '

- value~in_ contemporary industrial practice for the design >

of crankshafts for medium speed diesel engines tdue size of '

l 'the EDGs. Tr. 23,012, 23,071-72 (Pischinger). Por the -

-same reasons discussed in our November.15, 1984, findings i

(at 41-44)-on the inadequacy of the crankshafts'at 3500 kW-and 3900 kW, we do'not rely on Dr. Pischinger's additional safety factor calculation of 1.31 for~ operation at 3300 kW. Pischinger, ff. Tr.-28,416,'at 3, 5; Tr. 28,482~

'(Pischinger). None'of the parties inntroduced any evidence showing whether the replacement crankshafts complied with the IACS, Lloyd's or~ ABS rules for operation between 3300 kW-and'3500 kW.

2 / -. (Although the endurance test demonstrates the adequacy of the crankshafts at loads no higher than 3230kW, the ade-quacy of the crankshafts at or below 3300kW is not in Lissue in this procceding.

. . . . _ - . _ . _ _ . _ , _ . _ - . _ . _ - ~ - . _ _ _ , . . _ _ _ _ . _ _ . . _ _ _ . . _ . _ _ . - - - _ , _

3. The load during the 525 hour0.00608 days <br />0.146 hours <br />8.680556e-4 weeks <br />1.997625e-4 months <br /> segment of the 745 hour0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> endur-ance run on EDG 103 conservatively should be regarded as having been run at no more than 3230kW. We base this finding on the facts that (i) operators were instructed to maintain the EDG load at 3300kW-

+100kW during the run, (ii) readings of EDG lO3's output that were ,

taken every 30 minutes showed that 424 of the hours were recorded at loads of 3300kW, 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br /> were recorded between 3250kW and 3300kW, and 81 hours9.375e-4 days <br />0.0225 hours <br />1.339286e-4 weeks <br />3.08205e-5 months <br /> were recorded between 3300kW and 3400kW, and (iii) the accu-racy of the load meter instrument loop used to record most of the kW readings was shown by calibrations (performed just prior to and fol-lowing the endurance run) to be off by 70kW at between 3000kW and 4000kW indicated.3/ Youngling, ff. Tr. 27,153, at 28-29; Dawe, et al., ff. Tr. 27,153, at 38-39; Bridenbaugh and Minor, ff. Tr. 27,500, at 22, 30-31; Tr. 27,312-13 (Youngling); Bush and Henriksen, ff. Tr.

28,503, at 3, 12.4/

3/ An unspecified number of the kW readings during this seg-ment were made from a secondary measurement loop with an accuracy of .6 percent. Tr. 27,312 (Youngling).

-4/ We find LILCO's refusal to conduct the last 525 hours0.00608 days <br />0.146 hours <br />8.680556e-4 weeks <br />1.997625e-4 months <br /> of endurance testing at 3500kW particularly disturbing. This refusal demonstrates LILCO's unwillingness to verify its theories by subjecting the crankshafts to actual testing.

LILCO's lack of confidence in its analyses at 3500kW is further demonstrated by the fact that LILCO attempted to avoid any endurance testing whatsoever. LILCO unsuccessfully attempted to persuade the Staff that, be-cause the maximum projected continuous load on the EDGs after one hour of a LOOP /LOCA was approximately 2600kW, the EDGs should fall within the 185 psig BMEP exception to the Staff's interim licensing basis. Tr. 27,348-49 (Youngling). If LILCO genuinely had confidence in its (Footnote cont'd next page)

_ \

4. We disagree with LILCO's assertion that, despite the rela-t tive accuracy of the load meter instrument loop, it is reasonable to assume that the vast majority of the kW readings were actually taken at an accurate mean value of 3300kW due to the constant pulsing of the kilowatt _ meter. Tr. 27,309-10 (Dawe). LILCO claims that because EDG 103 was connected to the grid during the endurance run, there was a constant, rapid pulsing up and down on the load meter of approximately 60 to 100kW (Tr. 27,310-11, 27,321, 27,421 (Youngling)) and that, as a result, the meter was as much in the upper part of the 60 to 100kW band around 3300kW as it was in the lower part. Tr. 27,307 (Youngling). The difficulty we have in accepting this claim is that, because the kilowatt meter is rapidly bouncing up and down, the (Footnote cont'd from previous page) analyses, it should have run the test at 3500kW.

