Text

Proposed Findings of Fact Re Emergency Diesel Generator Contentions.Certificate of Svc Encl
	ML20112J882
Person / Time
Site:	Shoreham File:Long Island Lighting Company icon.png
Issue date:	12/03/1984
From:	Dreifus D; HUNTON & WILLIAMS, LONG ISLAND LIGHTING CO.
To:	;
References
	CON-#285-439 OL, NUDOCS 8504090241
	Download: ML20112J882 (164)
	v • d • e

-q L./}) -?

<gp UNITED STATES OF AMERICA '65 / ; .. o '

NUCLEAR REGULATORY COMMISSION M2 l dhh[b;$-[3{%IA-y Before the Atomic Safety and Licensing Board - M:C In the Matter of )

)

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

)

(Shc,reham Nuclear Power Station, )

Unit 1) )

LONG ISLAND LIGHTING COMPANY'S '

PROPOSED FINDINGS OF FACT CONCERNING EMERGENCY DIESEL GENERATOR CONTENTIONS Hunton r, Williams 707 East Main Street P. O. Box 1535 Richmond, Virginia 23212 2000 Pennsylvania Avenue, N.W.

P. O. Box 19230 Washington, D.C. 20036 333 Fayetteville Street P. O. Box 109 Raleigh, North Carolina 27602 8504090241 841203 PDR ADOCK 05000322 Q '

G PDR D L

c '

Q

+ .~

2 2 .

TABEI 0F CONTENTS Tab , ;. Page j

+

h*y

-1 PROPdSED FINDINGS OP: FACT CONCERNING QUALIFIED LOAD ,

I ~.

The Qualified Load is an Acceptable

. . Licensing Basis.............................................. 1 A. Alternative Methods Can Be Used to Comply With Regulations.................................. 1

' ~

B. DefinitiCd of Qualified Load and ,

Maximum EUeigency. Service Load.. . . . . . . . . . . . . . . ........7.2 C. The Maximum Emergency Service Load e is-Conservative.... ..................................... 3 1 -

II. The Qualified Load Accommodates Intermittent

. or Cyci'ic Loads.............................................. 8 m

III. Potential Instrumeht Error Does Not

~

Invalidate Qualified Load or Endurance Run.................. 13

~

IV. . The Tolerance Band of + 100 KW is Acceptable for Purposes of Surveillance

._

Testing and the" Endurance Run..........................,......~17 V. _ Qualified Load Need Not Include Operator

, Error Load......'............................................ 19 VI. ' Conclusions................................................. 36

2. PROPOSEDFINDINdSOFFACT.CONCERNINGCYLINDERBLOCKS I. Operating: Experience of the Shoreham EDGs.................... 1

, II. Metall'urgical Analysis and Testing _.4 4 Demonstrate _That the'EDG 101, 102 and 103 Re' placement Blocks Have Superior Pr,opertiesi to - the10riginal EDG 103 Block.' . . . . . . . . . . . . . . . . . . . . 3 -

. , _ s %

' The'Bi6ck Stress Analyses Are. Conservative....................7

~

III.

IV. Ligament1 Cracks Are Benign..................................l.9

. y 1 .

V. Stud-to-Stud Cracks Are Not Likely to-

Initiate and,.if_They Initiate, Will Not .

oJ" ImpairEDGlOperationDuringanEmergency....................12 E'.

+

se~

, -e

.E' h -

.~ % %

\' ",_

k

)' - ; ,

p

y

= -

Tab Page n ,

VI. Circumferential Cracks Are Not Present and Will Not Impair EDG Operation if They Initiate..... .................. ..... ........ ............ 20 VII. The Replacement Block is Adequately Designed and Tested...... .......... ........................... . . . 25 VIII. Conclusions. ....................... .................... 27 3 ADDITIONAL PROPOSED FINDINGS OF FACT CONCERNING CRANKSHAFTS 7

- I. Introduction............... ........... ............... . . . .1 II. Reliability of Crankshafts at 3300 KW. . .... .......... .. 1 E

A. 10E7 Cycle Endurance Run................................. 1 y ;, s i

B.

1

=

Kritzer-Stahl............. ............................. 2 4%s

g 3 C. DEMA. .. ........... ................. ............. . . .2 R i~.l".

I- III. Reliability of Crankshafts for Loads ;d.[l.d Y.

Above 3300 KW......... ........ ........ . . ............ . .. 3 4 ; . ., z

_ g '-l-y A. Loads Between 3300 and 3500 KW. ................ . . . . . .. 3 -

'M9

~

m. 4./ J B. Loads Above 3500 KW.. .... .......................... . .. 5 - '

t

[ Np IV. Future Testing and Inspections. ......................... . .. 6 t M. 7.

, , r. <,.

Qd, %. G ,

V. Conclusions............ ........... ....... . ... . . . . . . . . .. 6 '

6

-0 6..

4

PROPOSED FINDINGS OF FACT CONCERNING CRANKSHAFTS, November 5,1984 (submitted previously)

(@i?[. .,

@;C ;f,.,

b; : . ;)

_ I. Background................ ........... ... .............. . .. 1 $.hig; .

II. The Crankshafts are Only Required to Comply With DEMA.

t The Crankshafts do Not Have to Comply With the Requirements Of Any Other Design Society...... ........l...... '

3 III. The Crankshafts Comply With DEMA........ ............... .. .7 IV. The Crankshafts Have Been Approved By ABS................... 17

=

L" 11) h c

r b

Tab Page V. Fatigue Analys is . . . . . . ........ ....... .. ...... .. .. .18 VI. Kritzer-Sta'hl...... . .. ... ........... ... . . ... . .31 VII. Shot Peening...... ...... .. .. .... ........... ... ...... 34 5 LILCO's REPLY TO SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT CONCERNING CRANKSHAFTS, December 3, 1984 (submitted previously) l

! I. Introduction............... ........... ....... .... ... ... 1 II. Discussion.................... ..... .................... ... 2 A. Compliance With the Rules of Classification Societies. . . . 2

1. Relevance of Classification Societies ' Rules . . . . .... 2

2. Compliance With the Rules of Classification Societies...... .. ... ............. ........ ..... 4 (a) Lloyd's......... ..... .......... . .. . ...... 4 (b) IACS. ... .................. .... .. ........ .9 (c) ABS Rules on Torsional Vibration............... 12 (d) ABS Web Dimensions.................. .. ....... 16 (c) The Crankshafts are Adequate Under the Kritzer-Stahl Criteria. ........... .......... .17 B. DEMA.. ................................... ........... 20

1. DEMA is an Appropriate Standard... .... ........... .20

2. The Crankshafts Comply With DEMA...... ............ 27 l

l C. FaAA's Fatigue Analysis............... .. .............. 33 D. Shot Peening. .. ... ............. .... ............. .. 38 i

l (iii)

1. The Qualified Load is an Acceptable Licensing Basis A. Alternative Methods Can Be Used to Comply With Regulations L-1. GDC 17 requires that an onsite electric power system be provided to permit functioning of structures, systems and components important to safety.

The onsite electric power system must have sufficient capacity and capability to assure specified safety functions in the event of anticipated operational occur-rences and postulated accidents. 10 CFR S 50, Appendix A. To comply with these requirements of GDC 17, the diesel generators must be able to satisfy the func-tional requirements of the plant. Tr. 28131-34 (Berlinger). Regulatory Guide 1.9, endorsing IEEE 387-1977, see Tr. 27452-53 (Dawe), provides an acceptable approach, through continuous and short-term ratings, for demonstrating compli-ance with these aspects of GDC 17. This is not the only acceptable approach; it is one option. Alternate approaches are permitted. Tr. 27758 (Knox); Tr.

27759-60, 27960-61 (Berlinger); see also Tr. 27193-99 (Dawe); Tr. 27758 (Knox).

A short-term rating is not specifically required by regulation. What must be demonstrated, to find compliance with the capacity and capability requirements of GDC 17, is that the diesel generators are qualified to support a total elec-tric load which permits functioning of the equipment needed to perform required safety functions. The required loads should be assessed as they are expected to occur in sequence. See Tr. 27742-44 (Knox, Berlinger); Tr. 27751-53, 27758-59, 27996 (Knox); Tr. 27787 (Berlinger). The qualified load concept provides such an alternative, acceptable to the Staff, for demonstrating compliance with GDC

17. E.g. , Tr. 27759-60, 27772-73, 27945-46, 28181-82 (Berlinger) . The quali-l fied load concept involves testing of the diesel generator for 10E7 loading cy-cles, which is far more testing than is normally required by the Staff to demon-strate GDC 17 compliance for emergency diesel generators. Tr. 27987, 28170 (Berlinger); see also Tr. 28009 (Berlinger).

L-2. The qualified load concept was introduced by the NRC Staff, in the Safety Evaluation Report on the Transamerica Delaval, Inc. Diesel Generator

fi Owners Group Program Plan, August 1984, as an interim licensing basis for TDI y diesel engines. This basis has been established for a number of plants by the g

i NRC Staff and its consultants pending completion of the Staff's review of the L Owners Group efforts. Dawe et al. , ff. Tr. 27153, at 9-11; Tr. 27759-60, h 27960-62, 27964-66, 27983-84, 27990, 28181-82 (Berlinger) . In establishing the I interim licensing basis for TDI diesel engines, the Staff indicated that engines I

w operating below a BMEP of 185 psig could be licensed. For engines operating at

[ above 185 psig, a qualified load would have to be established. Tr. 27965-66

?

l (Berlinger); Dawe et al., ff. Tr. 27153, at 10.

d B. Definitions of Qualified Load and Maximum Emergency Service Load I

[ L-3. The qualified load is that load (1) which bounds the maximum emer-

?

gency service load (MESL) for each of the three Shoreham TDI EDGs, and (2) for which it has been demonstrated that certain key engine components have been op-erated successfully for at least 10E7 loading cycles. For Shoreham, the quali-a fied load is 3300 KW. Dawe et al. , ff. Tr. 27153, at 10.

f! L-4. For Shoreham's TDI diesel generators, the testing to achieve 10E7

loading cycles was chiefly for the purpose of demonstrating the adequacy of the -

> replacement crankshafts. Dawe et al. , ff. Tr. 27153, at 10. -

L-5. The maximum emergency service load (MESL), as defined in Amendment f 52 to the Shoreham License Application (Revision 34 to the FSAR), is the maximum n

load which would exist on an EDG during a loss of coolant accident in conjunc-tion with a loss of offsite power (LOOP /LOCA). A LOOP /LOCA event results in the

{ maximum automatic demand on the diesel generators and is therefore the event f considered in establishing the MESL for each diesel generator. Dawe et al. , ff.

Tr. 27153, at 8, 29. The MESL is a conservative estimate of diesel generator

loads anticipated in the event of a LOOP /LOCA, ea, Tr. 27756 (Berlinger); Tr.

1

j. 28208 (Knox). The peak load would last for a short period of time, with loads 1

L dropping considerably after one hour. Tr. 27182-85 (Youngling, Dawe); Tr.

28140-41, 28182-85 (Berlinger) .

r

. . . . - - ^

C. The Maximum Emergency Service Load is Conservative L-6. The MESL is determined for each EDG by summing the individual loads which will be simultaneously connected to that EDG for more than a short period of time following initiation of a LOOP /LOCA event. A combination of nameplate and measured component load values is used. Dawe et al. , ff. Tr. 27153, at 9.

Coincident operation of the equipment at either full nameplate or measured KW load is conservatively assumed. Tr. 27462 (Dawe) .

L-7. Plant equipment that responds automatically to an accident is the same after establishment of the qualified load as it was before with the excep- .

tion that one of the two reactor building service water pumps powered from r

EDG-103 is not automatically started and the spent fuel pool cooling water pumps have been administratively removed from service for the first operating cycle.

These changes do not affect the ability to accomplish required safety functions, see Tr. 27192 (Youngling); Tr. 27230-31, 27325 (Dawe) . There is also no change in the accident analyses and the assumptions of the accident analyses for Shoreham with the establishment of a qualified load. Tr. 27523-24 (Minor); see also Tr. 28295-96, 28311-12, 28355-56 (Clifford) .

L-8. The calculated MESL is 3253.3 KW for EDG-101, 3208.7 KW for EDG-102 _

and 3225.5 KW for EDG-103. Dawe et al. , ff. Tr. 27153, at 9. Thus, the 3300 KW qualified load bounds the MESL for each diesel generator.

L-9. The County witnesses claimed that many safety loads have not been '

estimated any more accurately than they were at the PSAR stage. Bridenbaugh and Minor, ff. T r. 27500, at 14. LILCO, however, has individually measured the ac-tual KW load values for many pieces of installed equipment, in only one in-stance, an emergency switchgear room, relay room and control room air-conditioning unit, was the measured component load greater than the nameplate -

value; 36.4 KW measured versus 33.9 KW nameplate. This larger measured value was used in calculating the MESL. In all other cases, the measured load was

5 _-

lower than the nameplate value, in some instances by as much as 40 to 80 per-cent. These differences aie attributable to conservatism in the sizing of the equipment for design service. Tr. 27202-04 (Youngling, Dawe). SC's claim that many safety loads have not now been estimated more accurately than at the PSAR stage therefore has no basis.

L-10. Suffolk County witnesses believed that the use of measured loads in

{ determining the MESL values was inappropriate. Bridenbaugh and Minor, ff. Tr.

27500, at 8. To the contrary, Regulatory Guide 1.9 differentiates betwaen the f prediction of loads at the construction permit and operating license stages in terms of ability to estimate loads with more accuracy at the latter stage.

1 Bridenbaugh and Minor, ff. Tr. 27500, at 13-14. The ability to be more accurate

?

at the license stage includes the ability to measure loads, and use of such mea-surements is appropriate in establishing the qualified load. Tr. 27450-51

(Dawe); Tr. 27795-96 (Knox).

7 L-11. Not all of the measured loads which were below nameplate values were used in developing the MESL. Tr. 27204-05 (Dawe) . LILCO measured the in-dividual loads for equipment representing about 60% of the KW value of the MESL.

l r_ in actually calculating the MESL, however, measured values were used for only about 30 to 40 percent of the KW value of the MESL. Tr. 27207-08 (Youngling, i

Dawe) . The measured values which were not used in the calculation of the MESL f were less than, but generally close to, the nameplate values. It was conserva-7 tive to use the nameplate value. Tr. 27204-05 (Daweb 27209, 27212-13 E

(Youngling) . The unused measured values would reduce the calculated MESL for each diesel generator by approximately 30 KW if they were substituted for their corresponding nameplate values. Tr. 27213-14 (Youngling) . The equipment for f l

which individual load values were not established by measurement are relatively small in terms of KW rating with the exception of the service water pumps. Tr. l 27216 (Youngling); 27222 (Dawe). Thus, even if an individual component in this u

L

i L

v

[ '-

Ir

$. category did operate at greater than its nameplate load demand, there is reason-g

[ able assurance it would not have significant effect on the calculated MESL.

L-12. All individual equipment load measurements were taken with a I

Dranetz meter. This is a state-of-the-art instrument with an accuracy of ap-

[ proximately 1 percent. Measured values used in calculating the MESL were _

lit

% unaltered for this instrument error. Tr. 27214 (Youngling). This would not, h however, introduce significant error into the MESL calculation. Thirty-fou r

?

percent of the MESL for EDG-101,1109.4 KW out of 3253.3 KW, is contributed by measured loads. Bridenbaugh and Minor, ff. Tr. 27500, at 20; Tr. 27210 A

f C (You ngling) . If the measured load values were assumed to be 1% in error, and

[ therefore the MESL was increased by 11 KW, it would still be bounded by the f.

a qualified load. The same is true for the MESLs for EDG-102 and EDG-103. See Bridenbaugh and Minor, ff. Tr. 27500, at 20. -

e L-13. The calculation of the MESL is conservative because in summing t

i loads to obtain the MESL, the assumption that all equipment operates simulta-B neously at the nameplate or measured value does not take into consideration the 3

( E actual sequence of operation or the operating conditions seen by the equipment.

j Dawe et al. , ff. Tr. 27153, at 19. '

j L-14. Nameplate values provided by manufacturers are accurate to well 5 within + 5%. They are often measured values from tests performed by the manu-i

? .facturer at the design conditions of the component. Compare Tr. 27217 -

g F (Youngling, Dawe) with Tr. 27606-08 (Bridenbaugh, Minor), 27620 (Minor). The 1

lj= use of equipment nameplate ratings in calculating the MESL is generally conser-1, 3 vative, however, because the components themselves are conservatively sized dur-ing the design of the plant. Thus, equipment as installed may be oversized such -

4

,{ that its design function is performed at less than nameplate power demand, Tr. -

p 27201-02 (Dawe), or the equipment is not required to develop its full design

I output following a LOOP /LOCA, thus resulting in a power demand less than L -

=

k _

s

=; _

L =

[ .

nameplate. See Tr. 27648-49, 27657 (Dawe) . The fact that all in-plant equip-ment load m.easurements except one showed that actual loads were lower than name- [_

plate values demonstrates this conservatism. [

=

L-15. Many conservatisms exist in the nameplate ratings as used to calcu- .-

s late the MESL. For example, the reactor building closed loop cooling water pumps, each with an 80 KW nameplate contribution to its respective diesel gener-ator's MESL, are designed to service many components for cooling during normal ;

operation . Following a LOOP /LOCA, most of these cooled loads are isolated and pump flow is reduced to about one-third capacity. The actual power demand in this case is estimated to be about 40 KW per pump less than the nameplate rating 1 e

used in calculating the MESL. Tr. 27648-49 (Dawe) . Similarly, the MESLs for EDG-101 and EDG-102 each include 235 KW for one RBSVS chiller at nameplate de- ?

mand while the MESL for EDG-103 includes 470 KW for two chillers at full load. l Tr. 27643 (Dawe). These chillers are oversized for the LOCA condition because,

in addition to being redundant equipment, they were sized for the greater heat load from a pipe break outside containment. There will be insufficient heat ~

l load to cause the chillers to operate at full load. Tr. 27668-71 (Dawe) . Thus, 1 a large part of the 235 KW included for each chiller in the MESL calculations will not actually be realized as load on a diesel generator following a 3

LOOP /LOCA.

Tr. 27650 (Dawe) .

L-16. The nameplate load used for each core spray pump in the MESL calcu-lations is 998 KW. This is for a pump runout condition of 6900 GPM. At .

1 -

Shoreham, maximum runout for core spray is 6400 GPM, with a corresponding pump ,

load of 988 KW. This represents another 10 KW conservatism in the MESL. Tr.

27652, 27657-58 (Dawe). _

L-17. That the load va!ues assumed for core spray and RHR pumps in calcu- _

\ 2 lating the MESLs are for runout conditions introduces yet additional conserva-4 l tism in the MESL. A break would be required in the core spray or low pressure j'

+

r

coolant injection (RHR/LPCI) line for the associated pump to be at runout while injecting. Such a break is, by itself, unlikely, and multiple, simultaneous breaks as assumed by the MESLs are far beyond the design basis of the plant. If l a co're spray pump is not at runout, its contribution to the associated MESL is l

reduced by 100 to 135 KW depending on reactor vessel pressure. If an RHR pump is not at runout, a reduction of 30 to 90 KW is realized, depending on the com-bination of RHR pumps in operation. Tr. 27657-59 (Dawe); see also Tr. 27677 (Dawe).

L-18. Following a LOOP /LOCA, the diesel generator will not be required to power all equipment included in the MESL at maximum demand simultaneously.

Thus, the assumption of coincident demand in the MESL calculation is also con-sc. vative. By design, the diesel generator loads are connected in sequence.

Further, and more importantly, the development of an individual load is not al-ways immediate, but rather, depends on the dynamic response of the plant. Tr.

27461, 27642, 27648 (Dawe) .

L-19. For example, the MESL calculation conservatively assumes the RBSVS chillers operate immediately at full capacity. In fact, it takes a significant period of time for heat load to build up. Tr. 27643, 27666-67 (Dawe) . Thus, by the time the chiller loads are beginning to build up, the core spray and RHR pump loads will have been decreased as a result of the operator action to reduce injection rates. Tr. 27228-32 (Dawe, Notaro, Youngling). Another dynamic ef-fect is exhibited between the loop level pumps and the ECCS pumps. As the flow

from the ECCS pumps increases, the loop level pumps reduce to minimum flow. Tr.

l 27646-48 (Dawe). Thus, the maximum demand of the ECCS pumps and the loop level pumps will not be superimposed. The operation of the battery chargers provides yet another exampla of this type of load savings. Tr. 27644-45 (Dawe) .

L

~

i -

p ll. The Qualified Load Accommodates Intermittent or C_y_clic Loads L

The MESL calculations do not include intermittent or cyclic loads.

L-20..

l h 1--

These are small loads that will operate only once or occasionally, for a short a period of time, following a LOOP /LOCA event. These loads do not impose a con- [

tinuous load on the diesel engines. Dawe et al. , ff. Tr. 27153, at 11; see also 1

L F Tr. 28194 (Knox). _

g

/ L-21. In considering whether an engine meets the 185 psig criterion, see -

$ Finding L-2, the NRC Staff stated they would consider excepting engines from the

? requirements where the load exceeds the 185 psig BMEP criterion for brief peri- [

[ ods of time. Dawe et al. , ff. Tr. 27153, at 10-11. By extension, it would be g i

b acceptable for the qualified load to be exceeded by small amounts for short pe-

, Y 1 riods of time, although that is not expected to occur. Thus, intermittent or T

't g cyclic loads can be excluded from the MESL used to establish the cualified load.

Dawe et al. , ff. Tr. 27153, at 11-12; Tr. 27742-43 (Berlinger) . But see Tr.

28003-06 (Berlinger); 27758-59 (Knox) .

f L-22. Three intermittent load groups (automatically actuated motor oper- .a-ated valves, diesel generator fuel oil transfer pumps and diesel generator air y compressors) were identified by LILCO and excluded in establishing the maximum [

emergency service load, and thus the qualified load, for each EDG. This exclu- h sion was appropriate and justified because these are short-ter:n, small magnitude loads. Dawe et al. , ff. Tr. 27153, at 7,12. The NRC Staff agreed with LILCO's %

4 identification of intermittent loads. Knox, ff. Tr. 27735, at 5; Tr. 27764-65, ;

5 27794 (Knox). Suffolk County witnesses believed an additional group of valves, E identified as nonoperating MOVs, should be included in this category. T'- FSAR r

states, however, that these valves do not operate upon a LOCA. Bridenbaugh and -

+

Minor, ff. Tr. 27500, at 9-11; see also Tr. 27525-27 (Bridenbaugh). The County s

?

w-j witnesses could not specifically identify any valve, or its load contribution, I :

f from this category that would in fact operate following a LOOP /LOCA. Tr. "

1

_~

.g.

27555-56 (Bridenbaugh). These valves are operated manually by the operator in the control room and none are required for plant response in the short-term foi-i lowing a LOOP /LOCA. Tr. 27305-07 (Dawe, Youngling). Thus, the set of intermit-

]

tent and cyclic loads has been properly identified and described by LILCO.

L-23. Automatically actuated motor operated valves are valves which both receive power from an EDG and have the ability to operate automatically follow-ing a LOOP /LOCA. Not all of these valves would be expected to reposition, and thus represent load on an EDG, following a LOOP /LOCA. Those that do operate generally do so only once. Operation occurs in the first several minutes after the diesel generators start, with most of these valves operating for less than one minute, and none operating for longer than three minutes. For various rea-sons, not all valves that do operate will do so simultaneously. Dawe et al. ,

ff. Tr. 27153, at 12-14; Knox, ff. Tr. 27735, at 5-6; Tr. 28195 (Knox).

L-24. Simple summation of each individual valve load yields a total valve load exceeding any that could reasonably be expected. Conservatively summed in this manner, the total valve load still could not exceed 65.7 KW for EDG-101, 64.3 KW for EDG-102 or 46.7 KW for EDG-103. Addition of these conservative loads to the MESL of each EDG yields loads of 3319.0 KW for EDG-101, 3273.0 KW for EDG-102 and 3272.2 KW for EDG-103. Thus, even assuming coincident valve op-eration, the 3300 KW qualified load would be conservatively predicted to be ex-ceeded for only one EDG, and then by only 19 KW, and for less than three minutes. Dawe et al. , ff. Tr. 27153, at 13-16.

L-25. Each EDG has two fuel oil transfer pumps which operate automat-ically to refill the fuel oil day tank, but only after a predetermined level setpoint is reached following diesel generator operation. Only one pump per diesel will operate at any time, the second operating only if the first fails.

The transfer pump will operate for approximately 22 minutes every 48 minutes during operation of its associated EDG at 3300 KW, and represents a load of only

0.2 KW. Dawe et al. , ff. Tr. 27153, at 16. Thus, this load does not operate in the first few minutes after the start of an accident when peak diesel generator load would be experienced, Dawe et al., ff. Tr. 27153, at 16-17, but if it were assumed to do so, 0.2 KW added to the MESL for each diesel would still result in loads below the 3300 KW qualified load.

L-26. Each diesel generator has two associated air compressors which start automatically to recharge the air start receivers after the EDG has ener-gized its associated emergency bus. The air compressors will operate for ap-proximately -15 to 30 minutes, depending on the number of start attempts, and represent a total load of 12 KW on each EDG. If summed with the MESL for each EDG, this load does not raise the load on any EDG to the qualified load. Dawe et al. , ff. Tr. 27153, at 17.

L-27. If all the intermittent loads were simply summed and added to the ME.SL for each EDG, the conservatively predicted total load would not exceed 3331.4 KW for EDG-101, 3285.4 KW for EDG-102 and 3284.6 KW for EDG-103. Thus, even with this conservative approach, the load on only one EDG can be predicted tc, exceed the 3300 KW qualified load, and then by less than 1% for no more than a few minutes. Dawe et al. , ff. Tr. 27153, at 18; Tr. 27746-47 (Berlinger, Knox). But see Tr. 28188-90 (Knox) (he does not know this for a fact).

L-28. The 3300 KW qualified load would not actually be exceeded even when the intermittent or cyclic loads are operated. The conservatism in the calcula-tions that lead to the MESL as well as the coincident summation of intermittent loads makes the prediction unrealistically high. E.g. , Tr. 27190-01, 27194, 27199-200 (Dawe); Tr. 27788-90 (Berlinger). The quantifications of these con-servatisms identified by the LILCO witnesses is far greater than the 31.4 KW by which the EDG-101 loads are conservatively predicted to exceed 3300 KW. Thus, there is sufficient margin to ensure the diesel generators have the capacity and capability to power the intermittent and cyclic loads, along with the equipment

included in the MESL, without exceeding the qualified load. Tr. 28282 (Berlinger); 27190-91, 27198-200 (Dawe).

L-29. An integrated electrical test (lET) has been performed at Shoreham with the TDI diesel generators. Although the IET does not precisely simulate i post-LOCA conditions, its results are a reasonably close approximation of 1

post-LOCA diesel generator loads. The IET starts from the introduction of acci-i dent (LOCA) and loss of offsite power (LOOP) signals, and proceeds through the design sequencing and extended operation of loads on the EDGs. Tr. 27412

( Dawe) . The major EDG electrical loads following a LOOP /LOCA are closely ap-proximated by achieving post-LOCA values for plant parameters such as ECCS flows . The loads observed during the IET are estimated to be within a few per-cent of the actual loads that would be observed following a LOOP /LOCA. Dawe et al. , ff. Tr. 27153, at 19-20; Tr. 27219-21 (Dawe) .1/

L-30. The peak loads measured during the IET were 2833.6 KW for EDG-101, 2806.9 KW for EDG-102 and 3072.0 KW for EDG-103. These loads are significantly lower than the predicted MESL and the qualified load for each EDG, see Findings L-2 and -8. In addition, both reactor building service water pumps powered from EDG-103 were started automatically during the IET. Had they been operated as they will be during plant operation, only one service water pump would have started, resulting in a peak load on EDG-103 as much as 358 KW less than the IET measured 3072.0 KW. Dawe et al. , ff. Tr. 27153, at 20-21.

L-31. The Suffolk County witnesses expressed concern that the differences between the IET results and the calculated MESL values could not be explained, l

l l

1/ Staff witness Knox testified that preoperational and periodic load accep-tance tests, as defined in IEEE Standard 387, are performed in which actual plant design loads are sequenced onto the diesel generators. He did not know if the IET constituted this test. Tr. 28213-15 (Knox); Knox, ff. Tr. 27735, at 7.

Although this record does not clarify this point, the Board notes that LILCO's description of the IET would indicate that it does constitute the load accep-tance test referred to by Mr. Knox and identified in Regulatory Guide 1.108.

-A f: __

s i

even when nameplate conservatisms and timing of the loads was considered. Thus, -d i "E'

they questioned whether the IET was representative of actual LOOP /LOCA loads. [_--

Tr. 27545, 27553 (Bridenbaugh); see also Tr. 27570 (Minor). The County witness-a es, however, did not present specific information to support their position. 3 g

l, See, ea, Tr. 27538-40 (Minor), 27552-54 (Bridenbaugh). ~.

w L-32. The NRC Staff witnesses, with the exception of Dr. Berlinger, were not familiar with the IET. Tr. 28151-54 (Knox, Clifford, Eckenrode, Buzy, E'

=

Berlinger). Dr. Berlinger was aware of the IET through information provided by i_

LILCO. He did not rely on the IET for his conclusions because he had not per- 3 3

formed a detailed review, and LILCO did not recommend using it as the basis for

{

establishing the qualified load. Dr. Berlinger also stated he did not have a j j basis for assuming that the IET was a 100% accu. ate representation of how the plant would respond to a LOOP /LOCA. Nevertheless, Dr. Berlinger believes the [ ---

I IET gives a better estimate of the diesel generator loads in response to an ac- g cident than does.the MESL, which he believes to be a conservative calculation. 1 Tr. 28152-56, 23272-73 (Berlinger) . b L-33. The LILCO witnesses provided specif;c information about the conduct $

3%

of the IET. Cumulative load on the diesel generators was measured during the ] _

J IET. The equipment which must operate in response to a LOOP /LOCA, incloding the -

-J

[ intermittent and cyclic loads as appropriate, was operating during the IET. Tr.

K r

1 27674-76 (Dawe) . The loads observed during the IET are significantly below the [_

predicted MESL because the MESL is conservative, not because equipment is not -Q -

operating. Tr. 27200-01, 27462-65 (Youngling, Dawe). g i L-34.

4 The significant difference between the peak loads observed dt ring ) 7

) the IET and the predicted MESL is due, in large part, to conservatism introduced A i N-into the calculation of the MESL by the use of nameplate loads and the assump- En j tion of coincident demand. Tr. 27461-62 (Dawe) . The conditions that exist dur- h e

-+

j ing the IET are more representative of the actual LOOP /LOCA than are the y j ,_

s ;

=

I; k conservative assumptions used in calculating MESL values. E.g. , Tr. 27673-74 Q (Dawe); see also findings L-14 through - 19. Thus, the IET results do provide substantial confirmation that significant conservatisms exist in the calculated MESL values, and that there is significant margin between the qualified load and the emergency power requirements of the plant, including the intermittent and cyclic loads identified by LILCO, to ensure the qualified load will not be ex-ceeded. E.g., Tr. 28282 (Berlinger).

L-35. IEEE-387 defines the design load as that combination of electrical -

lozds having the most severe power demand characteristic which is provided with electrical energy from a diesel generator unit for the operation of engineered safety features and other systems required during and following reactor shut-down. Tr. 27995 (Knox). In accordance with IEEE-387, the continuous and short-term ratings of the diesel generator should encompass the design load. Tr.

28172-74 (Berlinger). Here, the diesel generators have a qualified load which bounds the MESL. This qualified load, given the conservatisms in the MESL, i

bounds the maximum expected loading of the diesel generators, including inter-mittent loads. L-28, -30, -34. Thus, the qualified load, although not defined in IEEE-387, is essentially an overload or short-term rating in that it encom-passes the most severe loading conditions expected for operation of required systems. See Tr. 27181-82, 27190-91, 27193-95, 27197-200 (Dawe). But see Tr.

27993-98 (Knox) .

ll1. Potential Instrament Error Does Not Invalidate Qualified Load or Endurance Run L-36. Suffolk County contends that LILCO failed to consider the potential inaccuracy of the diesel generator load meters in establishing the qualified j load. Thus, the County witnesses testified that the qualification testing of EDG-103 established a qualified load of no more than 3230 KW (Contention 1

j (c)(iv)) and that there is a small uncertainty which may result in excess load-

ing of the diesel generators during future surveillance testing (Contention v

m

-im-m iemum m-mi m

(a)(ii)) . Bridenbaugh and Minor, ff. Tr. 27500, at 4, 21-23, 31; see also Tr.

27581-82 (Bridenbaugh). Based on the evidence, we disagree.

L-37. Total KW load for each diesel generator is displayed in the main control room by a Weston watt meter with a specified meter accuracy of 2o of full scale and an overall instrument accuracy of 2R of full scale when combined with the instrument loop. Dawe et al., ff. Tr. 27153, at 28.

L-38. As part of the Shoreham instrument calibration program, each EDG watt meter is calibrated annually, along with its associated instrument loop, in accordance with an approved station calibration procedure. Calibration is per-formed with a calibrated reference standard traceable to NBS. Dawe et al. , ff.

Tr. 27153, at 28, 29; Tr. 27266-68, 27384 (Youngling); Tr. 27309-10 (Dawe) .

Calibration of the diesel generator load meters will be a Technical Specifica-tion requirement. Knox, ff. Tr. 27735, at 10.

L-39. Calibration checks performed on the EDG 103 watt meter just prior to and following the 525-hour endurance run showed that the instrument was accurate to within 60 to 70 KW between 3000 KW and 4000 KW. Dawe et al. , ff.

Tr. 27153, at 28-29, 39. Calibration checks performed on the EDG 101 and 102 watt meters in September of 1984 showed that these instruments as well were accurate within 70 KW in the range between 3000 KW and 4000 KW. Tr. 27265, 27268 (Dawe). Thus, the load meters can reasonably be expected to be accurate to within 170 KW at or about the qualified load of 3300 KW.

L-40. The calibration records for the watt meters indicate that, within the 170 KW accuracy range, observed error is as often plus as it is minus.

Thus, when the diesel generator is operating, the observed value on the meter is a reliable mean. There is a high probability that even if the instrument is in error within 70 KW, it will be low as often as it is high. Thus, it is valid

'o monitor and control diesel generator loads using the control room load meters as read. Tr. 27309-10 (Dawe); 28427 (Pischinger).

l

15- 2 r

L-41. A confirmatory test was performed on EDG 103 to establish the qual- -_

r ified ioad by accumulating 10E7 loading cycles at 3300 KW. It consisted of 221 s hours of operation accumulated during preoperational testing at loads at or g above 3300 KW, and a 525 hour0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months endurance run performed at approximately 3300 KW. Z I-Dawe et al. , ff. Tr. 27153, at 38; Tr. 27313 (Youngling). In addition to r

calibration data for the EDG 103 load meter before and after the 525 hour0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months endur- -

=

ance run, Finding L-39, there is calibration data from before and after the pe- ,

=

riod of time during which the initial 221 hours0.00256 days 0.0614 hours 3.654101e-4 weeks 8.40905e-5 months of operation were accumulated. y These data show the meter accuracy to be within + 60 KW. Tr. 27315 (Youngling) .

~

Further, of the total hours accumulated during the 10E7 cycle confirmatory test

=-

(221 hours0.00256 days 0.0614 hours 3.654101e-4 weeks 8.40905e-5 months plus 525 hours0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months ), a significant portion of the hours was recorded f =

utilizing a digital test loop used with the process computer which has an accu- 5%

racy of approximately 0.6 percent. This would further reduce any effect of in- {

strument error. Tr. 27311-14, 27423 (Youngling). j i

L-42. In addition to the fact that load meter readings during the endur-i]

m ance run were reliable mean values, Tr. 27307 (Youngling), Finding L-40, a sub-g stantial portion of the 10E7 loading cycles were accumulated at recorded loads M greater than 3300 KW indicated, Finding L-50. These hours of operation above "". 2 3300 KW further assure that the confirmatory test demonstrates a qualified load W

of at least 3300 KW. See Finding C-8. m L-43. If the diesel generators are called upon to operate, maximum load- Q ing would occur following a LOOP /LOCA, the event which establishes the MESL. $

4-Initially, diesel generator loading is automatic and below the qualified load. l It is thus independent of load meter accuracy. Subsequent operator action ini- _h T-tially results in load reduction, such that loads are significantly below the p qualified load when discretionary loads are considered by procedure for op-eration . 4' Dawe ei 31. , ff. Tr. 27153, at 29-30; Tr. 27228-31 (Dawe, Notaro). m The amount by which the actual loads will be below 3300 KW has been demonstrated W 7_.

M

+

y by the MESL conservatisms and the IET, e.g., Findings L-15, -16, -17, -30.

Thus, although allowed by technical specification and procedure, it is neverthe-less unlikely that diesel generator loads will approach to within 70 KW the 3300 KW qualified load. Id. at 30. Should a diesel generator be operated at 3300 KW indicated, the indication is a reliable mean value, Finding L-40.

L-44. Further assurance is gained from the diesel generator load alarms to be installed by LILCO, Finding L-59. These alarms will take input from the instrument loops which input to the load meters, but will otherwise be indepen- _

dent of the load meters. The instrument loops, comprised of very accurate ele- -

ments, are accurate to about 0.5%. Tr.' 27339-40 (Youngling) .

L-45. During future surveillance testing, each diesel generator will be operated at 3300 KW for periods of time, Finding L-51. As previously discussed for the 525 hour0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months endurance run, the load meter utilized to conduct this testing provides a reliable mean value, Findings L-40, -42. Therefore, it is appropri-ate to conduct periodic surveillance testing at 3300 KW indicated on the control room load meters. Indeed, to the extent there might be any concern for test loads in excess of 3300 KW due to instrument error, the diesel generators will have been tested and subjected to subsequent required inspections. See Findings C-16, -17, B-21, -27, -30, -41, -43, -48, -50, -62. -

L-46. The ability of the diesel generators to accommodate relatively small loads in excess of 3300, should they occur due to instrument error, has been assured by substantial diesel generator operation at 3500 KW and above, in-spections already completed, and extensive engineering analysis. See Findings _

C-8 to -11, B-21, -22, -30, -40, -41, -43, -48, -50, -62. For the reasons we have explained, we conclude that the potential for small instrument errors did not invalidate the testing to establish the qualified load of 3300 KW, and will not render the qualified load inadequate in the future.

F L

IV. The Tolerance Band of 1100 KW is Acceptable for Purposes y

of Surveillance Testing and the Endurance Run

! L-47. Suffolk County contends that the test band of 100 KW utilized for L the 525 hour0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months endurance run and intended for use during future surveillance i

j .-

testing at 3300 KW renders the endurance run (Contention (c)(iii)) and the qual-9; ified load (Contention (a)(ii)) inadequate. The County witnesses testified, 3

[ however, that the test band did not significantly affect the endurance run in I

light of recorded data,2/ but that it was a concern that the qualified load f would be exceeded during tuture surveillance testing. Bridenbaugh and Minor, v ff. Tr. 27500, at 23, 30-31; Tr. 27582-83 (Bridenbaugh). The record demon-strates that the 100 KW test band is acceptable in both instances.

f L-48. Testing of a diesel generator at 3300 KW requires that it be con-

{

j_ nected to the offsite grid. In-plant loads alone are insufficient for such a f test. When the diesel generator is connected to the grid, it is difficult to

[ maintain a perfectly constant load value due to engine response to fluctuations 1 on the grid and an independent pulsation effect on the meter due to the mode of a

i governor operation. These effects exist only when the diesel generator is con-t nected to the grid, and not when it is operating independent of the grid to sup-f ply plant loads. Thus, experience has shown that a test band of at least 100 f KW is required. Tr. 27316-22 (Dawe, Notaro, Youngling); Tr. 27328-32 (Dawe, i

{ Youngling); Tr. 27389-92 (Youngling) . The Staff agrees that testing should be 3a accomplished at a mean load of 3300 KW, achieved with a 100 KW test band about h 3300 KW.3/ Tr. 27801 (Berlinger); Knox, ff. Tr. 27735, at 12. See also Tr.