LILCO's explanation for running the last 525 hours0.00608 days <br />0.146 hours <br />8.680556e-4 weeks <br />1.997625e-4 months <br /> of the endurance run at a nominal load of 3300kW makes no sense to us. The Staff had recommended that LILCO conduct the test at 3500kW.- Tr. 28,005 (Berlinger). LILCO purported-ly chose to test EDG 103 at 3300kW in order to qualify the EDGs at a load level that was as close as possible to the Staff's 185 psig BMEP criterion (approximately 2877 kW),

allegedly due to the Staff's concerns about operation above that level. Tr. 27,234-35, 27,346-47 (Youngling).

This rationale makes no sense because, according to the pr'emise of the Staff's interim licensing basis, had the endurance run been successfully completed at 3500kW (or at 3900 kW for that matter), the Staff would have been satisfied that the crankshafts had adequate design margin at that load regardless of whether that load level was close to or greatly exceeded the BMEP criterion. See Tr.

27,746 (Berlinger).

l

operator must interpolate the mid-point of the up and down pulsation

[

to obtain the kW readings. Tr. 27,321 (Youngling). That, when con-sidered with the fact that the kilowatt meter cannot be accurately read to within better than 50kW (Tr. 28,100 (Eckenrode)), leaves us with little assurance that the indicated kW readings are conservative.

5. The load during the 220 hour0.00255 days <br />0.0611 hours <br />3.637566e-4 weeks <br />8.371e-5 months <br /> segment of the 745 hour0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> endur-ance run at a nominal load at or above 3500kW5 / conservatively should be regarded as having been no higher than 3430kW to 3460kW. We base this finding on the fact that the recorded load levels for this seg-ment were measured with instruments having accuracies to within +70kW and +40kW. Tr. 27,315, 27,422-23, 28,801 (Youngling).6/

6. Thus, when instrument accuracy is conservatively taken into consideration, the 745 hour0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> endurance test on EDG 103 was run for 220 hours0.00255 days <br />0.0611 hours <br />3.637566e-4 weeks <br />8.371e-5 months <br /> at 3430kW to 3460kW and 525 hours0.00608 days <br />0.146 hours <br />8.680556e-4 weeks <br />1.997625e-4 months <br /> at 3230kW. Because LILCO 5/ During this segment of the endurance run, EDG 103 operated for approximately 119 hours0.00138 days <br />0.0331 hours <br />1.967593e-4 weeks <br />4.52795e-5 months <br /> at a nominal load of 3500kW, 101 hours0.00117 days <br />0.0281 hours <br />1.669974e-4 weeks <br />3.84305e-5 months <br /> at a nominal load between 3500 and 3900kW, and 7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> at a nominal load of 3900kW. Bush, ff. Tr. 28,503, at 16. Because there is no evidence in the record of the actual nominal loads recorded between 3500kW and 3900kW (see, e.g., Tr. 28,608-09, 28,666 (Bush)), we must conser-vatively consider these loads to have been recorded at loads not significantly greater than a nominal load of 3500kW. Id.

6/ According to LILCO, a majority of the loads recorded dur-ing the 220 hours0.00255 days <br />0.0611 hours <br />3.637566e-4 weeks <br />8.371e-5 months <br /> were recorded with the more accurate in-strument, but LILCO could not specify the exact number of hours. Tr. 27,423 (Youngling).

chose to conduct the second segment of the test at the lower load

. level, we are left with the difficult task of determining the specific

. load level at which the adequacy of the crankshafts can be established ifrom the endurance run.7/ The parties performed a number of calcula-

.tions-to determine a specific load level at which the adequacy of the crankshafts can be established from the endurance run. None of these calculations provides reasonable assurance of the adequacy of the crankshafts above 3230kW.

7. Dr. Bush concluded from the loads at which EDG 103 operated

.during the 745 hour0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> endurance test that the replacement crankshaf ts have a fatigue limit of 3430kW. Bush, ff. Tr. 28,503, at 16; Tr.

28,'531 (Bush).8/ Central to Dr. Bush's conclusion is his assumption, based on the data contained in Table 1 to his testimony, that the 7,/ We. note that the average load during the 745 hours0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> was ap-proximately 3300kW. This calculation is based on the fact that approximately-70 percent of the 745 hour-endurance test was run at 3230kW and 30 percent was.run.as high as 3460kW.- There is no evidence in the record,. however, from which to conclude whether such averaging is a scientif-1cally reliable method of determining the specific load

' level at which the adequacy.of the crankshaft can-be established.

8/ Dr. Bush conservatively took into consideration instrument accuracy of +70kW. Bush, ff. Tr. 28,503, at 14. LILCO erroneously claims that during cross-examination Dr. Bush revised this estimate to 3480kW based on an instrument ac-curacy of .06 percent. See LILCO's Additional Proposed Findings of Fact.Concerning Crankshafts, at 4. The testi-mony cited by LILCO does not support this proposition. In fact, Dr. Bush did not testify that instrument accuracy was in fact .06 percent. See also n.3, supra.

I

beginning of the fatigue limit for the replacement crankshafts is close to 10E6 cycles. Bush, ff. Tr. 28,503, at 16-18; Tr. 28,534, 28,649 (Bush).9/ .Because that data is not sufficiently probative or applicable to the replacement crankshafts, Dr. Bush's conclusion has no reliable basis and cannot be relied upon in determining the endur-ance limit of the replacement crankshafts.

8. The data relied on by Dr. Bush does not include fatigue limit data for the same type of steel from which the replacement crankshafts were fabricated. Tr. 28,650 (Bush). Given the consider-able variation that often occurs between samples of steels of the same specification and the absence of any data on the particular typelof steel from which the replacement crankshafts were fabricated,10/ we can draw no reliable conclusions about the onset of the fatigue limit of the replacement crankshafts from this data.ll/ Furthermore, Dr.

9/ This data concerning several ferrite steels purportedly shows that, regardless of the surface or heat treatment or composition of the steel, the beginning of the fatigue limit for ferrite steels is close to one times ten to the sixth cycles. Tr. 28,649 (Bush). The data itself indi-cates, however, that a considerable variation exists in the onset of the fatigue limit for the steels cited, ranging from 0.2 to 3.0 times 10E6 cycles. Tr. 28,747 (Bush). This in itself casts some doubt on the reliability of Dr. Bush's conclusion.

10/ Dr. Bush admitted that the fatigue strengths of samples of even the same material specifications often vary considerably. ~ Tr. 28,740-41 (Bush). Indeed, of the two items in the Table which Dr. Bush balieved to have mechan-ical properties similar to those of the replacement crank-shafts, the onset of the fatigue limit for one was over twice as long as the other. Tr. 28,652-53 (Bush).

11/ The data complied by Dr. Bush was not the result of an ex-tensive survey of the scientific literature but was (Footnote cont'd next page)

Bush's conclusion fails to account for the size effect on the fatigue limit of the replacement crankshafts.12/ Therefore, because Dr.' Bush failed to consider these and other factors,13/ we attribute no weight to his conclusion that the endurance limit of the replacement crank-shafts is 3430kW.14/

(Footnote cont'd from previous page) compiled in approximately 30 minutes from the witness's personal library. Tr. 28,742 (Bush). In fact, Dr. Bush indicated that, although he had known where he could have obtained fatigue limit data on the replacement crank-shafts' steel, he failed to look for the data. Id.

--12/ The data relied on by Dr. Bush does not include fatigue limit data on crankshafts the same size as the replacement crankshafts. Despite the fact that there is a significant relationship between the size of a crankshaf t and its fa-tigue limit, most of the data was not obtained from crank-shafts but from small test specimens or samples of materi-als. Tr. 28,640-42, 28,739-40-(Bush). Only 7 of the 21 items in Table 1 pertain to crankshafts, but that data was derived from automobile and aircraft engine crankshafts which generally are much smaller than the replacement crankshafts. See Tr. 28,640-42 (Bush).