13

il f 2/ The County witnesses also testified that the accuracy of certain data f recorded during the endurance run (ea, 3326 KW) called into question the va-L lidity of that data. LlLCO witness Youngling identified the source of the data f in question to be a very accurate digital measurement test loop, thus resolving E this question. Compare Bridenbaugh and Minor, ff. Tr. 27500, at 30-31 with Tr.

5 27311-12 (Youngling) .

t-3 3/ Staff witness Henriksen, in prefiled testimony, alluded to blocking of the g governor to control the load more closely. Bush et al. , ff. Tr. 28503, at 11.

(footnote continued)

r E 27416, 27425, 27449-50 (Youngling); Tr. 27482 (Dawe).

L-49. . A + 100 KW tolerance band around the 3300 KW test load is required as a practical matter because if the operator must test at 3300 KW with no con-trol band, then, at any instance when the meter read even slightly above or

below 3300 KW, the operator would be in violation of the Technical Specifica-tion . Tr. 27318, 27328-29 (Dawe) . In fact, the operators are trained to run the tests at a mean load of 3300 KW and not operate unnecessarily at loads al-lowed by the band. Tr. 27318 (Notaro); Tr. 27329 (Dawe) . This has been demon-strated by the fact that the majority of readings taken during the 525 hour0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months en-durance run were at 3300 KW. See Finding L-50.

L-50. Review of operating logs shows that during the 525 hour0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months endurance run, which was part of the 10E7 loading cycle demonstration of the qualified load, 81 hours9.375e-4 days 0.0225 hours 1.339286e-4 weeks 3.08205e-5 months were recorded at loads above 3300 KW but not exceeding 3400 KW.

Twenty hours were recorded at loads below 3300 KW but not below 3250 KW.

Readings were taken every 30 minutes. The balance of the 10E7 loading cycles was accumulated during 221 hours0.00256 days 0.0614 hours 3.654101e-4 weeks 8.40905e-5 months of operation, at or above 3300 KW, prior to the

endurance run. Dawe et al. , ff. Tr. 27153, at 38-39. Since significantly more hours of operation were accumulated above 3300 KW than below, the + 100 KW test ,

band did not prevent demonstration of a qualified load of at least 3300 KW. See Finding C-8.

L-51. Once the EDGs are placed in service, they will be subjected to pe-riodic surveillance testing as prescribed by the Technical Specifications.

While the Technical Specifications have not yet been issued, LILCO and the NRC (footnote continued)

The evidence shows that blocking of the governor on the high side would not

eliminate downward fluctuations, and could result in testing at a mean value b

less than 3300 KW, Tr. 27491-94 (Youngling); 28703-04 (Henriksen). Further, Mr.

L Henriksen was not familiar with the details of the governor used at Shoreham nor

! the cause of load fluctuations during testing. Tr. 28720-21 (Henriksen).

i e

i.i

? Staff consider that appropriate surveillance testing for each EDG should include one hour of operation per month at a mean load of 3300 KW (3300 KW 100 KW), a

24 hour2.777778e-4 days 0.00667 hours 3.968254e-5 weeks 9.132e-6 months run at a mean load of 3300 KW (3300 KW 100 KW) every eighteen months, a load rejection test and a load acceptance test in which actual plant design f- loads, including cyclic and intermittent loads, will be sequenced on to the EDGs or simulated. Tr. 27413-14 (Youngling); Tr. 28213-14 (Knox); Knox, ff. Tr.

3 27735, at 7-8. The Staff and LILCO agree that a test band of 100 KW is appro-h priate for testing at 3300 KW, and will result in a test at a mean load of 3300 KW. Findings L-48, -49.

, L-52. The ability of the diesel generators to accommodate loads in excess

=

P of 3300 KW due to the test band during surveillance testing has already been ad-equately demonstrated. Finding L-46. Further, the testing itself will demon-strate the ability of the diesel generator to accommodate the loads at which it is tested.

=

g L-53. Thus, on the basis of the evidence presented, the 100 KW test

-- band did not result in an inadequate test to determine the qualified load, and f

it will not affect the adequacy of the qualified load in the future. Indeed, it is illogical to consider testing in the future at some load above the qualified load to accommodate the test band, since the test band would still be required.

[

=

V. Qualified Load Need Not include Operator Error Load 4/

F 4/ While the Board initially denied LILCO's motion to exclude Contention E Ta)(iv) (Morris, J., dissenting), it noted that it would reconsider this issue.

E Memorandum and Order dated January 18,1985, at 4-6; see also Tr. 27124, 27257-61, 27958 (Brenner, J.). LILCO submits that reconsideration is warranted

, because the record demonstrates that Contention (a)(iv) challenges the Commis-7 sion's regulations. See 10 CFR S 50, Appendix A (single failure criterion),

b By SC's lights, the thrust of (a)(iv) is that the EDGs must be designed and y qualified to accommodate single worst case operator error loads as well as safe-g ty function loads. Previously, SC argued that accommodation of operator error E loads in diesel generator capacity was a part of the safety function. Response k of Suffolk County dated January 25, 1985, at 2-3. The record shows, to the con-(r trary, that GDC 17 does not require the assumption of worst case single operator

, (footnote continued) le 1 -

L-54. EDG compliance with GDC 17 is determined by ensuring that the plant design load does not exceed the EDGs' capacity and capability. Knox, ff. Tr.

27735, at 4, 7. EDGs meet GDC 17 where, as here, they are qualified for the de-sign load. Tr. 27758-59, 28282 (Berlinger); Findings L-28, -30, -34, -35. The design load, as defined in IEEE-387-1977, does not include loads attributable to operator error. Tr. 27796-97, 28174 (Knox); Tr. 28277-81 (Berlinger, Hodges);

Tr. 28351-53 (Clifford); Finding L-35. Thus, operator error loads need not be considered in determining design adequacy of EDGs or in making capacity and (footnote continued) error as a design basis for an individual EDG. Rather, GDC 17 requires that EDG design and qualification accommodate only that equipment necessary to maintain safety functions. See Findings L-54, L-55. A requirement that each EDG include operator error loads as a design basis, and be sized and qualified accordingly, would effectively require a system design that would provide emergency power as-suming not only a LOOP /LOCA event and a single failure of a diesel for any rea-son, but, in addition, an operator error which otherwise would be assumed to cause a second diesel to fail. Such a design basis assumes multiple failures beyond the single _ failure contemplated by NRC regulations. See Tr. 27429-32,

=

27459 (Dawe); but see Hodges, ff. Tr. 27729, at 6.

Nor is there any reason to doubt the applicability of the single failure criterion. If, as LILCO contends, it has met its burden with respect to the re-maining diesel contentions, then the EDGs are qualified to perform their safety functions pursuant to GDC 17 and are thus at " Ground Zero" (Tr. 27958) such that the single failure criterion is applicable and dispositive of (a)(iv). Further, no exceptional circumstances have been alleged or shown to avoid this conclu-sion . See LILCO's brief dated January 15, 1985, at 4 12. Though not directly in point, Public Service Co. of New Hampshire (Seabrook Station, Units 1 r, 2),

LBP-82-76,16 NRC 1029 (1982) is persuasively analogous. There, a contention was rejected as not having a regulatory basis where intervenors charged that plant design should be revised because that design did not accommodate a single failure in the emergency feedwater system together with an operator error to correct the failure. Id. at 1059-60.

Concerns regarding procedures and training do not save (a)(iv). The deter-mination that the EDGs meet GDC 17 can be, and was, made by the Staff solely on the basis of design, independent of procedures and training. See Findings L-61 to L-63. Moreover, even if procedures and training were necessary for resolu-tion of the contention, LlLCO's testimony has demonstrated that procedures and training do provide assurance that operator error is unlikely and can be prompt-ly corrected. Findings L-65 to L-78; see also Findings L-56, L-59.

Accordingly, LILCO respectfully requests reconsideration of LILCO's motion l to exclude Contention (a)(iv).

l l

capability determinations pursuant to GDC 17. Tr. 27948-54, 28149, 28277-81 (Berlinger, Hodges); see also Tr. 28197 (Hodges).5/ EDG ratings do not have to include operator error loads as a design basis, nor must the EDGs be sized to accommodate such loads; rather EDGs should be sized to accommodate the operation of necessary equipment to maintain safety functions. The 3300 KW qualified load accommodates all the equipment load at Shoreham to maintain safety functions.

Tr. 28350-53 (Clifford); Tr. 28282 (Berlinger). In addition, testing at a mean qualified load of 3300 KW has established the capacity and capability of the EDGs to support the loads reasonably expected to occur following a LOOP /LOCA.

Tr. 27787, 28282 (Berlinger); see also Tr. 28356 (Clifford).

L-55. Further, since IEEE-387-1977 specifies that diesel generators may be used to the limit of their power capabilities as defined by the continuous and short-term ratings, no margin between the design load and rating is re-quired. The short-term rating is not intended or required to accommodate opera-tor error loads. Tr. 28173-74 (Berlinger, Knox); see also Tr. 28198 (Hodges);

Tr. 27455-56 (Dawe) (former 3900 KW short-term rating did not accommodate opera-tor error when fourth service water pump connected as automatic load). Thus, the short-term rating is not intended to accommodate operator error loads; it is 5/ Notwithstanding some confusion and contradiction (particularly in the tes-timony of Mr. Knox), compare Tr. 27796-97, 28174 (Knox) with Knox, ff. Tr.

27735, at 9, the record as a whole supports the conclusion that EDG ratings are not required to accommodate single worst case operator error loads. E.g . , Tr.

27948, 27952-53, 28149, 28277, 28280 (Berlinger); Tr. 28309-10, 28350-53 (Clifford). Significantly, Dr. Berlinger, correcting earlier testimony, stated that the capacity and capability of EDGs up to this particular case have not been determined by adding worst case operator error loads to determine a short-term rating. Tr. 28277, 27280 (Berlinger). Given that regulations do not re-quire an individual diesel generator to be sized to accommodate the worst single load which could be added in error, it is not surprising that the Staff witness-es had no basis on which to testify as to whether or not, for previously li-censed plants, diesel generator ratings encompassed loads due to operator error.

See Tr. 27956-60 (Knox, Hodges, Buzy, Clifford, Eckenrode, Berlinger); 28036-37 (Berlinger, Buzy, Clifford, Eckenrode); 28196-201 (Hodges, Buzy, Knox). There may well be cases where ratings do not encompass cuch operator error loads. Tr.

28200 (Hodges, Knox) .

i

s

? instead intended to accommodate the design load where that load's profile may exceed the continuous rating for some period of time. Similarly, in the quali-

=

fied load concept, there is no requirement that the diesels be tested for 10E7 cycles at a load which accommodates operator error. Tr. 28175 (Berlinger). The qualified load concept is therefore no different from the IEEE-387-1977 and Reg-

\ ulatory Guide 1.9 approach with respect to the treatment of operator error.

Neither requires that the diesels be sized to accommodate operator error.

L-56. The single worst case loads that could be started erroneously by an operator following a LOOP /LOCA when added to the MESL yield a total load of

[

3459.4 KW for EDG-101, 3414.8 KW for EDG-102 and 3583.5 KW for EDG-103. Simi-larly, the maximum loads that would result from an operator error following a LOOP are 3741.8 KW for EDG-101, 3575.2 KW for EDG-102 and 3707.9 KW for EDG-103.

(

Operator errors of adding the single worst case loads following a LOOP or a I LOOP /LOCA are unlikely to occur. The equipment that makes up the single worst

[ case load is, in each instance, not needed to mitigate or control the event.

Further, the design of the plant greatly minimizes the need for operator action 5 during the initial time in a LOOP /LOCA event when the EDGs are at their maximum load. During this period, plant design is such that operator action is to veri-E

. fy operation of automatically actuated equipment. Dawe et al. , ff. Tr. 27153, 5

at 32-35.

f K

L-57. Even if an operator error were to occur, it is unlikely that the resulting total loads would be as high as those listed in Finding L-56 because s

of conservatisms associated with the MESLs and the individual erroneous loads.

Dawe et al. , ff. Tr. 27153, at 32-35; Tr. 27482-84 (Youngling) (LOOP op irator error load conservative by a factor of 2). For example, a more realistic as-

, sessment of the effect of operator errors following a LOOP /LOCA is obtained by using the IET measured loads rather than the MESL. In this event, the single worst case operator error in the LOOP /LOCA, even if it occurred, would not

. result in loads above the qualified load. Dawe et al. , ff. Tr. 27153, at 32-33.

h

'3-i L'

j( L-58. Further, even if an operator error were to occur to add the single worst case load, there is reasonable assurance the EDGs would not fail. See Findings C-12, B-21, -30, -43, -48, -62. While the NRC Staff did not consider that testing at 3300 KW alono qualifies the EDGs for loads above 3300 KW, the

~

Staff's overall evaluation considered accumulated hours of operation and testing F above 3300 KW as well as those at 3300 KW. Therefore, the Staff was able to conclude that the EDGs can sustain the conservat'vely estimated operator error loads in Finding L-56 above in addition to the conservative LOOP /LOCA load pro-file assumed by the NRC Staff. See Tr. 28204-05 (Berlinger); Bush, et al., ff.

Tr. 28503, at 13, 21-22. See Findings C-14, B-23, -36, -43, -62.

L-59. There is also reasonable assurance that if, through operator error, a load were added to increase the total EDG toad above 3300 KW, the error would be promptly recognized and remedied.g/ A crew of licensed operators, cognizant i

of plant conditions, is present in the main control room. They have available to them clear and concise indication of load for each diesel generator. The

! diesel generator load meters in the control room will be banded at 3300 KW. The operators are trained and knowledgeable in the diesel generator qualified load.

~

Dawe et al. , ff. Tr. 27153 at 33, 35; Tr. 27297-98 (Youngling). In addition, LILCO has committed to provide a distinctive visual and audible alarm for each diesel generator in the main control room that will be set no higher than 3300 KW for operation. Tr. 27298-302, 27334-35 (Youngling) .

i s/ SC witnesses implied operator error was a particular concern in a LOOP /LOCA i as it could be hours or days before plant conditions are fully understood.

Bridenbaugh and Minor, ff. Tr. 27500, at 24. In the context of the diesel gen-erators, this is not meaningful. The diesel generators are required only if,

, and for only so long as, offsite power is lost. The Low Power Licensing Board j h. s determined that offsite power is unlikely to be lost, but even if Icst, it

_ can be restored within 30 minutes. Long Island Lighting Co. (Shoreham Nuclear

Generating Plant, Unit 1), LBP-84-45, 20 NRC 1343,1367-70,1389-90 (October 29, J; 1984). While this does not alter the requirement for an onsite power source i under GDC 17, it correctly places in context the potential duration of operation

) with only onsite power sources. See Tr. 27303, 27428-29 (Dawe); Dawe et al. ,

i ff. Tr. 27153, at 37.

r E

_ L-60. Even in the unlikely event that an operator error results in the

=

failure of one .EDG, the on-site emergency power system is designed to meet the single failure criterion. The on-site power system accommodates the loss of one g EDG for any reason, including operator error, because all required safety func-( tions can be satisfied within the qualified load by the remaining two EDGs. Ac-cordingly, there is no requirement or nucessity to include single worst case op-erator error loads within the qualified load.

Tr. 27429-32, 27459 (Dawe); Tr.

28149, 28279 (Berlinger); Tr. 28174 (Hodges); Tr. 28350-52, 28356 (Clifford);

see also Findings L-5, -7. ,

L-61. Issues related to procedures and training are separate from the issue of the design adequacy of the diesel generators. The procedures and training are reviewed to evaluate the operation of the plant, given that the de-T sign is adequate. The procedures are not part of the design. Tr. 27882, 5 28354-55 (Clifford); 27885-87 (Berlinger) .

L-62. The Staff, with Pacific Northwest Laboratory, reviewed the design to determine whether or not the diesel generators satisfied the requirements of GDC 17. The Staff conclusion that the diesel generators do provide a reliable source of standby power in accordance with GDC 17 is independent of, and does i not rely upon, the procedures. Tr. 28274-75, 27884-86 (Berlinger) . The Staff considered some short-term excursions above 3300 KW as being possible but not probable, and determined the diesels could support such excursions. Tr.

- 27948-49, 28176-80 (Berlinger) . See Bush et al. , ff. Tr. 28503, at 13, 21-22.

T L-63. Assuming a given plant design, the Staff review of procedures eval-uates the capability of the operators to operate within the plant design while

& maintaining the required safety functions accommodated by the design. Thus, the F

{

procedures provide additional assurance, beyond the requisite reasorable assur-

ance provided by design, that the EDGs will satisfy functional requirements in response to an accident or severe transient. Tr. 28343, 28347, 27882-83 E

(Clifford); Tr. 27885-86 (Berlinger). The extensive testing and analysis of the EDGs _ confirms 'their capacity and capability to provida the required electrical p:wer to the plant to' satisfy all safety requirements in sccordance with GDC 17

, during and 'following a LOOP or a LOOP /LOCA. This reasonable assurance provided by the design and confirmed by the testing _and analysis is adequate to support '

. ths issuance 'of a license to operate at the qualified load. Thus the design, without' regard to the procedures, satisfies the requirements of GDC 17. Tr.

27787, 27884-87 (Berlinger). Therefore, the finding of reasonable assurance of no undue riskito the public health and safety _ can be made on the basis of the dasign alone.

L-64. Suffolk County provided no specific evidence addressing procedures or training. The Suffolk County witnesses testified that they had examined the procedures governing operation of the EDG equipment in the emergency situation and found the operations to be relatively complex, offering many opportunities for error. The testimony, however, consisted of no more than a brief summary discription of four procedures. Bridenbaugh and Minor, ff. Tr. 27500, at 25-28.

Tha County witnesses had only limited, general experience with emergency proce-du ras . Tr. 27504-11 (Bridenbaugh, Minor). Their examination of the procedures consisted of some review, but not detailed analysis. Tr. 27562-64 (Bridenbaugh, Minor) . This testimony does not aid in understanding, or provide a basis for cvaluating, the adequacy of procedures.

L-65. In contrast, LILCO's witnesses testified in -substantial detail con-c:rning the adequacy of the procedures and training as they relate to main-taining diesel generator loading below the qualified load. LILCO's testimony in this area was provided by witnesses with significant experience related to Shereham. Dawe _et al., ff. Tr. 27153, at 2-5. These witnesses had participated i

l In the preparation of both the procedures and training. See, ea, Tr. 27353

! (D:we); 27372 (Notaro). LILCO witness Notaro, currently the Outage and

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

? n *

+ j

. t y --261 7

' *, g

, g_

my

~

, Modifications. Manager, is a licensed Senior Reactor. Operator for Shoreham, and w, -

E , was formerly,< as Shoreham Chief Operating Engineer, responsible for the formu'la- 2 tien and implementation:of training programs for all Shoreham personnel. Dawe

!u .-

1 ~e t al. , 'ff./ Tr. 27153,' Attachment A; Tr. 27380 (Notaro) . These witnesses u n -

t %stified that the procedures and training would minimize the likelihood of an ~ '

l .