1]/ Even though the geometry, torsional stresses and stress concentrations are important factors relevant to calculat-ing the fatigue life of a crankshaft, Dr. Bush failed to consider these factors. Tr. 28,655-56, 28,740 (Bush). In addition, Dr. Bush's other testimony contradicts his as-sertion that the type of surface treatment has no effect on the onset of the fatigue limit. Bush, ff. Tr. 28,503, at 17; Tr. 28,649 (Bush). Dr. Bush testified that shotpeening a crankshaft will delay the onset of its fail-ure if it operates above its endurance limit. Tr. 23,191, 28,642 (Bush).

14/

~~

As Dr. Bush testified, it would be extremely important to his calculation if the onset of the fatigue limit of the replacement crankshafts was 10E7 cycles rather than 10E6 cycles. Tr. 28,648-49 (Bush). Indeed, if the onset of the fatigue limit was at 4 or 5 times 10E6 cycles instead, (Footnote cont'd next page)

9. Drs. Bush and Pischinger also performed cumulative damage calculations of the fatigue limit of the replacement crankshafts from the loads at which EDG 103 was operated during the 745 hour0.00862 days <br />0.207 hours <br />0.00123 weeks <br />2.834725e-4 months <br /> endurance run. Bush, ff. Tr. 28,503, at 12, 17; Tr. 28,420 (Pischinger).15/ we give no weight to these calculations because they did not adequately

l. consider the actual sequence of loads on EDG 10316/ and therefore can-not form the basis for a reliable determination of the fatigue limit of the crankshafts. Dr. Bush's calculation considered seven various, but admittedly arbitrary, load sequences (Tr. 28,658-60, 28,611 (Bush)),17/ and Dr. Pischinger's calculation failed to consider the (Footnote cont'd from previous page)

Dr. Bush could not conclude that the replacement crank-shafts would survive operation at continuous loads up to 3500kW. Tr. 28,712-13 (B,ush).

--15/ Dr. Bush did not base his conclusion concerning the fa-tigue limit of the crankshafts on his cumulative damage calculation. Instead, Dr. Bush used his calcualation solely to confirm his conclusion, based on the actual ex-perimental evidence of the endurance run, regarding the fatigue limit of the crankshafts. Tr. 28,634, 28,717 (Bush).

16/ The cumulative rate of damage to a crankshaft at a given stress level is a function of prior cyclic stress history, and the particular sequence in which various loads are ap-plied to a crankshaft has a significant effect on its fa-tigue limit. Tr. 28,513-14 (Bush); Tr. 28,421 (Pischinger). Indeed, the predicted fatigue limit of a crankshaft would " vary markedly" depending upon the actual sequence in which loads have been applied to it. Bush, ff. 28,503, at 16.

17/ The fact that all seven of Dr. Bush's postulated load se-i quences gave the same result (Tr. 28,609-10, 28,660 (Bush)), does not necessarily suggest, as Dr. Bush (Footnote cont'd next page)

( .. ..

load sequences whatsoever. Tr. 28,420 (Pischinger).18/ Furthermore,

~

Dr. Pischinger's calculation failed to take into consideration possi-ble instrument error of up to 70 kW. Tr. 28,427 (Pischinger).

10. In summary, because the analyses and calculations performed to determine the load level at which the EDG were " qualified" lack reliability and probative value, we conclude that the endurance run establishes that the EDGs are qualified only for continuous operation at 3230kW. That is the only load level at which there is reasonable-assurance that EDG 103 operated for lOE7 cycles.

(Footnote cont'd from previous page) implies, that the actual load sequence had no effect on

=the. fatigue limit of the replacement crankshafts. It is just.as.likely that, given the expected significant effect of the load sequence on the fatigue limit, Dr. Bush did not vary the postulated sequences sufficiently to reflect the actual load sequence and its effect on the fatigue limit.

18/ We have no basis to credit Dr. Pischinger's assertion.that his-calculational method compensates for the effect of load sequencing because it considered stress. cycles below his estimate of the fatigue limit (i.e., stresses corre-sponding to loads between 3300 kW and approximately 3500 kW). Tr. 28,421 (Pischinger). We note that Dr.. Bush's calculational-method also considered the stress cycles below the estimated fatigue limit but nonetheless failed ,

to adequately take into account the loading sequence. Nor do.we attribute any weight to Dr. Pischinger's suggestion that the actual sequencing of loads on EDG 103 (in which higher loads -- 3500 kW to 3900 kW - .were applied first) may have been beneficial (Tr. 28,446-48 (Pischinger)) be-cause, as Dr. Bush testified, the application of those stresses should not have a pronounced effect on the fa-tigue limit. Tr. 28,612, 28,614, 28,662, 28,725-26 (Bush).

i

11. LILCO and the Staff also presented evidence on the ability of the replacement crankshafts to handle short duration loads up to 3900kW. We agree with the Staff that the endurance run does not pro-vide an adequate basis for drawing conclusions about the ability of the crankshafts to handle loads greater than 3430kW to 3460kW on the EDGs. Bush, ff. Tr. 28,503, at 5; Tr. 27,989 (Berlinger). We do not agree, however, that the adequacy of the crankshafts to handle loads up to 3900kW for short periods has been established with any degree of reasonable assurance.

12. Dr. Bush concluded that the crankshafts can survive loads of 3900kW for up to an hour because EDG 103 already has operated for ap-proximately 7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> at 3900kW and because he believes that, if a crack were to initiate from a short duration load of 3900 kW, it would not propagate to failure during additional operation at lower loads during an accident. Bush, ff. Tr. 28,503, at 21-22; Tr. 28,663-64 (Bush).

We find that these factors do not constitute an adequate basis from which to conclude whether the crankshafts can handle loads up to 3900kW for an hour.

13. First, the fact that the crankshaft on EDG 103 has survived 7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> of operation at 3900kW, and thus has operated for more hours than would be required under industry practice to establish its short-time capability at that load (Knox, ff. Tr. 27,735, at 4), is not a reasonable basis for concluding that the crankshafts are adequate for

<n short-time operation at-that load. Additional hours of operation are

- required to provide reasonable assurance that the EDGs, with their troubled history of reliability, have the capacity to operate at that load. We note that the original crankshafts had successfully complet-ed their short-time load capability qualification testing at the fac-tory at loads over 3900 kW before failing during later testing. Tr.

28,018-19,28-026 (Knox).

14. Second, Dr. Bush could not quantify the probability of crack initiation at 3900kW and did not calculate the rate of crack propaga-tion at 3900kW. Dr. Bush made it very clear that quantifying the probability of crack initiation, propagation and crankshaft failure would require a very detailed analysis which he had not performed.

Id.; Tr. 28,668-70-(Bush). Such calculations could have provided a reliable basis for determining the capability of the crankshafts at 3900kW. Bush, ff.-Tr. 28,503, at 22. Because Dr. Bush's crack propa-gation' analysis is essentially rx) more than a gut feeling unsupported by_any scientific. analysis, we cannot find that reasonable assurance of the adequacy of the crankshafts has been demonstrated.19/

4

15. In any event, the adequacy of.the crankshafts has not been demonstrated because, as previously noted, there is no reasonable 19/ We do not rely on Dr. Pischinger's calculations on the ca-pability of the crankshafts to withstand loads up to 3900kW for short periods for the same reasons stated in our. November 15, 1984,. Findings of Fact and Conclusions of Law, at 39-44.

11 -

l l assurance that procedures will ensure that operators will reduce the load from 3900kW to below 3300kW within one hour. Thus, even assuming that the crankshaft can survive loads of 3900kW for up to one hour, there is insufficient evidence in the record to conclude that the crankshafts can survive such loads for longer than one hour.

II. CONCLUSION Based upon all of the foregoing, we conclude that LILCO has failed to establish by a preponderance of the evidence that the re-placement crankshafts are adequate for operation at loads above 3300 kW.