_6rror by an operator that would result in a diesel generator load in excess of

/ -

{ -'tho 3300 KW qualified ' load.,

Dawe et al. , ff. -Tr. 27153, at 24; Tr. 27155

, ((Youngl_ijg) . s A '

1 ,

L~-66. The LILCO4 witnesses identified a number .of emergency operating pro-c: dures and system pr'ocedu,res that had been reviewed and, in some cases, revised g e 'i -

j as a result of establis'hing th's qualified load. Tr_. 27156-61, 27262 (Notaro). .

i. Significantly, ~ these are not new procedures. They aPg the saide procedures wh,ich n w r I

ths operators have utilized, trained with,~and practiced with '"at the simulator l A

' for ,some time. .The' ?.hanges which have been made ape mainly' added cautions to c :

4 l highlight the diesel s'enerator load. Tr. 27233 (Dawe); Tr. 27367 -(Youngling); '

i s .

Tr. 2/372; 27395, 27454-55 (Notaro).

l 'Mr. Notaro described these cautionias

. 3, 4 j bsing eninnc6ments'that provide information to the operators. Tr. 27473-74 l (Nota ro) . *

. ., .c L-67. In , response to a LOOP /LOCA, four procedures (loss of offsite power, 1:v 1 control, emergency shutdown and cortainment s control) may be en c'd simul-

<l t:neously. . Tr. 27277-78, 27368 (Not.m..)l ,p 7Mr.y Notaro testified that th.ere is no

-a ,-

m:nageability concern with the simvltaneous;usiif these procedures by th'e oper-

, l ~, .+.-y at:rs. He has verified at the simulator that th's procedures are manageable,- and W ~ j, .

,J

' i.h3 has observed-the Shoreham cperators utilize these sams procedures, without i & ,

2:n tha cautions, _ for a' number of years. The NRC-hasj testid the oycrators :in their .

p y v

! 7elaHity fFuse and manage the procedures, and they Chave been licensed. The op- , s y , .

2

[ htor(3,nre not confused.or_ misled by then~j multiple procedures.p Tr.%27434-35,-

m . M .. . , j-!

.?,7404-06 (Notaro); see also Tr. 27277 (NotarAI.T4fhsv?are tipical of 'the '

v

, .f procedufos for all BWR' plants.- Tr. 27885 { Berlinger) . c id Q > ,

%./

,,}

N n a

'v .

' h.

if. . 4 mel a- l aak 1. . - - . . . . s. u:

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

l 4

L-68. The plant is operated by a shift crew comprised of a number of li-csnsed operators, including two Senior Reactor Operators in the positions of Watch Supervisor' and Watch Engineer. The responsibilities of each member of the crsw are set forth in an administrative procedure. The Watch Supervisor is re-sponsible for using the procedures to ensure the operators are performing their functions properly in response to a LOOP or a . LOOP /LOCA. The Watch . Engineer, 4

' with . responsibility fo'r the command control function, verifies that the crew j ovsrall .is complying with the procedural requirements. The operators have pre-

dstarmined functional responsibilities such as level control, containment. con-

trol, heat sink availability and AC power distribution. Tr. 272G9-75;(Notaro).

4 L-69. The plant is designed for automatic operation for the first ten minutes following a LOOP /LOCA. During that time, the operator's function is to ,

verify that immediate actions identified in the procedures have occt/ red and to- f.

Initiate manually any of these automatic actuations that fail to' occui .

~

Tr.

27279, 27286-90 (Notaro). All the automatic actuations. are included in the MESL, and taus diesel loads would be below the qualified load. See Findings

-7, -8.

L-5, Although thire is no-formal prohibition to other actions. in this psriod of time, training and procedures do not require other actions forLthe first 10 minutes. In addition, some of the non-automatically. operated equipment

-is actually precluded from operation by design. Tr. 27279-81 (Notaro, ' Dawe) .

L-70. .The LILCO witnesses testified to a number of other matters which Y dtmonstrate manageability and adequacy of the procedures' as they relate to load-ing of.the diesel generators.1The operators function as a crew.' Th'e operator rssponsible for AC power-distribution monitors the diesel generator loads -

throughout the event and communicates with the' other members of the crew. Tr.

'27393-94 (Notaro). Each operator,' before adding load to the-diesel, verifies

' .thet the qualified load will. not be exceeded by directly communicating with the cp rator responsible for monitoring the diesel. :This communication is' e'asily -

__ w -r-- we w .- m , 1w t 4

l

-l accomplished within the control room. Equipment associated with a given diesel is clearly ident.ified by'_the equipment numbering on the control panels. Tr.

27399-400 (Notaro, Youngling). Equipment status changes are also communicated to the rest of the crew. Watch Engineer approval for such activity is neither warranted nor required, although he is aware of the overall plant st5tus. The I

~ only action requiring the approval of the Watch Engineer is the shutting off of an ECCS pump that started in response to an automatic signal. Tr. 27406-11, Field operators do not perform functions outside the control

~

27282 (Notaro).

room without direction from the control room operators. The control room opera-tors have the knowledge of diesel generator loading necessary 'to direct the fisld operators. Tr. 27379, 27436 (Notaro). Control room operators can monitor equipment status by means of indications available on the main control boards and, additionally,. equipment status is maintained in the control room log.

See Tr. 27375-78 (Notaro). Compare Tr. 28094-95 (Clifford) with Tr. 27905-06 (Eckenrode). Thu.s, .the procedures are m'anageable and do n'ot confuse or unduly burden the operators.

L-71. The . witnesses discussed two typos of procedures, emergency op-erating procedures and system procedures, used to guide the operators in the conduct of ' plant operations. The pertinent-emergency operating procedures have bosn revised to include cautions as a reminder of diesel generator loading con-

'ditions' when equipment operation is~ called for.

Although LILCO was still re-

viswin's the system ' procedures for a final decision on whether cautions were -

nuded, :the witnesses did not expect them to-be' necessary. The system proce-duras merely direct the "how to"Lof: system operation once the decision to oper-ata has :been made. This decision making is . guided by. the emergency ~ operating

' ~

procedures 1which have . cautions designed to ensure diesel generator loads are.

censidered before actions that can increase load are -taken. E. g . , -Tr. 27165, .

27171s 27473-74 (Notaro); 27170-72 (Dawe).

, .w

4 L-72. ~ Review and revision; of th'e procedures were performed by the cogni-i zant LILCO' Plant Staff section and then by the Review of Operations Committee. !

Immndiate and subsequent actions were reviewed and addressed. In initial revi- -

l ' sions,' generic caution statements were used for sequences of actions. In subse-quent' revisions, individual cautions were used for specific actions. The LILCO 4 witnesses testified'that either format was acceptable, but- they believed the

.Ittter was preferred by the NRC Staff reviewers.

The cautions -are stated in 50 ,

t KW increments to resolve any qu'estion concerning an operator's ability to read '

i . . i

!= accurately the diesel generator load meters in the control room. Tr. 27355-60

~(Notaro, Dawe); ~ 27476-77 (Dawe); 27291-94, 27421 '(Youngling); see also Tr.

28101-06 (Clifford, Eckenrode) .

L-73. The revised procedures were also verified at' the Limerick simulator

- by Mr. Not'aro and the Shoreham training supervisor. LlLCO personnel have trained at this simulator for four years, and thus they had the- necessary. famil-

- isrity with the simulator to check the procedures despite differences between -i tha simulator and Shoreham. See Tr. 27401-02 (Notaro). Mr. Notaro testified 7 that there was no difficulty in managing the revised procedures. There was also 1

no difficulty with taking time to evaluate diesel loading, or communicating be- ,

j' twasn operators, prior to taking' actions in accordance with procedures. They

wara able to verify this, despite the fact that' diesel' load limits' at the si-mulator, as is: expected at Shoreham, are not. seriously'. challenged by the equip-mtnt required to be operated. Tr. 27372, ,27400-05, 27476-71, 27.484-85 (Notaro).

This was a verification of the ~ procedures as opposed:to a verification of each-opsrator's , abilities'. Staff' witness Eckenrode-agreed with the process- LILCO '

2 ustd 'to verify thel manageability of the procedures, . but -stated that' the Staff

.wis looking for a further . verification that the operators, as well 'as the super-L lvisers, could manage the-procedures. Tr. - 28081 -(Ecken rode) .

m. p > s 4

e m n, , -, - . - , s ,- - y ~e

~

L-74. LILCO.is confident that the 3300 KW qualified load will not hinder tha operators' ability to accomplish ' actions specified in the procedures. E.g.,

Tr. 27164, 27467-68 (Notaro); see also Tr. 28358 (Clifford), Findings L-15, -16,

-17, -30, -34. :Nevertheless, LILCO has provided procedural guidance to the op-srators identifying the nonessential loads -that sh'ould be verified as not op-

srating,.or that preferentially can be deenergized, if the diesel generator load approaches or exceeds 3300 KW. See Tr. 27159-60, 27374-75, 27471-72, 27478 ,

. (Notaro); Dawe et al., ff. Tr. 27153, at 25-26.

l- L-75. LILCO .has also incorporated the qualified load into its training l

p rog ram. LILCO initiated the training by' placing revised procedures on a re- ,

{.

' quired reading list for senior reactor operators and reactor operators. - Re-

.quired reading;is an acceptable method of training. LILCO is also conducting i

classroom and simulator training' for the licensed operators as part of the . ,

j raqualification training program. A specific lesson plan related to the quali-

.fisd load has been developed for classroom training. At .the simulator, the op-orators will use the revised procedures, thus operating with the 3300 KW quali-1 l ..fied load. Classroom training related to'the qualified load began in

[ mid-February 1985 and will take six weeks' to complete for all six operating ,

I crsws . The simulator training follows the classroom training in the next six-wask cycle. Tr. 27177-79, 27262, 27353, 27361, 27373, 27398 (Notaro); 'Dawe- e_t

.als ff. Tr. 27153,' at 27.

L-76. The-LILCO training organization is responsible for ' certifying that training has been conducted properlyf and completed satisfactorily.

2 The training is certified by independent reviewers. The NRC reviews and ' evaluates the r2 qualification ~ training program on an annual; basis. After -the = initial cold li-4

. csnse examination,Tthe NRC may . test some. operators during requalification training' in - addition to LILCO' testing. . General Physics Corpo' ration, .which oper .

, ~ atzs- the _ simulator used by. LILCO, evaluates the examination; process,- examination

'Tmrgyy- Tv rw-W-f w'

i.

"qusstions,. responses and grading as an independent consultant. In addition, the

~

- LiLCO QA program andlthe Nuclear Review Board evaluate the training program.

Ths Nuclear Review Board also includes an independent consultant with extensive training . experience. Tr.- 27381-83 (Notaro, Youngling) .

L-77. .With the exception of agreements made with the NRC reviewers during

hchnical meetings held February 27 through March 1,1985, LILCO had essentially completed the revisions to the procedures of interest. Tr. 27162-65, 27175-76 (Notaro); see also Tr. 28288, 28320-22 (Clifford). Even before meeting with .the

~

Staff'. reviewers, LILCO's . witnesses believed LILCO had been responsive to Staff concerns and requests for. information about procedures. Tr. 27395-96 (Notaro).

. One major area of disagreement existed. LILCO dP not believe a task' analysis nssded to be performed to confirm the correctner 4 of the procedures, because the oparators' tasks had not changed. Neither the equipment to be operated nc 'he indications and controls in the control room had changed. In a'ddition,- LILCO is ,

rsquired to conduct a full control room design review, including task analysis, to be completed- by the first refueling outage. Tr. 27363-67,' 27396-97 (Notaro, Youngling); 27405 (Notaro); 27902 (Eck'enrode). Notwithstanding, during the.

course of the hearing', LILCO committed to the Staff to perform a job task analy-i . -

ses specific to diesel' generator operation within the qualified load, which the

. Staff witnesses found acceptable. E.g. , Tr. 28291-97 (Clifford, Buzy, Ecksn rode) .

L-78. -On .the . basis of LiLCO's' evidence, we' find. that procedures and training provided by LILCO.do minimize the likelihood of operator error that"

~

could result in exceeding the qualified load for a dies'el generator.- Although.

. .wa can find compliance .with GDC 17-without this determination, we find that the-

- procedures-~and. training testified to by LILCO. give. additional assurance that

'thsre is. no undue risk.to therpublic healtlk and' safety. The NRC Staff witnesses

-; sxpressed 'a number of concerns, . but we'are not persuaded that these concerns i -

I

. s .

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

i l

]

l pr:clude this finding. Concerns voiced by the Staff in its testimony cannot be

- tsksn as conclu.sions of deficiencies or inadequacies; rather, as the Staff tes-timony reflects, they are, more accurately, specifications of matters as to L which the Staff needs more information so that it can complete its review with a l'

thorough understanding of LILCO's bases for development and use of the proce-l

durss as they relate to the qualified load. Indeed, as the Staff acknowledged, E

additional information might well eliminate concerns, see Tr. 28211 (Clifford).

Thus,- LILCO's testimony, coming as it does from a licensed Senior Reactor Opera-tor and individuals knowledgeable about plant design and response, is sufficient to support our finding that LILCO's procedures and training adequately address the qualified load and render operator errors unlikely. Additional LILCO com-4

. mitments to the Staff can only improve this situation.

L-79. We agree with the Staff witnesses that an assessment of the adequa-cy of the emergency operating procedures as they relate to the. qualified load rsquires an understanding of plant operation and plant response in accident and transient conditions. See Tr. 28210-12 (Buzy, Clifford, Eckenrode). In early January 1985, the Staff commenced a review of procedures and training relating to the qualified load, which included a brief site visit. In' the time available prior to the hearings in February 1985, Staff witnesses were unable to obtain all the information necessary to understand the details of plant performance and plant response and the role of procedures and training. Clifford et al. , ff.

Tr. 27732, at 7-8; Tr. 27710-12, 27895 '(Buzy, 'Clifford, Eckenrode); Tr. 28219 (Clifford) . Given the time available and subsequent revisions by LILCO, the Staff reviewed some procedures only preliminarily and others not at all. See Tr. 27841-42, 28062-69 (Clifford). .Thus, the Staff testimony prior to the sec-ond site visit (February 27-March 1) expressing concerns with respect :o proce-i duras and training was not based on complete information or understanding.

. . ~ .. . .- - .- . .. ..- - - - -

L L-80. As-~a result of the'need for more'information, the Staff sent LILCO

~

La rcquest for additional information on a number of matters as to which the Staff had a concern. Clifford et al. , ff. Tr. 27732, at 9. A number of these

! concerns were reviewed in the hearings. See, ea, Tr. '27822-23 (Buzy); Tr.

27877-80, 27917-18_ (Clifford, Hodges);. Tr. '28082-83, 28095-99 (Clifford); Tr.

27914-15 (Hodges); Tr. 28040-41, 28052-53 (Clifford); Tr. 28107-08 (Clifford);

, ; Tr. = 27901, 27905-06 (Clifford, Ecken rode) .

L-81. In~ connection ~ with the need' for further information,' the Staff vis-j ittd the site a second time during the period February'27 to March 1,.1985. The

,_ results of this second site visit 'are reflected in the Staff's testimony of

~

March' 5, 1985.~ This. testimony refledts that many' staff concerns have been re-solved. Sec, e.g. , Tr. 28288-92 (Clifford, Eckenrode). For example, most of ths. Staff's' specific: concerns regarding cautions were generally resolved with '

only a small number remaining to be resolved by 'the task analysis. . Tr. 28307-08 4

(Clifford) . Further, while the Staff had found in ~ the past that LILCO's overall training program was< adequate and appropriate, Tr. 27822-23, -28108 (Buzy), the -

Staff had not had an adequate opportunity to. review LILCO's' revised lesson plans

~

that addressed the qualified load until the second. site _ visit during the week of~

Fcbruary 27. - As a result of this further? review,"the Staff' expressed satisfac-4 tion with LILCO's approach to training _with respect'to .the qualified load and

- natsd that the classroom exercises ' implemented by_ LILCO were extremely well; 1

~

1 structu red. Tr. 28298-99 ~ (Buzy). ' Based on^ the results of this: subsequent re :

J visw, the Staff concurred with LILCO that the training program ' adequatelyad -

~ drassed .the.3300-.KW qualified load. Tr. 28299, 28388 (Buzy). Simila rly,, the ,

. ' Staff originally; expressed a concern regarding restriction of the operators' p..

fi2xibilitylto' utilize loads in accordance withf procedures, but following the sscond site visit, Staff witness Clifford agreed operators were ableito take the

~

. tctions they were ~expecteditoitakelto operate the, plant within _its design and '

I ,

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

avoid loading the EDGs above the qualified load. Tr. 28290-96 (Clifford). ;

Thus, the Staff.also concluded that operators' flexibility to utilize loads in accordance with procedures was not as restricted as they had originally thought.

Tr. 28311, 28356-62 (Clifford).

L-82. During the. Staff's second site visit, LILCO presented a program for a job task analysis pertaining to the qualified load to be performed by an out-sida consultant. The Staff has reviewed the proposed job task analysis program and believes it is appropriate and complete to resolve any remaining concerns.

Tha Staff also believes that LILCO and the contractor are qualified to perform tha job task analysis. Tr. 28290-92, 28297 (Clifford, Eckenrode). Staff wit-nssses Clifford and Buzy support LILCO's conclusion that the operators can oper-ata the plant and maintain all safety functions within the design of the plant and the qualified load, but believe that the results of the task analysis are nasded to confirm this conclusion. Tr. 28295-96 (Clifford, Buzy). Staff wit-ncss Clifford believes the task analysis is appropriately considered as con-firmatory. Tr. 28315 (Clifford). The Staff witnesse's do not believe resolution of procedures and training to the Staff's satisfaction is in any way precluded.

Sn Tr. 28295-97 (Clifford, Buzy, Eckenrode). Accordingly, in light of LILCO's tsstimony on the adequacy of procedures and training to minimize and correct er-rors, and because the task analysis is not required to show compliance with GDC 17, the completion of the task analysis is not required for resolution of this contention . Thus, we agree.with Mr. Clifford that the task analysis and _resolu-tion of the details of procedures are appropriately considered as confirmatory.

- Wa are further confirmed in this view by the Board's experience which has demon-strited that litigation is not well suited for the review and refinement of pro-c:du res .7/

7/ As these findings reflect, LILCO believes GDC 17. compliance is established ind: pendent of the procedures. The-GDC 17 determination can be made on the

~

(footnote continued)

(footnote continued) 4

. basis of. design alone .because design need not accommodate operator error on .a par diesel b' asis. See Findings L-54, -55. .

Based on the present record, there-fora, the Board can make the requisite findings to support the issuance of a li-canse notwithstanding that. the task analysis, designed to resolve remaining

~

Staff procedures concerns, is not yet completed. LILCO submits that the task analysis is unnecessary for resolution of the contention. At most, it is a con-firmatory item appropri_ately left to the Staff for post-hearing resolution.

This conclusion is. supported, inter alia, by the following:

(i) The Staff and LILCO testimony that the qualified load encom-passes all loads required for performance of safety func-tions, see Findings L-3 to--35;

' (ii) LILCO's well-founded testimony that its procedures and training a'dequately address the qualified load and render; unlikely operator errors that'would cause the qualified load to be exceeded, see Findings L-63 to -78;. see also Finding L-56;-

(iii) LILCO's commitment to install a distinctive visual an'd audi-ble alarm which, together with procedures, and training,-.will ensure that unlikely. errors that cause the qualified load to be exceeded will be-promptly remedied, see Finding L-59; (iv) - The Staff's testimony, ; uncontested by SC, th'at the task analysis is the appropriate confirmatory means of resolving any '. remaining Staff procedures concerns-, see Findings L-77,

-82; and

) (v) The Staff's testimony, also unco . tested by SC, that -LILCO's procedures, which are typical BWR emergency operating proce-dures, are in some~ aspects already adequate to address the qualified load, .see Findings L-67, -81,. and in others, they; are susceptible of. revision to address the qualified, load to

~

L

! ~t he Staff's satisfaction, - see Finding L-82.

l LILCO ' recognizes that post-headng resolution o'f contested issues -should be sparingly employed. - But where, as 'here. : there is a clear course of action. the

, ' Staff can be directed to follow, i.e., to monitor. and confirm successful task '

" antlysis completion, post-hearing. resolution' cloes not constitute 'a delegation of-

dIcisional authority. - See Long Island Lighting Co. (Shoreham Nuclear' Power l Sta-tien,- Unit 1), LBP-83-57,18 NRC 445, 519. (1983). .Given the confirmatory nature -

of the task analysis' and. (i) through. (v) above, this case is readily disti.n -

guishable from those instances' where post-hearing resolution has been held:inap-

'. p rop riate.- . For example, sin Commonwealth Edison Co. (Byron Nucleari Power Sta--

tion,. Units L1 r, 2),EALAB-770,19 NRC 1163_ (1984), the -Appeal Board held ineppropriate the delegation to the Staff of the efficacy and outcome of ;remedi-al QA programs central to 'a finding ~of' reasonable' assurance of proper construc-

, . tien. There, unlike* here,: the requisite finding _of reasonable assurance could L , -not bef made 'on the-existing record; the efficacy 'and outcome .of the remedial-program .was central to such La- finding. - By contrast,~ the. task analysis here isi

~

,- r- -m 7- ' %-.-., e W - , - e ,-mm. .-m. -,---

l

-VI. Conclusions L;83. The qualified load is an acceptable licensing basis. Set at 3300 KW, the qualified lo'ad accommodates the loads the diesel generators must supply to perform their safety function in accordance with GDC 17. Findings L-1 to -4,

-34.

L-84. . Intermittent and cyclic loads are accommodated within the qualified lord. . They . represent small, short duration loads which can be conservatively

< predicted to result in exceeding the qualified load by no more than 31.4 KW for

]

- only. one EDG and for less than three minutes. Conservatisms associated with the MESL used to establish the qualified load, as demonstrated by the IET; n. ore than I

offset the 31.4 KW such that the qualified load of 3300 KW will not be exceeded.

Findings L-5 to -35. . ,

]

L-85. Diesel load meter error has been adequately considered in -estab-lishing the qualified -load. Calibration of the EDG load meters will be required by Technical Specifications and is controlled by procedure. The load meters;in-

~

troduce minimal- error, which is as likely to be low .as high. Instrument ' error.

did not result in an inadequate confirmatory test to establish the qualified

~

4 load- the indicated KW load values can b~e taken reliably as mean -values and,

! . givea-the substantial hours.of operation above 3300 KW, any potential remaining csrtainly not central to al finding of . reasonable assurance. The nature of the

- issues and their suitability for on-the-record litigation 'suggest-another dis -

' tinction. ' Resolution of ' task analysis and procedures issues, .unlike issues of dnsign changes'.and QA . remedial programs, would not likely be. aided by further proceedings and cross-examination. See = Southern California Edison Co. (San-l - Onofre -Station, Unis 2 - &- 3), - LBP-82-39,:15 NRC 1163,1216 (1982); see also

- Louisiana Power & Light Co. (Waterford,.~ Unit 3), ALAB 732, -17 NRC 1076,1106-07

(1983)-(emergency plan implementing procedures not intended for litigation, not -

rsquired for reasonable assurance). ,

For all ti.ese reasons,;the~ Board can and should resolve this contention in-i- dIpendently of the task l analysis and specific procedures, in any event, the

- Beard should leave the ta'sk analysis and procedures to the Staff for post-hearing resolution, with instructions to report to-the Board and th'e ~ parties'on tho task analysis;and its results. '.Should the results'be at odds with these findings, further' proceedings may be appropriate. c Absent this, however, further ,

litigation' on procedures 'is unnecessary and unproductive.

w J- - e , =ar- + w- e.r*m- y.sw ww - 'w y y

l l~

Grror is not significant. The qualified load is not inadequate due to instru-l m:nt error: the diesel generators will not be operated for long periods at 3300

, KW, the instrument provides a reliable mean value, and the diesel generators have demonstrated capability to support any additional load above 3300 KW at-tributable to instrument error. Findings L-36 to -46.

L-86. The' 100 KW tolerance band for testing is a practical necessity.

Rzcords of the endurance run demonstrate that a significant majority of the en-durance run was conducted at an indicated mean value of 3300 KW. To the extent that it was necessary to deviate from this indicated mean value, the endurance run records demonstrate that hours of operation in excess of 3300 KW far exceed those below 3300 KW. The performance of the endurance run also demonstrates that the 1100 KW tolerance band will not result in excessive loading of the EDGs during future surveillance testing because, despite the tolerance band, the vast majority of recorded loads were at 3300 KW. Findings L-47 to -53.

L-87. Individual diesel generators are not required by regulation (GDC

17) to be sized or qualified to accommodate, in addition to the design load, the worst case single load which could be placed on the EDG by operator error. To require qualification of the diesels-at a load to include the single worst case opsrator error load would effectively require the Shoreham design to accommodate ar. additional error beyond the single failure required by regulation. Findings L-54, -55. ;

L-88. Operator error has been demonstrated to be unlikely. Additionally, loads attributable to operator error are unlikely to. be 'as high as conserva-tively predicted. Even if such ' operator error were to . occur, EDGs have suffi-cisnt capacity and capability, as demonstrated by prior operation,. inspection and analyses, to support additional operator error loads without failure. Fi- l n:lly, -should a diesel . generator fail for any reason, including ' operator error, the onsite emergency power system meets the single failure criterion because the l l

j i

I l

~

remaining two EDGs provide power to sufficient equipment, within the qualified lord, to perform all necessary safety functions. Findings L-56 to -58.

L-89. GDC 17 compliance in this instance is demonstrated by the design of tho plant and the qualification testing of the EDGs at and above 3300 KW. This d:tsrmination is independent of procedures and training related to the qualified l

l lo-d. Thus, resolution of the Load Contention does not depend upon resolution

of any issues related to procedures. Findings L-61 to -63.

L-90. The procedures and training related to the qualified load do pro-vida additional assurance beyond the adequate assurance provided by the design.

LILCO's witnesses knowledgeably explained the role of the procedures and training in operation with the qualified load. They minimize the potential that tho. qualified -load will be exceeded, and they give assurance that prompt steps will be taken to reduce the load if it is exceeded. Findings L-64 to -78.

L-91. The confirmatory task analysis undertaken by LILCO will resolve any 4 rcmnining Staff concerns and confirm that LILCO's training and procedures pro-vida yet additional assurance that the qualified load will not be exceeded and that prompt remedial steps will be taken if it is. Findings L-79 to -82.

L-92. The qualified load is the only " rating" required to satisfy GDC 17.

Lord excursions above 3300 KW are not anticipated or required to perform the re-quirtd safety functions. To the extent 3300 KW may be exceeded during surveil-

, Irnca testing, the testing demonstrates the capacity. and capability of the die-scl generator. To the extent 3300 KW might be exceeded due to instrument error, the additional load is minimal and the diesel generator capacity and capability have been adequately demonstrated by prior operation, inspection, .and analysis.

Future testing above 3300 KW is unwarranted.

1. Operating Experience of the Shoreham EDGs B-1. in the spring of 1984, the DRQR program inspections identified liga-

! msnt cracks in the blocks of all three EDGs, and stud-to-stud cracks and one l

l - stud-to-end crack in the original EDG 103 block.1/ McCarthy et al. , ff. Tr.

c

-24372, at 13-15; Tr. 24,603-04 (Schuster). The location and depth of the liga-

msnt cracks was measured using a series of liquid penetrant, eddy current, and visual inspections of the block tops, stod holes, and cylinder liner landings.

McCarthy et al. , ff. Tr. 24372, at 13. No ligament crack in EDG 101 and 102 ex-tsnded to a depth below 1.5 inches nor onto the liner landing. Id. at 14-15; LILCO Exs. B B-17. As of March 11, 1984, the original EDG 103 block had no ligament cracks deeper than 1.5 inches, and the deepest stud-to-stud crack, which was between cylinders no. 4 and 5, was measured by eddy current to have a

'dtpth of 1.4 to 1.6 inches. McCarthy et al. , ff. Tr. 24372, at 14-15; Tr.

28823-24 (Johnson); Tr. 28825-27 (Rau).

B-2. On April 14, 1984, the EDG 103 block experienced an abnormal load excursion for approximately 25 seconds. The engine was operating with the fuel rack set at 3500 KW when the power demand from the site load was accidentally picksd up. The engine speed slowed until the output breaker tripped due to low-sngine rpm; the diesel continued to run at no load for an additional 10 minutes before it was shut down. McCarthy et al. , ff. Tr. 24372, at 17-18; Tr.

24,655-61 (Youngling, Seaman). The engine was later restarted and the qualifi-cation testing continued at 3900 KW for about 1.75' hours, at which time the en-gina was shut down due to a crack at cylinder no.1. 'When the engine was shut d:wn, it was operating satisfactorily and producing power. McCarthy et al.,. ff.

Tr. 24372,- at 17-18; Tr. 2M34 (McCarthy); Tr. 24661.1 (Youngling).

1/ .The term " ligament crack" refers to a crack that extends from the cylinder h:ad stud counterbore to the cylinder' liner counterbore and lies in a vertical plane. The ' term " stud-to-stud crack" refers to a crack between two adjacent stud holes'. " Stud-to-end cracks" are similar to stud-to-stud cracks, but they -

cxt;nd from a cylinder head stud counterbore to the end of the block. See McCarthy et al. , ff. Tr. 24372, at 14-15.

l B-3. After shutdown of the engine on April 14, 1984, inspection of the l; EDG: 103 block revealed that .the deepest stud-to-stud crack, located between cyl- l

~

. indsrs no. 4 Jand 5, had extended from a depth of .1.4 to 1.6 inches to a maximum dzpth of 3 inches. McCarthy et al., ff. Tr. 24372, at 18; Tr. 28823-24

-(Johnson); 'Tr. 28905-06 (Rau); LILCO Exs. B-18, B-25. Between March 11 and I . - .. .

l April 14,1984, additional ligament.and stud-to-stud cracks had initiated and l

' propagated at other block top locations; however, none of the ligament cracks

- extsnded onto. the. liner. landing. McCarthy et al., ff. Tr. 24372, at 18-19; Tr.

, 25538 (Johnson); LILCO Exs. B B-18, B-25. The EDG '103 block was replaced by a new block', which was installed in June 1984. Johnson et al. , ff. Tr.

-28799, at 5.

. -B-4. In' September 1984, destructive sectioning,. magnetic particle, and !

ultrasonic examinations revealed the presence of shallow circumferential cracks l'

.in the' eriginal 'EDG .103 block. McCarthy et al. (Supp.), ff. Tr. 24372, at 2, i

11; Anderson et al., ff. Tr. 25565,~ at 10-11; SC Ex. S-10. These cracks were located in the; sharp corner formed by the cylinder liner counterbore and the cylinder liner landing. They extended- at about a 45 angle from the corner to a 3 maximum depth of 3/8 inch. McCarthy et al. (Supp.), ff. Tr. 24372, at 2,11.

B-5. As of September -22,1984, the EDG 101 and 102 blocks. had each accu-mulated more than 1200 hours0.0139 days 0.333 hours 0.00198 weeks 4.566e-4 months of operation. .On the EDG 101. block', about '440

~

hours were at or above full load (3500 KW), including 25 hours2.893519e-4 days 0.00694 hours 4.133598e-5 weeks 9.5125e-6 months at or above 110%;

L of full load. -Tr. 28887 (Rau); .LILCO Ex. B-13. On :the EDG' 102: block,- about 475

. hours were at or above full load (3500 ;KW), including 30 hours3.472222e-4 days 0.00833 hours 4.960317e-5 weeks 1.1415e-5 months at or' above 110%

Tr. ; 28887-88 (Rau); LlLCO - Ex. B-14.

cf full load. The original EDG 103 block also accumulated.more than 1200' hours of operation, of which about 428 hour0.00495 days 0.119 hours 7.07672e-4 weeks 1.62854e-4 months si

-wara 'at or above full load (3500 KW), including 30 hours3.472222e-4 days 0.00833 hours 4.960317e-5 weeks 1.1415e-5 months at or above 110% of -

full load. -LILCO Ex. B-15.

~

-ll. Metallurgicai Analysis and Testing Demonstrate That the EDG 101, 102, . and 103 Replacement Blocks Have Superior Properties to the Original EDG 103 Block B-6. 'The. original Shoreham blocks were specified to be ASTM A48-64 Class l

40 gray cast iron. McCarthy et al. , ff. Tr. 24372, at 9-10. The replacement

(:

block was. specified to be Class 45 gray cast iron. Id. at 69. Extensive metal-

.lurgical analyses performed on these cylinder blocks demonstrated significant

, - - microstructural . differences between the original EDG 103-block on the one hand

! and the EDG 101,102, an'd 103 replacement blocks on the other hand. These dif-

, . forances reflect that the original 103 block had extensive Widmanstaetten graph-

-ita2/ that -adversely affects mech'anical properties. McCarthy et al. , ff. Tr.

~

24372, at 29-31, 35-41; Tr. 24731-32, 24738-39, 24941-49 (Rau, Wachob); Tr.

l 24766, 24769-70, 24951-52 (Wachob); Tr. 24760, 25346-47.(Rau); see LILCO Exs.

i j - B-33, B B-37.

i B-7. Two different metallurgical techniques were used to analyze.the .

bloc ks . First, metal samples were removed from the EDG 101, 102, original.103, and replacement 103 block tops. These samples were then metallographically pol-ishsd and examined under a microscope to evaluate their microstructure. Second, plastic replicas were taken of polished- surfaces of .the EDG 101,102, and origi-nal 103 blocks. Both-these techniques revealed extensive quantities of'

, - Widmanstaetten graphite throughout the original EDG 103 block and typical class

. '40 gray cast iron microstructure throughout the 101, 102, and replacement 103 blocks. - McCarthy et al. , ff. Tr. :.24372, at 29-31, 41-42; . Tr. 24741, 24746, 24752-55 (Rau);-. Tr. 24748-54, 24756-57, 24769-71 (Wachob); LILCO Exs. B ~

[ B-38.

4 2/ . Widmanstaetten graphite is a degenerate form of graphite that occurs.infre-qu ntly in _ heavy-section gray cast iron. A combination of very slow cooling rctrand tramp elements can combine ~to form Widmanstaetten graphite. McCarthy ct al., ff. Tr. 24372, at 30-31: Tr. 24745, 25010 (Wachob); Tr. 25059-60 (Rau, j,. -Wrchob);; Tr. 25064_ (Rau); see Tr.- 25870-71 (Bush).

E

- , , =.. . ,- .m.. . . __ - . , - . - . . , - . .

-B-8. The metal samples tested were cut from identical sites on each of ths ' EDG 101', 102, and original 103 block tops: the block _ top corners adjacent to cylinders no. 4 and 5 on the exhaust side and the crotch between cylinders no. 4 l

l .and 5 on the exhaust. side. Tr. 24738-39, 24941-44, 26,651-52 (Wachob, Rau); Tr.

24951 (Wachob). On the replacement 103 block, one metal sample was taken from an identical _ site' adjacent. to cylinders no. 4 and 5. Tr. 24951 (Wachob). Va ri-j; -ous 'metallog'raphic' preparation procedures were employed to examine the samples, and the results were evaluated and compared to assure that the observed -

microstructure had not beeir affected by artifacts produced by the polishing pro-cad u'res . Tr. -24947 (Wachob); Tr. 24948-49 (Rau). The. samples, and all of the

'approximately 10 replicac evaluated from the EDG 101 and 102 blocks, showed typ-ical gray _ cast iron microstructure.3/ McCarthy et al., ff. Tr. 24372, at 41;

-Tr. 24749, 28830 (Wachob); Tr. 24771 (Rau); Tr. 24945-48 (Wachob, Rau).

B-9. LILCO and the Staff agree that the samples andireplicas taken from -

4

.tha EDG 101 and '102 blocks provide .a representative sample for determining that extsnsive Widmanstaetten graphite is not present.4/ Tr. 25063-65-(Rau); Tr.

l 26651-53 (Wachob, Rau); Tr. 26287-88 (Bush). 'At least two factors support this conclusion. First, the formation of Widmanstaetten graphite is influenced by

, tha rate of cooling which is virtually uniform throu'ghout the heavy-section por-l

. tions of a large' casting sucn as the blocks.

Thus, the microstructure in one l 1

block top location Lwould be representative of the microstructure throughout the- 1

' block top. Second, the extensive additional metallography a~nd mechanical l

3/ Although small, isolated locations in the EDG 102-b!ock contain some u nconfi rmed . Widma n staetten microstru ctu ral featu res . The areas represent such a E

sm:ll fraction of the cell wall in that le.ation and a negligible fraction of eths cell walls in the structure; that' they have no significant impact on' mechani- I

-cal properties. Tr. - 24755, . 26657 -(Rau) l i . .

l

- 4/L Although Dr. Bush would_'have preferred to see' additional metallurgical site L

cvaluation,' he agreed that-there is a _ very definite difference in the micr::structur_e of EDGs:101 and 102 and the~ original EDG 103 microstructure; Tr.

26287-88 : ( B u sh) . .

,_, ,.,;.,m, -4 _a.&---= - - , . - W - **'*a " '*

'tssting performed on the original EDG 103 block confirmed that, at a range of d pths beneath the block top, extensive Widmanstaetten graphite was present. I I

- Thus, each location sampled, including the identical locations sampled in the EDG 101 and 102 blocks, confirmed that the sample locations were representative

of the microstructure of- the entire block. McCarthy et al. , ff. Tr. 24372, at 32; Tr. - 25063-65 (Rau); Tr. 24743-45, 26651-53 (Wachob, Rau); Tr. 24745-47,

, - 24949-50 -(Rau, .Wachob); see LILCO Ex. B-39; see also Tr. 24612-15 (Wachob).

B-10. SC witness Anderson asserted that' FaAA's sampling technique did not- .

- provide sufficient evidence that all portions of the EDG 101 and 102 block tops have typical gray. cast iron microstructure. Anderson et al., ff. Tr. 25564, at 1

171; Anderson et al. (Rebut), ff. Tr. 25564, at 1; Tr. 25552-53 (Anderson). He bassd that opinion, in part, on his belief that the material of each block is

not ' homogeneous . However, Dr. Anderson's opinion is entitled to little weight

- sinco he offered ~no independent metallographic evaluation of the Shoreham EDGs to rsfute FaAA's testing, which demonstrated that the. blocks have a ' virtually-

. uniform microstructure. Also unpersuasive is Dr. Anderson's testimony that the

samples are not reliable because -they-are not.a significant portion by weight of

' the entire block. As L'lLCO and Staff witnesses agreed, reliability is assured

- by sample location, not sample weight. Tr. 24756-57 (Rau); Tr. 24745-46 (Rau, Wachob); Tr. 26651-53 (Rau, Wachob). Compare Anderson et al. (Rebut), ff. Tr.

26326, at I with Tr. - 26032-33,- 26287-88 (Bush); Tr. 26651-53 (Wachob, Rau). In

- fact, Dr.- Anderson subsequently agreed that sample location is a more important-factor than- the sample ' weight. Tr. 26649-51:(Anderson).