12 -

1 l

l

. UNITED STATES OF AMERICA  : .' r 70 NUCLEAR REGULATORY COMMISSION '

Before the Atomic Safety and Licensing Board'qg g g gy .33

) W .CE . .:. ;;,ii r In the Matter of ) j d,M,(,jl"V #-

)

LONG ISLAND LIGHTING COMPANY ) Docket No. 50-322-OL

)

(Shoreham Nuclear Power Station, )

Unit 1) )

)

CERTIFICATE OF SERVICE I hereby certify that copies of Suffolk County and State of New York Proposed Findings of Fact and Conclusions of Law on Emergency Diesel Generator Contentions have been served on the following this 15th day of April 1985, by U.S. mail, first class, except as otherwise noted.

Lawrence J. Brenner, Esq.* Edwin J. Reis, Esq.*

Administrative Judge Bernard M. Bordenick, Esq.

Atomic Safety and Licensing Board U.S. Nuclear Regulatory Comm.

U.S. Nuclear Regulatory Commission Washington, D.C. 20555 .

Washington, D.C. 20555 Mr. Stuart Diamond Dr. George A. Ferguson* Business / Financial -

Administrative Judge NEW YORK TIMES- ~

Atomic Safety and Licensing Board New York, New York 10036 School of Engineering Howard University E. Milton Farley, III, Esq.*

2300 6th Street, N.W. Hunton & Williams Washington, D.C. 20059 P.O. Box 19230 2000 Pennsylvania Ave., N.W.

Dr. Peter A. Morris

• Washington, D.C. 20036 Administrative Judge Atomic Safety and Licensing Board Mr. Jay Dunkleberger U.S. Nuclear Regulatory Commission New York State Energy Office Washington, D.C. 20555 Agency Building 2 Empire State Plaza Edward M. Barrett, Esq. Albany, New York 12223 General Counsel Long Island Lighting Company James B. Dougherty, Esq.

250 Old Country Road 3045 Porter Street, N.W.

Mineola, New York 11501 Washington, D.C. 20008 I

1 5 -

Robert E. Smith, Esq. Stephen B. Latham, Esq.

Rosenman Colin Freund Lewis & Cohen Twomey, Latham & Shea 575 Madison Avenue P.O. Box 398 New York, New York 10022 33 West Second Street Riverhead, New York 11901 Mr. Brian R. McCaffrey Long 1sland Lighting Company Odes L. Stroupe, Jr., Esq.k Shoreham Nuclear Power Station Hunton & Williams P.O. Box 618 P.O. Box 109 North Country Road 333 Fayetteville Street Wading River, New York 11792 Raleigh, North Carolina 27602 Joel Blau, Esq. MHB Technical Associates New York Public Service Commission 1723 Hamilton Avenue The Governor Nelson A. Rockefeller Suite K Building San Jose, California 95125 Empire State Plaza Albany, New York 12223 Hon. Peter F. Cohalan Suffolk County Executive Martin Bradley Ashare, Esq. H. Lee Dennison Building Suffolk County Attorney Veterans Memorial Highway H. Lee Dennison Building Hauppauge, New York 11788 Veterans Memorial Highway Hauppauge, New York 11788 Fabian Palomino, Esq.#

Special' Counsel to the Atomic Safety and Licensing Board Governor Panel Executive Chamber U.S. Nuclear Regulatory Commission Room 229 Washington, D.C. 20555 State Capitol Albany, New York 12224.

Docketing and Service Section Office of the Secretary Atomic Safety and Licensing U.S. Nuclear Regulatory Commission Appeal Board 1717 H Street, N.W. U.S. Nuclear Regulatory Washington, D.C. 20555 Commission Washington, D.C. 20555 Jonathan D. Feinberg, Esq. Stewart M. Glass, Esq.

Staff Counsel Regional Counsel New York State Public Federal Emergency Management Service Commission Agency 3 Rockefeller Plaza 26 Federal Plaza Albany, New York 12223 New York, New York 10278 J

Mr. John C. Gallagher Timothy S. Ellis, Esquire #

Chief Deputy County Executive Hunton & Williams County of Suffolk 707 East Main Street H. Lee Dennison Executive Richmond, VA 23212 Office Building Veterans Memorial Highway Hauppauge, New York 11788 Dougla J. Scyefdt KIRKPA ICK & LLOCKHART 1900 M Street, N.W., Suite 800 Washington, D.C. 20036

• By Hand Delivery i By Federal Express DATE: April 15, 1985

.. . .. . _ _