- B-11. It is well recognized, and confirmed by -tensile tests performed on

, i tbs original EDG -103 block, that the presence of Widmanstaetten graphite signif-ic ntly degraded the mechanical properties of the original EDG'103 block com-4

parcd .to typical Class 40 gray-cast -iron. 'McCarthy et al. , ff. Tr. 24372, at-

31,- 33-35;; Tr. '25674, 25702. ( Anderson); Tr. 25781, 25807 (Bush); Tr. 24746

. - . .- -.-- .. . -. . - = - .

s

, L(Wachob); Tr. 24747, 24760-61, 24776-77 (Rau); LILCO Exs. B-40, B-42, B-44. As a-result, the original EDG 103 block was significantly more susceptible to fa-tigua crack initiation and propagation. Tr. 25317, 25346-47 (Rau). Tensile

i strength tests on specimens removed from the original EDG 103 block top regis-

tarad as low as 14.5 ksi, or about 40% less than the anticipated minimum value

.of 25 ksi for typical Class 40 gray. cast iron of this thickness. McCarthy et ab, ff. Tr. ~ 24372,- at 35-36, 41; Tr. 25552-53 ( Anderson); LILCO Ex. B-40. Fa-i tigua initiation testing .showed that the presence of Widmanstaetten graphite re-ducsd the fatigue life of the original EDG 103 by a factor of 10 to 1000 times.

McCarthy et al. , ff. Tr'. 24372, at 41-42. Fatigue crack propagation tests showed that the Widmanstaetten graphite' in the. original EDG 103 block increased tha rate of fatigue crack propagation .b'y a factor of-10 to 100 times. McCarthy .

ct al. , ff. Tr. 24372, at 40; see- LILCO Ex. B-44.

B-12. -in contrast, the EDG 101 and 102 blocks exceed th~e requirements for i Class 40 gray-cast iron. Tr. 24769-71 (Wachob); Tr. 24642-43, 24652 (Rau). The i

B-bar test results demonstrate that the . ultimate tensile strengths -(UTS) of the EDG 101 and 102 blocks are 45 and 47 ksi, - respectively. Tr. 24766-67 (Wachob);

Tr. 24642-45, 25029-30 (Rau); McCarthy et al., ff. Tr. 24372, at 10, 35-37; LILCO Ex. B-12. Since"metallographic testing of the EDG 101 and 102 blocks dem-onstrates that they .have a normal microstructure for Class L40 gray ' cast iron,

- and the B-bar tests exceed the minimum strength requirements for Class 40 gray 3 cast iron, the strength of the . blocks also exceeds the minimum requirements for 1

Class' 40 gray cast iron.5/ -Tr. 24642, 26652, 24770-72 (Rau) .

i 5/ - lIn' order to establish' the material properties of the blocks based on the .B-i Sar test results, it is necessary to verify that the block material:has a normal micrestructure because the cooling rate of the B-bars is-faster than the large l

' block casting due-to the difference in' size. Since; slow' cooling rates are .a >

principal factor in.. the formation of.Widmanstaetten ' graphite, B-bars may have n strsngth properties that are not representative.of the strength of the overall-l block material. Thus, .although the B-bar for the original EDG 103 block indi-l^ ^

(footnote continued).

- , - -, ..-,,a. :a .. 1., , . , . - _ , - -- . - .

B-13. The B-bar test for the EDG 103 replacement block indicated a UTS of 54 ksi, which is well in excess of the specified Class 45 requirement, and, in-dard, in excess of requirements for Class 50 gray cast iron. Tr. 24764-69 (Rau, Wachob). Since FaAA's metallographic testing confirmed that the replacement block has a normal microstructure, the strength of the replacement block exceeds E

the requirements for Class 45 gray cast iron. McCarthy et al. , ff. Tr. 24372, i at 36-38, 41-42, 69-70; Tr. 24767-69, 28849 (Rau); Tr. 24951-52 (Wachob); see LILCO Ex. B-42.

Ill. The Block Stress Analyses are Conservative B-14. .The primary. Ioadings that influence block cracking result from the stud preload, thermal stresses, and pressure stresses associated with cylinder

. firing during operation. To quantify these stresses, strain gauge measurements-wara made on the original EDG 103 block to evaluate the total stresses developed in the block top region. McCarthy et al. , ff. Tr. 24372, at 15-16, 22-23, 27; Tr. 24511 (Youngling); see also LILCO Exs. B-22, B-23. .

B-15. ^ The'. recorded strain gauge data were used to compute the-stresses at tha locations on the blocks where the gauges were placed and, in conjunction with finite element analyses, to compute the stresses present elsewhere in the block top. McCarthy et al. , ff. Tr. 24372, at 27-28; see LILCO Exs. B-22, B-26

. B-31; Tr. 24518 (Wells).

B-16. FaAA conducted two-dimensional and three-dimensional finite element strass analyses of the block top. The results of these analyses were used to

'dstsrmine scale factors that ' conservatively relat, ,e measured stress at-strain gauge no. -13; located between the cylinder heads in the stud-to-stud region, .to-(footnote continued) estrd a UTS of 42 ksi, the strength of the block material was severely' degraded dus to the presence of'the extensive Widmanstaetten graphite microstructure.

Tr. 24642-45, 24772-73, 25029, 25346 (Rau); Tr. 24745 (Wachob); McCarthy et al.,~ '

ff. Tr. 24372, at 36;'see also LILCO- Ex. B-40.

i l :

tho stresses at the edge of the stud holes where ligament and stud-to-stud cracks have been observed to initiate. McCarthy et al. , ff. Tr. 24372, at 42-44; Tr. 24650, 24724 (Rau); see LILCO Exs. B-22, B-27, B-30, B B-48.

B-17. Mechanisms of crack initiation in the block top were determined using strain gauge measurements and conservatively derived finite element analy-sis scale factors. Three mechanisms- of crack initiation were identified that cen act separately or in combination in the block top. They are (1) low cycle fatigue, associated with the stress range developed during start-up to high load Isvcis, (2) high frequency fatigue, associated with stress variations resulting from cylinder firing durin'g operation, and (3) overload rupture associated with tha highest tensile stress resulting from a combination of pressure, thermal, and preload stresses. McCarthy et al. , ff. Tr. 24372, at 44-45; Tr. 24690-95 (Wells, Rau).

B-18. To ascertain whether fatigue crack initiation was possible in blocks with minimum typical materials properties for Class 40 cast iron, the strcsses calculated from FaAA's conservative finite element analyses were plot-tsd on two modified Goodman (Smith) diagrams. See LILCO Exs. B-49, B-50. The Goodman diagrams predicted the possibility that stresses in the block top were sufficiently high for fatigue crack initiation to occur in the EDG 101 and 102 blocks. McCarthy et al., ff. Tr. 24372, at 45-46; Tr. 24648-51 (Rau).

B-19. The finite element analyses and materials properties used in the Goodman diagram analysis of fatigue crack initiation have been demonstrated by actual operating experience at Shoreham and other nuclear plants to be extremely c:nservative. McCarthy et al., ff. Tr. 24372, at 46-47; Tr. 24654 (McCarthy);

-Tr. 26291-92 (Bush). In addition, the scale factors based upon the results of tha conservative finite element analyses introduce further conservatism into the Goodman diagram analysis of possib'e crack initiation. Tr. 24640-41, 24649-50 (Reu); Tr. 29112-13 (Bush).

.g.

B-20. The Goodman diagrams are~ far too conservative and were not intended to be used to predict the specific load levels at which cracks would initiate.

Tr. 24649-50 (Rau); Tr. 24707-08 (McCarthy). Tf e conservatism is confirmed by the fact that ligament cracks have not occurred at all locations even in the original EDG 103 block with degraded properties. Tr. 24654 (McCarthy); Tr.

24649-50 (Rau); see LILCO Exs. B-16, B-17, B-25. Further conservatism is shown by the' fact that the Goodman diagrams indicate the possibility of stud-to-stud j' cracking in only a few loading cycles, yet stud-to-stud cracks have not initi-attd in the 101 or 102 blocks despite extensive high load service. Tr. 24648-51 (Rau); Tr. 26062, 26065-66, 26291-92 (Bush); Tr. 24654 (McCarthy); see LlLCO Exs. B-16, B-17.

-lV. ' Ligament Cracks Are Benign, B-21. Fa'AA's three-dimensional finite element analysis of the EDG' 101 and 102 block tops demonstrates that stresses in. the ligament region are highest at-tha corner formed by the block top and the stud hole. The stresses decrease i

rapidly with distance beneath the surface of the block top and become fully com-prassive. Tr. 24465-66, 28820, 24689 (Rau). Thus, the analysis demonstrates that Iigament cracks initiate, if at all, at the block top surface and propagate-1 down the stud hole toward the threads and across the ligament region- to the.cyl-L

-indar liner counterbore. McCarthy et al., ff. Tr. 24372, at .14; Tr. 24465, 24689 (Rau). As the cracks propagate downward, they slow down and arrest at.the

. cylinder liner landing. McCarthy et al. , . ff. - Tr. 24372, at 14, 58; see ' Tr.

c-l 24689 (Rau); see also LILCO Ex. B-19; Tr. 25930 (Berlinger). i B-22. Operating experience confirms the conclusions of FaAA's finite ele--

!- mint analysis. Testing on - EDG 102 confirmed that no discernible ligament crack prcpagation-occurre'd .when the engine was fast started 100 consecutive times,

-including three fast starts.to full load in less than 60 iecon' s:in d ac'cordance

, with FSAR requirements. Further, after more than- 1000 hours0.0116 days 0.278 hours 0.00165 weeks 3.805e-4 months of operation, .)

1 s 1 1

. . - ~ .. . . . .. -.- - . - . - -

' including more than 350 hours0.00405 days 0.0972 hours 5.787037e-4 weeks 1.33175e-4 months at or above 3500 KW, none of the ligament cracks

on the-EDG '101~ or: 102 blocks propagated onto the cylinder liner landing or ex- ,

i tsnded ; deeper than 1.5 inches 1.in the stud hole counterbore.!/ McCarthy et al.,

ff. !Tr. 24372, at l'4-15; Tr. 24404 -(Johnson); Tr. 24507-08 (Schuster, Wells);

[

Tr. 24499-501 (Youngling);. Tr. 28821 (Rau); see LILCO Exs. B B-14, B .

B-17; 'see also Tr. l24399-400- (Schuster, Johnson); Tr. 24505-06 (Youngling).

B-23. ~ Dr. : Bush reviewed FaAA's three-dimensional -finite element analysis. ,

i . .

+

' Ha was generally in agreement.with ~the analysis and the conclusion that stresses-in the ligament region decrease with distance from the block top and become com-fpressive, causing the ligament cracks to arrest. Tr. 25845,. 25854, 25880, j- (Bush); see also Tr.- 25853-54. (Berlinger);_ but see Tr. 26059 (Bush).

Although 1 .Dr. - Bush expressed some ' reservation th'at secondary thermal stresses 1were not i* ~

completely taken into account in the analysis, his reservation was limited to -

tha exa'ct point at:which stresses. became compressive ~ and d'id not affect his con-

~

r clusion-that ligament cracks move into a compressive stress field- and arrest.

4 Tr; ~ 25845-49 (Bush).

~

B-24. ~ SC's witnesses! testified that ligament cracks were not benign and. ~

thet they could.cause catastrophic engine failure. ' Anderson ~ et al. ,1ff. Tr.

25564, at .151-52,3154-55. They -assert- that ligament crack propagation might

} Lcause? coolant _ leakage.and subsequent EDG failure if a coolant leak depleted the coolant water reservoir. . . Anderson et al. , ff.M Tr. 25564, L at 152-53. SC's tes-

.timony' is speculative an'd entitled to no' weight because- SC's witnesses have 'no lkncwledge of the: stresses' in the block top. Theyl have- n'ot performed any finite ilemsnt analyses or' fracture mechanics analyses _of crack progression to g/.- No ligament crack's on the' original EDG 103.blockcextended onto the cylinder 1

1 L linsr landing but. onel crack' adjacent to the three ~ inch deep stud-to-stud crack -

.. bstween cylinders' no. 4 and 5 extended to'aidepth.of: 21/2 inches on the stud

' hcla _ side of the ligament. . McCarthy. et al., ff. Tr. 24372,Lat'14-15;iLILCO Ex.

E : B-25;E Tr. 25538 (Joh'nson).

e o.

a s ,

^

y , ,- y -

,e - w

d;tarmine whether the stresses are high enough to cause ligament cracks to prop-agste to the cooling water jacket. See Tr. 25619, 25625-26 (Anderson); Tr. i

- 25631-40 (Christensen, Bridenbaugh, Hubbard, Eley, Anderson); Tr. 26377-78 l

(Elsy, Bridenbaugh); Tr. 25630-31 (Stipulation re Christiansen, Eley). Fu rther,

-SC's witnesses lack a detailed knowledge of the geometry of the block top. See Tr. 26375 (Eley) (Eley does not know how deep a ligament crack must propagate to reach the cooling water jacket); see also Tr. 26359-60 (Anderson).

B-25. FaAA's analyses showing that ligament cracks move into a com-prsssive stress field and arrest, and the operation of the EDGs for more than 1000 hours0.0116 days 0.278 hours 0.00165 weeks 3.805e-4 months without the development of ligament cracks beyond the liner landing, support the conclusion that ligament cracks will not propagate to the cooling watar jacket. Tr. 25231 (Wells); Tr. 25238 (McCarthy); see also LILCO Exs. B-13

- B-18, B-23. However, even assuming ligament cracks continued to propagate, the cooling water system would not be depleted. First, any coolant water leak-

. ago would be minor and would not cause an operational proHem. Tr. 26055, 26187 (Hsnriksen); Tr. 25231-32 (Wells); Tr. 25238 (McCarthy); Tr. 25232 (Youngling);

su also Staff Exs. 9 - 10; Tr. 26376 (Eley). Second, the EDGs are equipped with a low level water alarm, which sounds after a loss of 20 gallons to warn tha operator of a low coolant leve!. Tr. 25232 (Youngling); Tr. 25272 (McCa rthy); ' Tr. 25490-91 (Youngling); Tr. 26376 (Eley). Third, virtually unlim-it:d makeup coolant water can be added to the engine during operation through a 1.5 inch water pipe capable of delivering 70 gallons per minute from storage tcnks having capacities of 100,000 and 600,000 gallons. Tr. 25272 (McCarthy);

Tr. 25492 (Youngling); Tr. 26188 (Henriksen).

B-26. As. SC admitted, leakage from a ' ligament crack would become signifi-

, cant only if it exceeded the makeup rate of the cooling water system. Tr. 26380 (Elry) . Since the availability of virtually unlimited amounts of makeup water at 70 gpm assures continued long-term EDG operation even if a substantial-leak rt

dsyslops, the evidence justifies' the conclusion that the cooling water system will not be depleted and that-ligament cracks will not affect reliable EDG op-

.eration . Tr. 26188 (Henriksen); Tr. 25232 (Youngling).

B-27. Additional assurance that ligament cracks do not pose any threat to

, rslitble operation is provided by LILCO's commitment to inspect the ligament re-gions of the EDG 101 and 102 blocks with liquid penetrant and eddy current by rrmoving two cylinder heads- from each engine at each of the first four refueling

" ou tages'. -See Attachment 1.

V. Stud-to'-Stud Cracks Are Not Likely to initiate and, if They Initiate, Will Not impair EDG Operation During An Emergency B-28. FaAA performed a cumulative damage analysis to determine whether a

) ~s tud-to-stud crack would propagate during a LOOP /LOCA to the point that it would impair the ability of the EDGs to perform their intended function.

~

j McCarthy et 6, ff. Tr. 24372, at 48-49. Cumulativo damage analysis is. an accepted method for quantifying the fatigue life of a component subjected-to variable loading i history. Cumulative damage analysir, quantifies the fatigue damage for a known j .btnchmark period and then compares the cumulative damage predicted for a postu-

! latId operating condition against the benchmark period to determine whether it is more or less severe than the benchmark. McCarthy e' al. , ff. L T r. 24372, at l '48-50; Tr. 24692-95 (Rau) .

l -B-29. The benchmark period for the Shoreham blocks is the load history and the response of the original EDG .103 block between March -11 and April .14,-

1984. : A fatigue damage index was calculated for the load history experienced

during the benchmark period. McCarthy et al., ff. Tr. 24372, at 49; Tr. l

~

24780-84 (Rau). -The fatigue damage index was then related to the amount of j stud-to-stud crack growth, which is. bounded % the growth between cylinders no.

i. . 4 and.5 from 1.5 to 3 inches during ~the benchn._ '< period. Tr. 24695 (Rau).

Using cumulative damage analysis, the fatigue dam.;., index for the amount of crack growth' during the. benchmark period is ther compm.ed to the fatigue damage 1

y , 4 .- . , , . , , . - . , - , , , - - , --,-,_.-w- ,

r :indsx .that'would be accumulated by other postulated load requirements, i.e., a

' LOOP /LOCA. Tr. 24694-95 (Rau); McCarthy et al., ff. Tr. 24372, at 49-52.

B-30. To analyze the suitability of the Shoreham blocks for n'uclear ser-

vico, FaAA initially computed the amount of cumulative damage required for a l postulated . LOOP /LOCA-involving maximum operation at .3900 KW and 3500 KW (the "3900/3500 ' KW LOOP /LOCA"). _ McCarthy et al. , ff. Tr. 24372, at 54; LILCO Ex.

l - B -51. FaAA's cum ~ulative damage analysis of the EDG 101 and 102 blocks demon-V strates that the fatigue damage index that would be accumulated on,these blocks

' during a postulated'3900/3500 KW LOOP /LOCA is less than 2% of the cumulative

damsge for-the benchmark period. McCarthy et al., ff. Tr. 24372, at 53-54; Tr.

24695, 25313-14 (Rau). Thus, the EDG 101 and 102 blocks can withstand 50 con-sacutive 3900/3500 KW LOOP /LOCAs befcre accumulating the same amount of fatigue j: crack growth experienced by the original EDG 103 block without any affect on en-gins operability durin's the benchmark period. Tr. 25313-14, 28810 (Rau). Sig-i nificantly, this conclusion is based on an analysis that begins with the conser-e vative assumption of a 1.5 inch deep stud-to-stud crack in the EDG 101 and 102

' blocks . Tr. 25316, 28852 (Rau).

. B-31. The actual load requirements of the EDGs for a LOOP /LOCA:are lower i

than those on which FaAA's original' cumulative damage calculations.were based.

LILCO has demonstrated that a qualified load of _3300 KW ' bounds .thel load that any

, EDG must supply to perform its safety function. Thus, a conservatively postu-8

~

Istad LOOP /LOCA involves maximum operation at 3300 KW:(the " qualified load -

LOOP /LOCA") . Johnson et al. ,' ff. Tr. - 28799, at 12; Dawe 'et al. , -ff. Tr. 27153,

! = at 18; see also Findings L'-27, L-28.

[ 'B-32. Operation of the EDGs at the. qualified load of 3300 KW rather than -

st loads tof 3500 or 3900 KW will produce lower . stresses . in 'the ' block top. 'As a

rssult, the ,)ossibility of ligament, stud-to-stud and circumferential crack

. initistion is reduced and, -even.if crack initiation occurs,. the rate of any

cr:ck propagation will be slower. Johnson et al., :ff. Tr. 28799,. at 10-11; Tr.

I

! 28902-05 (Rau).

l B -33. . Cumulative damage analysis has also been performed for the quali-fiad I'oad LOOP /LOCA .which . indicates that it w'ill produce less fatigue damage L '

l ' than the 3900/3500 KW LOOP /LOCA previously analyzed. Johnson et al., ff. Tr.

28799, at 10-13. A crack would require 20% more time at 3300 -KW than at 3500 KW

{ .to propagate an equal amount. Tr. 28904-05 (Rau) . Further, at 3300 KW, the

. crack propagation rates are 3.5 times slower than at 3900 KW. Tr. 28904-05 -

(Rau). Thus, there is 'even greater demonstrated margin for the qualified load LOOP /LOCA th'an for the conservative 3900/3500 KW LOOP /LOCA, which has a demon-j strated margin of 50 consecutive LOOP /LOCAs. Tr. 25313-14, 28810 (Rau); Johnson at al. , ff.- Tr. 28799, at 12-13.

l B-34. The conservatisms in.the cumulative damage analysis are such that the actual margin is more than 50 consecutive LOOP /LOCAs. These conservatisms include the following:

(A) The analysis does not take credit for the additional cumu-j lative damage' necessary to initiate a stud-to-stud crack or i to cause the crack to propagate from initiation to a depth

.of 1.5 inches' (Tr. 25316-17, 28876-77 (Rau); Tr. 29092 (Bush));

(B) The superior fatigue resistance of the EDG 101 and 102

! blocks reduces the possibility of stud-to-stud crack

!' initiation - (Tr.- 25317 (Rau); Tr. 29092, 29129,-30 (Bush);

q LILCO. Ex. B-42);

'(C) The EDG 101 and 102. blocks - have - not initiated stud-to-stud e cracks despite more than 400 hours0.00463 days 0.111 hours 6.613757e-4 weeks 1.522e-4 months of operation at or above -

l 3500 KW (Tr.4 28885-88 (Rau); LILCO Exs. B B-14));

i _( D) The' analysis does not take credit for the additional cumu-lative damage necessary for crack growth beyond 3 inches; greater. crack depth' would be req ~uired to impair EDG op-eration (Tr. 25318 (Rau); Tr.~ 25234-37-'(Wells); Tr.

25237.-38 (McCarthy));

-(E) . ' The analysis attributes all crack ~ growth curing the benchmark period to fatigue and does not take credit for any rapid crack propagation that'might have occurred during t

the unusual load excursion -(Tr. 29076-7G '(Bush));

i i

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

i (F) Actual LOOP /LOCA loads, as demonstrated by the IET, are less than the 3300 KW qualified load, and substantially below those loads assumed in the 3900/3500 KW LOOP /LOCA cu-mulative damage analysis (McCarthy et al., ff. Tr. 24372, at 54; Johnson et al., ff. Tr. 28799, at 12-13; Load Find-ings L L-30, L L-34);

(G) The actual load sequence during the benchmark period did not result in crack growth retardation whereas higher ini-tial loads during a LOOP /LOCA' would retard crack propaga-tion (Tr. 28831-33, 28897-99 (R'au'));

(H) The model is not " limited," so operation at all loads is assumed to contribute to crack propagation when in fact. low toads may not contribute to' crack growth because they are below the fatigue threshold (McCarthy et i., ff. Tr.

24372, at 57-58; Tr. 25324-25 (Rau)); and (1) The analysis dces not take credit for the lower stresses av and reduced probability of stud-to-stud crack initiation which exist in the absence of ligament cracks (Tr. 29090, 29093 (Bush); LILCO Exs. B B-50). '

r ap B -35. Although Dr. Bush would have performed the analysis differently, he .f agreed that FaAA's cumulative damage methodology was conservative. Tr. 26223, 26313, 29077-78, 29094-95 (Bush) . Dr. Bush believes the unusual load excursion may have been a major contributor to crack initiation and propagation. Tr.

29041, 29043-44, 29046, 29076-77 (Bush). FaAA's analysis conservatively assumed all cracking during the benchmark period was due to fatigue, and did not take credit for any rapid crack growth during the unusual load excursion. Conse-quently, Dr. Bush believes FaAA's analysis predicts artificially high crack growth rates, and therefore provides a conservative bound on crack propagation '

rates. Tr. 29075-78 (Bush) .

B -36. On the basis of his own semi-quantitative analysis, Dr. Bush agreed .

with FaAA's conclusion that the blocks are qualified for nuclear service. Tr. ..

25861, 29090, 29116-17 (Bush); Bush, Staff Ex.14, at 24-26. Dr. B u s h's . analy-sis is based in part on the operation of the EDG 101 and 102 blocks for at least '

n 400 hours0.00463 days 0.111 hours 6.613757e-4 weeks 1.522e-4 months at or above 3300 KW, which is the equivalent of 5 x 10E6 loading cy- .

! cles, without stud-to-stud cracks having initiated.7/ In Dr. Bush's opinion, -

7/ Dr. Bush indicated that he reverted to an earlier date (apparently in the

r

-_ 3,#

s 4

this ooeration demonstrates that it is unlikely that stud-to-stud cracks will it. t eate in the EDG 101 and 102 blocks. Tr. 29J52-53, 29129-30 ( B ush) . Fu r-ther, even if cr .eks were to ir itiate, they would not propagate to the point of loss of tanction during a LOOP /LOCA. Bush, Staff Ex. 14, at 32; Tr. 29065-67 (8ushT.

B-37. SC asser ts that Fc AA's cumulative damage anab, sis was based on faulty pr etnises, inst fi%nt data, and error m us assumptions ' nderson et al t, SC's criticisms are enti':eu to no weight because if. Fr. 25564, at 166-70. :

SC's .vit.iesses performed no independent cumulati'.e Jamage analysis on the blocks

~~

e,,d Eave no experience performir.g cumulative damage c.a l y s es . 8/ Tr. 25637-39

( Anderson); Tr 25639 42 ( B ric'.- ibaugh , Ch ristens 3r. , Eley, kubbard). Fu rthe. , .'

SC's principal witness on cumulative damage nevr.r even reviewed Fa AA's cumula-tive a :rrage calculations. Tr. 25637 ( Andersen) . .

B G8. SC's criticisms of FaAA's cumulatis e damage analysis are ncorrect . -

and ref!ect a lack of un Jerstanding. LILC^ witnesses refuted each of SC's cri+ -

cisrm First, FaAA did consider the s'ur;-te-and crack from cylinder no. ! and ,

den.aastrated by cumulative damage analysir, that propagation of ch:s crack during a rustulated LOOP / LOC A would be less tF an tnat for a stud-to-stt._ crack. Tr.

24808, 24811-13 (Johnson, Rau). Socoad, the cu.culative damage ar.alysis 2s i i fact r,en iinea,- Tr. 25323-2/ (Rau) . T h . -d , by not limiting" the ana'ysis and i not taking c woit for i a :aticas ir crack growtii rates at various poi: 's in time . ,

due to load sequencing, FaAA actuahy increased the consers cism in it - 3

. / . .

\

A pr il 1964 tinie f reme) to obtair his inferrration showinc 100 hou rs i r <.geratio, ,

at 3300 kW .t s : f Mpten?ha,- 22, ir l, the EDG IN a . 3 102 blockc had in f set #

acc er.iulate u s F ) ; Tc Rd86-88 (Rau); LiLCO Exs. B-13 t-

- B - 14.

~

3/ SC witnes ses Anderson a .d Bride nbaugh claimed they wve ciuslified to res :aw the results of cumuja ive damage a'aly s s, Tr 26428 ( Anderson, Br:denbaugh), ~

but their ass.ertions are not supported ay any credible evidon e in the record.

. cumulative-damage analysis.

McCarthy et al. , ff. Tr. 24372, at 57-58; Tr.

25324-25, 28831-33, 28897-99 (Rau). Fourth, the cumulative damage model was not

-bsscd ~on inadequate crack propagation data. FaAA used accurate data obtained by

-dirret testing on the' original EDG 103 block and on Class 40 gray cast iron with a normal thick-section microstructure like that present in the EDG 101 and 102

blocks. Tr. 28828-30 (Rau, Wachob); Tr. 29071-73, 29118 (Bush): see LILCO Ex.

B-44. Fifth, FaAA did not rely upon imprecise crack measurements. The deepest crack at the beginning of the benchmark period was measured by eddy current to ba between 1.4 and 1.6 inches. Tr. 28823 (Johnson) . Although he originally tastified that the crack depth could have been shallower than 1.6 inches, Tr.

24792 (Johnson),9/ Dr. Johnson, a certified Level 111 NDE inspector, subse-quantly verified the accuracy of the measurement by determining that the eddy .

current signal was five times the reference level. The strength of the signal confirms that the nominal 1.5 inch measurement was not an overestimation of the actual crack depth caused by background noise from Widmanstaetten graphite. . Tr.

28823-24 (Johnson). The deepest crack after the load excursion was accurately determined to be 2.8 to 3 inches by destructive sectioning and four independent NDE techniques. Tr. 28825-27 (Rau); McCarthy et al. (Supp.), ff.. Tr. 24372, at

10. Thus, the measurements of the stud-to-stud crack used by FaAA for the bsnchmark period cumulative damage analysis were accurate. Sixth, it is not nscessary to identify when ligament or stud-to-stud cracks initiated because the cumulative damage analysis does not. take credit- for the time required for crack initiation. Rather, it begins with the conservative assumption that ligament 94 FaAA has testified that Widmanstaetten graphite can lead to overestimation cf crack depth if the background noise from the Widmanstaetten graphite ~ is mis-intsrpreted as an-extension of the crack. Tr. 24579, 24583- (Johnson).

Ovsrestimation of some crack depths due to Widmanstaetten graphite will not af-

- fset the cumulative damage analysis because there is no uncertainty in the nomi-nel 1.5' and 3.0 inch measurements, which as the largest crack measurements, were usid to compute the fatigue damage index. Tr. 24789-92, 24812-13 (Rau)'.

l l

I

- cracks and stud-to-stud cracks having a depth of 1.5 inches are present.

Tr.

28894, 28895-98, 28908-10, 28911 (Rau); Tr. 29074-75 (Bush).

B-39. SC also asserts .that FaAA should have performed a fracture mechan-ics analysis. Anderson 'et al., ff. Tr. 25564, at -170. In fact, FaAA's cumula- ;

'tivo damage analysis is a fracture mechanics analysis that conservatively bounds tha' rate of crack growth. . Since this analysis has demonstrated a significant margin, 50 consecutive 3900/3500 KW LOOP /LOCAs, it is not necessary to perform a mo'ra detailed fracture mechanics analysis merely to verify that the blocks will

, -in fact _ perform their intended function with an even larger margin. Tr. 24803,

, : 28818 (Rau).

B-40. The' EDG '101 and 102 blocks have operated at or above 3500 KW for more than 400 hours0.00463 days 0.111 hours 6.613757e-4 weeks 1.522e-4 months , which is more than 5 x 10E6 loading cycles, without devel-

oping stud-to-stud = cracks. This operation, combined with the superior fracture and fatigue properties of these blocks compared .to the original EDG' 103 block, support the conclusion .that stud-to-stud cracks are .unlikely to initiate in the cDG 101 and 102 blocks. McCarthy et al., ff. Tr. 24372, .at 60, 74; Johnson et ah, ff. Tr. 28799, at 12; Tr. 28810-11, 28884-88, . 28853-54 (Rau); Tr. 29052-53

_ (Bush); LILCO Exs. B-13 '- B-14, B-42; see also Tr. 29129 (Bush).

B-41. In the unlikely event that a stud-to-stud crack initiates, it will

! be detected by the visual and eddy current inspections performed on the EDG,101

and 102 block tops after any. operation over 1800 KW. - See Attachment 1.

McCarthy. et al. ,- ff. ITr. 24372,~ at 60-61; Bush, Staff Ex. -14 at.25; Tr._29098

~

(Bush) Tr. 25897-98 (Berlinger). _ If a crack-is L detected-, the engine ~ will. be removed from ' service and- the crack will be evaluated. .lf the crack is no more' then l'.5 inches . deep,4 the EDG remains acceptable for emergency standby service i

bicituse the cumulative -damage analysis has demonstrated a. margin _ of at least.50 i-consecutive: LOOP /LOCAs even assuming the existence; of a l5 inch ' deep crack. .

- McCarthy~. et al , ff.- Tr.- 124372,'at 71.

b a c n - ,

B-42. The Staff's position is that in addition to evaluating the stud-to-stud crack, further analysis should be conducted before the EDG is returned to sarvice. Bush. and Henriksen,- ff. Tr. 25775, at 29a-30. This position is not, howsver, supported by any analysis. Also, Dr. Bush acknowledges that FaAA's cu-mul:tive damage analysis provides a conservative bound on crack growth rates.

Sm Id.; Tr. 29076-78 (Bush) . Therefore, we conclude that the EDG can be re-turned to emergency standby service if a detected stud-to-stud crack is no duper than 1.5 inches because the margin demonstrated by FaAA's cumulative dam-aga analysis is maintained. Tr. 25313-14, 28890 '(Rau); McCarthy et al. , ff. Tr.

24372, at 53-54, 71.

B-43. Even if loads as high as 3900 KW were to occur for brief periods of tima during a postulated LOOP /LOCA, LILCO and the Staff concur that the EDGs' ability to perform adequately would not be impaired. Johnson et al., ff. Tr.

28799, at 10-13; Bush and Henriksen, ff. Tr. 28503, at 14, 21-23.10/ The EDGs' ability to perform their intended function at loads above 3300 KW has been shown

- by FaAA's cumulative damage analysis. The margin of 50 consecutive 3900/3500 KW.

LOOP /LOCAs involves- a total of 10 hours1.157407e-4 days 0.00278 hours 1.653439e-5 weeks 3.805e-6 months at 3900 KW and 40 hours4.62963e-4 days 0.0111 hours 6.613757e-5 weeks 1.522e-5 months at 3500 KW. See McCarthy et al., ff. Tr. 24372, at 54; Tr. 25313-14, 28890 (Rau). Further, the ability of the EDGs to run at loads up to 3900 KW has been demonstrated directly by operation of the EDG 101 and 102 blocks, at 110% iof full load, for more' than

. 25 hours2.893519e-4 days 0.00694 hours 4.133598e-5 weeks 9.5125e-6 months each without developing stud-to-stud cracks. Johnson et al., ff. Tr.

28799, at 12; LILCO Exs. B B-14. .Thus,. the EDGs have adequatch . demon-strated their ability to withstand loads over 3300 KW, including loads up to 3900 KW for more than an hour, without impairing the EDGs' ability to perform its ' intended function during a postulated LOOP /LOCA.

10/ The Staff's testimony refers specifically to the crankshaft,' but it' is im .

. plicit that no loss of- EDG function will occur-because of block cracking. See Bush,'. Staff Ex.14, at 32.

P' VI. .Circumferential Cracks Are~ Not Present and Will Not impair EDG

Operation if They Initiate I B-44. Liquid penetrant _ and' ultrasonic inspections performed on the EDG

1101 and 102 blocks demonstrate that.these blocks have no circumferential cracks.

Tr. ' 28815-16,' 28870-72 (Schuster); Tr. 28116-17 (Johnson); Tr. 28813 (Rau). Al-th'ough liquid penetrant inspections on the EDG 101 block revealed some back-

' ground indications, .these indications occurred as a result of liquid penetrant =

, collccting .in a carbon deposit that had not been completely removed. Tr. 24447 (Schuster); Tr. 24448-50 (Wells, Schuster). Evaluation of these indications i with an ultrasonic inspection technique -and reinspection by hquid penetrant 1

, aftar complete cleaning confirmed that there were no circumferential cracks in EDG 101. Tr. 28815-16 (Schuster, Johnson); Tr. 24447-48 (Schuster); Tr.

1 26692-93, 26871-72 (Rau).

B-45. Ultrasonic inspections are highly reliable for circumferential crack detection because they are not af_fected by deposits collecting in the cor-nar or on the cylinder liner counterbore. Tr. 24449-50 (Schuster); Tr.

} 26692-93, 26871-72 (Rau); Tr. 28816 (Johnson); Tr. 28872-73 (Schuster). Al-though ultrasonic inspection probes have a " blind spot" that prevents them from dstccting small cracks immediately.below the probe, LILCO's ultrasonic inspec-

. tion technique avoids the effect of. this blind spot:by examining- for circumfer - >

cntial cracks from below the liner landing' ledge. Tr. 28908 (Johnson); Tr.

26695, 26874 (Rau); Tr. 26874 (Bush). A crack, if present, would not be within tha ultrasonic probe's_ blind spot. Tr. 28908 '(Johnson) . Accordingly, the ul-

~ trasonic inspections reliably _-established.that circumferential cracks are not

'prssant in the' EDG 101 and 102- blocks.

B-46. Dr. Bush. te'stified that he had ~no assurance that the EDG 101 and L i

102 blocks did. not ha've circumferential cracks.~ Bush and Henriksen, ff. Tr.

- 25775, at 7-9; Tr. 26020,' 26155 (Bush). .'At the time of Lhis testimony, however, L

Dr. Bush misunde'rstood the procedure use'd by[LILCO in: performing its ultrasonic t .,-s.-- e yM. - , - - . .m v v , 4 , +,4 ,-

t2 sting. Tr. 26874-75 (Bush) . Dr. Bush later agreed that the ultrasonic proce-

. dura used by LILCO was technically feasible for detecting circumferential

- cracks . Id. Therefore, because ultrasonic inspections reliably detect circum-farsntial cracks and appropriate procedures were used, we can conclude there are no circumferential. cracks in the EDG 101 and 102 blocks.

B-47. SC expressed general concerns regarding the reliability of LILCO's inspections, but presented no evidence challenging the reliability of the spe-cific inspection procedures used by LILCO for detecting circumferential cracks in the EDG 101 and 102 blocks. See Anderson et al., ff. Tr. 25565, at 11-12,

14. SC's testimony is so general that is has no probative value and is entitled 4

to.no weight since the witnesses sponsoring the testimony are not certified NDE inspectors. See Tr. 25645 (Anderson, Hubbard).

B-48. Even if circumferential cracks were to develop in the EDG 101 and 102 blocks, they would not affect the suitability of the EDGs for nuclear stand-by service. Tr. 28813 (Rau); Tr. 26020 (Bush, Berlinge.'); Tr. 26023 '(Bush);

Bush, Staff Ex.14, at 25-26. FaAA conservatively assumed the presence of cir-cumferential cracks 360 around each cylinder, and analyzed these assumed cracks using the results of its finite element analyses. These analyses demonstrated that circunferential cracks would slow down, arrest, and therefore not impair EDG operation. McCarthy et al. (Supp.), ff. Tr. 24372, at 12-14; Tr. 28812-13 (Rau).

B-49. Specifically, FaAA's finite element analyses show the largest str:sses driving a circumferential crack would cause the crack to propagate from tht corner at approximately a 45 angle. Tr. 25100, 25343-45 (Rau). The analy-srs- further indicate that the stresses decrease rapidly with distance from the corner formed by the cylinder liner counterbore and the cylinder liner landing,

, thsreby reducing the crack driving force. McCarthy et al. (Supp.), ff. Tr.

24372, at 13; Tr. 25100,' 25343-45, 28819 (Rau). At a depth of less than 0.4-r

R

inch, the stresses become fully compressive, and any circumferential cracks will

-a rrest. Tr. l 25344-45, 28819 (Rau) .

I B-50. FaAA's conclusion that circumferential cracks will grow slowly, ar-rsst, and not cause any operational problem is confirmed by the operating histo-ry of the original EDG .103 block. McCarthy et al. (Supp.), ff. Tr. 24372, at I 13 14; Tr. 288'1 2-13, 28819 (Rau). _ Despite more than 1200 hours0.0139 days 0.333 hours 0.00198 weeks 4.566e-4 months of operation, incieding more than 400 hours0.00463 days 0.111 hours 6.613757e-4 weeks 1.522e-4 months at or above 3500 KW, the circumferential cracks in.

tha original EDG 103 block,- which had markedly _ inferior fatigue and fracture l

properties, did not propagate to a depth beyond 3/8 inch and did not impair en-gina operation. McCarthy et al. -(Supp.),. ff. Tr. 24372, at .13. The _ superior materials of the EDG 101,102 and 103 replacement blocks are less susceptible to circumferential crack initiation, and should such cracks initiate, they would

, propagate more slowly and arrest at a shallower depth than the deepest crack (3/8 inch) in the original EDG 103 block. Id. at 13; Tr. 25091, -25350-51 (Rau);

LILCO Exs. B-42,; B-44; -_see also Findings B-11, -12, -13.

B -51. Although' Dr. Bush testified that he would not be surprised if cir-cumferential cracks initiated in the EDGs, he concluded, based on his . engineer-4 ing judgment, that the stresses decrease rapidiv with distance into the. block i-top and move'into a compressive stress field. Tr. 26021, 26149-52, 26225, 26279 (Bush). He also concluded that this compressive stress field is strong enough I

so that circumferential cracks, if they initiate,!will not propagate to the. !

' point that theyJimpair engine operation. . Bush et al. , ff Tr. _25775, at 8; _Tr. .

i- I

26019-21 (Bush) . j

. B-52. - SC witness Anderson testified th'at he observed multiple, small, disconnected cracks. branching out below the tip of the 3/8 inch circumferential

!- crick' on the original .103 block, and that- he did not see extensive -amounts of Widmanstaetten graphite in the sample he examined from the original EDG- 103 j- . block. Anderson 'et al. , :ff. Tr. 25565, 'at 11-12. These observations were a

4 .- . ,,, - , . . . ,, . _ + . . , _ . _ , . - . - _ . _ , . , . _ . . E_,-~ , , . _ , . , ,

o unr;; liable .because they were made on a rough' cut surface that had not been met-

,allographically polished. See Tr. 26354 (Anderson); Tr. 25097-98 (Rau, Wachob);

Tr. 26666 (Rau). Complete, accurate and. detailed examination of gray cast iron esquires careful metallographic polishing because flakes of graphite are broken

-out of the iron when it is cut,-leaving artifacts which appear as shallow holes or trenches in the surface of the iron. Tr. 26663-64 (Anderson); Tr. 26666-68 (Rau). These artifacts make it impossible to draw reliable conclusions about tha presence or size of cracks or the amount of.Widmanstaetten graphite present.

Tr. 25097-98, 25138-40, 26666 (Rau) . Liquid penetrant, magnetic particle and cddy current testing of the sample examined by Dr. Anderson established that there.were no cracks deeper than 3/8 inch. Tr. 25139-40, 26667 (Rau). Because Dr. Anderson's only basis for concluding .the circumferential crack in the origi-nel EDG 103 block was propagating was his unreliable visual observation of branching cracks, Tr. 26409 (Anderson), there is no sound basis for his conclu-sion. that circumferential cracks propagate.

B-53. SC's witnesses also testified that the development of a large cir-cumferential crack could permit some up and down movement of 'the cylinder liner against the gasket that seals the liner to the cylinder head. They postulated that this could cause leakage of combustion gases into the jacket water, and that crack- propagation through the liner landing would cause the cylinder liner to fall into the crankcase. Anderson et al. , ff. Tr. 25565, at 13; Anderson e_t al, ff. Tr. 26326, : at '3.

This testimony,. based on Dr. Anderson's incorrect and unsupported conclusion that circumferential cracks propagate, is~ not. probative.

B-54. Even.if' crack propagation beyond 3/8 inch were assumed to occur, .

- SC's claim that combustion gases could escape into the cooling water system is 3 far-fetched. .Tr.' 26216-17 (Henriksen). SC witnesses have performed no 'calcula-tions or analyses of stresses in the block top to support. their claim. T r .- ;

26355,. 26370-71, 26373-75 (Eley, Anderson). ' Since at least one-third of the

)

circumference.of the liners is supported by eight gusset-reinforced stud bosses, Tr. 25100, 25246-47 (Wells); see also LILCO Ex. B-9 and Staff Ex. 9, a circum-fsrsntial crack'would have to propagate vertically 4 to 5 inches to cause appre-ciable motion between the cylinder liner and .the block. The chances of this oc-curring are remote. It contradicts both.the physical observations and FaAA's finite element analyses, which demonstrate cracks propagate at about a 45 angle, move into a compressive stress field, and arrest. Tr. 25095-96, 25100-01, 25246-47 (Wells); Tr. 28812-13, 28819 (Rau); LILCO Ex. B-64.

i B-55. Even if combustion gases did leak, they would not necessarily enter the water jacket because there is virtually no driving force to push the gases into the cooling system. Tr. 26217-19 (Henriksen). Moreover, if combustion gases did enter the cooling system, they would cause no operational problem be-cause the gases would be released into the expansion tank. Tr. 26218-19 (Hen riksen) .

B-56. SC's claim that the cylinder liner landing could separate from the block, causing the cyinder liner to fall into the crankcase, is improbable be-cause, as noted'previously, a crack would have to propagate vertically 4 to 5 inchss through the gusset-reinforced stud bosses to cause the liner landing to ssparate from the' block. If a circumferential crack propagated at a'45 angle from the liner landing through all the ligament material to the stud hole, it I would still not affect the ability of the block material to support the cylinder lin:r. Tr. 25100-02, 25104-06 ' (Wells, Rau); see LiLCO Ex. B-9.

B-57. The evidence supports the conclusion that the EDGs are qualified for nuclear service, even if 1circumferential cracks should initiate. McCa rthy.

ct al. (Supp.), ff. -Tr. 24372, at 12-14; Tr. 28812-13, 28818-19 (Rau); Bush and Hanriksen, ff.- Tr. 25775, at 7; Bush, Staff Ex.14 at 25-26 H/ ; Tr. 26020, H/ . Staff Ex.14 is the marked up version of Dr. Bush's testimony, which was inititily bound into the record following Tr. 28503. See Tr. 29020.

i

-26023 '(Bush);' Tr. ,26020. (Berlinger) . LlLCO and the Staff have agreed that a e

scheduled ~ program of monitoring the_ blocks for circumferential cracks is not re--

i ' quiredIbut that LlLCO .will inspect the block and liner _ landing area for circum-farantial cracks in the event a cylinder liner is ren,oved. Bush, Staff Ex.14 '

i- - st' 26; 'see Attachment 1.

i Vll.- The Replacemeni: Block is Adequately Designed and Tested

'B-58.

Based on ' LILCO's endurance testing at 3300 KW, SC has stipulated it ~

doss not challenge the adequacy of the replacement block for loads not exceeding

_3230 KW, which assumes an instrument error of'70 KW. See Tr. 28800 (Stipula-tion) . _ SC's stipulation that the replacement block is adequate at 3230 KW is, .

- for all intents and purposes, a recognition that the replacement ble.k is ac-esptable for nuclear service at the qualified load of 3300 KW.

LILCO Exhibit

- B-30, which plots the principal stresses vs. load recorded by strain gauges nos.

4 l 11-13, demonstrates that the difference in stresses in the block between -3230 KW and 3300 KW is almost imperceptible. See LILCO Ex. B-30. Given that the dif-forance in stresses between 3230 KW and.3300 KW is insignificant leven if a 70.KW ~ -

mster error is assumed, and further given the evidence that the meter actually provides a reliable mean load (see Load Finding L-40), SC's stipulation supports the conclusion that the replacement block is adequate for nuclear serviceSat the 3300 KW qualified load.

B-59. ' Apart from' SC's stipulation, the evidence -demonstrates the replace-

mant block .is_ a proven design - that- has been adequately . tested. FaAA's review of -

ths. replacement block shows that-this = block is a current production model, not a nsw design. as. alleged- by SC. The product enhancementsL incorporated in the re-pitczment block -- lengthening the stud bosses, thickening the block' top, and increasing the ~ clearance gap -- are relatively minor, yet they ' reduce the _

1 strssses in the block top and.make the block more resistant to fatigue crack -

l ; initiation. Johnson'. et al. , ff. Tr. -28799, .at 8; see McCarthy et al. , ff. Tr.

n W -W r--~r- Te' h* tM-+4 ++ 1 =- af- e m rW-- . yj p-e v M 1r ww

[ L24372}at.68-71. In addition ~, the use of Class 45 gray cast iron in the re-

> placement l block further reduces the possibility of fatigue cracking. McCarthy

. at al. , ff. Tr. 24372, at 69-70.12/

B-60. The impreved fatic ue resistance provided by the product enhance-4 mants . incorporated in the replacement block has been tested and proven in the i .

TDi[R-5-test engine. The R-5 test engine has been operated for more than 5000

~

[ t hours' at loa ~ds exceeding the full . rated load (3500 KW) of the Shoreham engines.

i .

McCarthy et al. , - ff. Tr. 24372, a.t 70-71;- Johnson et al. , ff. Tr. 28799, at. 8; Tr. 24879-84-(Wells). inspections ~after this operation revealed only one. liga-

-msnt crack, and this crack occurred in a cylinder where an improper cylinder

? .linari had been installed. Tr. 24885 (Wells); Tr. 25373-81 (Wachob).

i B-61. The adequacy of the design enhancements incorporated into.the re-

-placsment block has also been- demonstrated by operation of.the replacement block

et Shoreham for more -than .849 hours0.00983 days 0.236 hours 0.0014 weeks 3.230445e-4 months '. The block has been operated for more than

' 577 hours0.00668 days 0.16 hours 9.540344e-4 weeks 2.195485e-4 months at or above' 3300 KW, including more than 70 hours8.101852e-4 days 0.0194 hours 1.157407e-4 weeks 2.6635e-5 months at or above 3500 KW,

.without developing ligament or stud-to-stud cracks. . Johnson et al., ff. Tr'.

28799, at 5-6. This operation ' confirms that +he design enhancements have ~re-i

!- ducsd the possibility of fatigue crack initir'on. . It is also a direct demon-

}- - stration that the replacement block has oeen adequately tested. Id. - at 8-0.

p B-62. ,FaAA's cumulative damage analysis also demonstrates that the re-n plactment block is~ capable of performing its intended function. FaAA's conser-f vative analysis: of the EDG 101 and 102 blocks at the 3900/3500 KW LOOP /LOCA-

Llcads1 kas demonstrated that these blocks, which have known -ligament cracks, can i: Since the repir.coment block has-

~

withstand 50 consect.tive 3900/3500 LOOP /LOCAs.

r .

. - supsrior mechanical' properties and has not developed ligament cracks after op-orating 'at an approximately equivalent number of hours as the EDG 101.and 102 i

l' I;

212/ ' Tensile tests on the B-bar for the replacement block demonstrated--that the-

. Ust iron actually' exceeds the requirements for : Class 50 material. :Tr. 24764-65 L f(Wachob);?Tr. - 24766 (Rau); see also Tr. - 24874 751(Wells); . Finding ~ B-13.

.a _ .a. . - - . ._a . .1,_ . . _ . . - ___~ __ . -

blocks,-it has demonstrated even greater margin against fatigue cracking. Ld .

. at 8-9; McCarthy et al. , ff. Tr. 24372, at 70. Thus, the replacement block will p2rform'its intended function at 3300 KW, as well as at loads up to its overload rating for brief periods of time. Johnson et al., ff. Tr. 28799, at 11-12; McCarthy et al., ff. Tr. 24372, at 75; Bush, Staff Ex.14, at 24-25.

Vill. Conclusions

'B-63. Metallurgical testing of the original EDG 103 block revealed exten-siva amounts 'of Widmanstaetten graphi+e that severely degraded the mechanical ,

properties of the original block material. Microstructural evaluations and B-b:r. testing demonstrate that the EDG 101,102, and 103 replacement blocks have vastly superior mechanical properties compared to the original- EDG 103 block.

Tharefore, the possibility of fatigue crack initiation and propagation is .

grsatly reduced. Findings B-6 to B-13.

B-64. Ligament cracks are benign and will not impair the ability of any EDG to perform its required safety function. Finite element analysis and op-sration of the EDG 101 and 102 blocks for more than 350 hours0.00405 days 0.0972 hours 5.787037e-4 weeks 1.33175e-4 months at or above 3500 KW, including more than 25 hours2.893519e-4 days 0.00694 hours 4.133598e-5 weeks 9.5125e-6 months at 110% of full load without ligament crack propagation beyond the cylinder liner landing, confirm that the cracks arrest.

Morcover, ligament cracks did not initiate in the EDG 103 replacement block de-spita more than 577 hours0.00668 days 0.16 hours 9.540344e-4 weeks 2.195485e-4 months at or above 3300 KW, including more than 70 hours8.101852e-4 days 0.0194 hours 1.157407e-4 weeks 2.6635e-5 months at or above 3500 KW. Further, even if ligament cracks were to initiate in the re-placcment block and propagate in all three engines, they would not impair EDG cp;rability. Findings B-1, B-3, B-5, B-21 to B-27.

l B-65. The EDG 10.1: and 102 blocks do not have circumferential cracks. Op-cration of the original EDG 103 block, which contained extensive amounts of Widmanstaetten graphite, for over 400 houra at or above 3500 KW did not result ,

in crack propagation to a depth beyond 3/8 of an inch.

t This operation combined with analysis confirms that circumferential cracks arrest. _Th e refore,

!+

[ circumferential' cracks, even if they snould initiate, will not impair the abili-

ty 'of the EDGs to perform their required safety function. Findings B-4, B-5, B-44 to B-57.

, B-66. Stud-to-stud _ cracks are not likely to initiate in the EDG 101, 102,

. or 103 replacement blocks. In the unlikely ' event that a stud-to-stud crack should initiate, cumulative damage analysis demonstrates the cracks will not l

propagate to the point that they. impair EDG operation during 50 consecutive "3900/3500 KW ' LOOP / LOC As. " Even greater margin is demonstrated for a "qualfied load LOOP /LOCA." _ Findings B-1, B-3, B-5, B-28 to B-43.

B-67. The Shoreham cylinder blocks have been demonstrated by analysis,

testing, and operation to be suitable for nuclear service at the qualified load of 3300 KW and for postulated short-term loads up to 3900 KW. Therefore, the cylinder blocks meet the requirements of GDC 17. Findings B-63 to B-66.

B-68. Contrary to SC's contention, it was not necessary to test the re-placzment block for the entire 745 hour0.00862 days 0.207 hours 0.00123 weeks 2.834725e-4 months confirmatory test period. The adequacy of the replacement block for nuclear service at the qualified load and up to its

. dasign rating-has been verified by review of the block design enhancements, re-i visw of TDl's testing on the R-5 test engine, _ actual operation at Shoreham, and by cumulative damage analysis. See Johnson _ et al. , ff. Tr. ' 28799, at 8-9.

f B-69. SC's contention that the testing on the EDG 103 replacement' block

. is not transferable to the EDG-101 and 102 blocks is irrelevant since LILCO.does

- nct assert that the testing is transferable and does not rely on this testing to

' qualify the EDG 1101 and '102 blocks. Rather, finite element analysis, cumulative

! damage analysis, metallurgical testing, and operation ha.ve demonstrated the -

suitability of 'the EDG 1011and 102 blocks. Further. endurance testing of the-i blecks is unnecessary and'would!not, in any event, provide probative evidence with respect to fatigue crack initiation'

. Tr. 28854, 28876-77, 28881-83 (Rau);

4

, - Tr. _29082 (Bush) .

1

- ~~.. -. _- .. . - m. . , -L-----. -

g, ~ . -, , - a s -,_. - . - _ ,

?

s 9

4 h

ATTACHMENT 1 i

i f

e

- k nven- -w-- vm..,m - ar ,w-er- .st - en we n.aa w -

m- va--, -

-e-e ~

BLOCK TOP INSPECTIONS I. LILCO COMMITMENTS l

1. - During any period of continuous operation following autcmatic diesel generator initiation, Lilco will perform daily vicual inspections of the area between adjacent cylinder heads cnd the general block top. Lilco will also perform visual inapections of the same areas under intense light during the I conthly surveillance testing.

( :2. Lilco will inspect the top surface of the block oxposed by removal of two cylinder heads each from the EDG 101

!- and 102 engines at each of the first four consecutive refueling outages. Inspection will be-by liquid penetrant with eddy j current as appropriate. Based on the results of these i inspections, Lilco may request such inspections be terminated j aftor the fourth outage.

3. Lilco will perform eddy current testing between adjac nt cylinder heads after any operation of amp EDG,st i graster than 1800 KW. hieg gL let

. 4. Lilco will perform a liquid penetrant and, as i _cppropriate, UT inspection of the cylinder liner landing at any

time a cylinder liner is removed for any other reason.

o . . .

" II . NRC STAFF RECOMMENCATIONS

! 1. The foregoing Lilco commitments satisfy NRC Staff recommendations with respect to block top inspections. Thus,

' thore a're no NRC Staff recommendations not, accepted by Lilco.

In codition to the inspections set forth in paragraphs I.2 and i I.3 above, the current SER'also recommends that two cylinder ,

! hocds be removed from EDG 103 at each of four consecutive !

rofueling outages for purposes of inspecting the block top l oroes. The NRC Staff no longer'censiders this necessary and

! intends to issue a revised SER to reflect that removal of two cylinder heads each from EDG 101 and 102 at each of'four conoecutive refueling outages for purposes of inspecting the block top is sufficient.

i s

2. It is also agreed by and.between the NRC Staff and j Lilco that at the conclusion of the fourth refueling outage,

' tho necessity for further inspections in accordance with' parcgraph I.2 above, if any, will-be re-evaluated.

~

3. It is agreed by and between the NRC Staff and Lilco I that because there are no ligament cracks in the EDG 103 ccplacement bloch, eddy. current testing between adjacent cylinder heads of the EDG 103 block (paragraph I.3 above) is not reou red.

e i

I i

t .

, ,. _ . . . . . l

1 l

1. -Intro'duction These findings are in addition to previous crankshaft findings submitted l l

by LILCO in November and December 1984 and result from additional testimony l pr:ssnted at reopened hearings in February and March,1985.

11. Reliability of Crankshafts at 3300 KW A. 10E7 Cycle Endurance Run C-1. Suffolk County submitted no crankshaft testimony in the reopened pr:creding and concedes 'the adequacy of the crankshafts at 3300 KW based on thsir consultants' determination that the replacement crankshafts comply with DEMA, Lloyd's and ABS at this load. Tr. 28417-18.

C-2. The 10E7 cycle confirmatory test on EDG 103 consisted of 220 plus hours of operation at or above 3300 KW prior to October 8,1984, and an endur-anco run of 525 hours0.00608 days 0.146 hours 8.680556e-4 weeks 1.997625e-4 months at approximately 3300 KW between October 8,1984 and Novcmber 2,1984. Dawe et al. , ff. Tr. 27153, at 38-39; Pischinger et al. , ff.

Tr. 28416, at 5. Together these hours constitute 10,000,000 (10E7) Ioading cy-clss. Dawe et al. , ff. Tr. 27153, at 38; Pischinger et al. , ff. Tr. 28416, at 5.

C-3. Testing for 10E7 loading cycles at a particular load is an accepted m thod of establishing the adequacy of the crankshafts at that particular load.

Bsrlinger et al., ff. Tr. 23126, at 18; Tr. 23477-79 (Henriksen, Sarsten). This is the ultimate test for the crankshaft. Berlinger et al. , ff. Tr. 23126, at 21.

C-4. Suffolk' County claims that meter error and fluctuation caused by EDG 103 being connected to the grid during the .10E7 confirmatory test run resulted in the qualification run being performed at a level nc higher than 3230 KW and implies that surveillance. testing of the EDGs in the future would have to be as-l sum:d to occur at a load of 3470 KW. Bridenbaugh and Minor, ff. Tr. 27500, .at c 21-23. This is incorrect. See Findings L L-53.

C-5. The nondestructive examinations performed upon the EDG 103 crank-shaft following the 10E7 loading cycle confirmatory test demonstrate that the crankshaft suffered no fatigue damage as a result of 745 hours0.00862 days 0.207 hours 0.00123 weeks 2.834725e-4 months of operation at or about 3300 KW. Pischinger, et al., ff. Tr. 28416, at 8 and 11. This estab-lishss that the Shoreham replacement crankshafts have unlimited life at 3300 KW.

Pischinger et al., ff. Tr. 28416, at 11; Bush and Henriksen, ff. Tr. 28503, at 12.

4 B. Kritzer-Stahl C-6. Kritzer-Stahl is a very conservative, German design criteria for cvaluating the adequacy of a crankshaft. .Tr. 22767-69, 22772, 22775-77 (Pischinger) .1/ The factor of safety calculated under Kritzer-Stahl for the re-plac ment crankshafts at 3300 KW is 1.318. Pischinger et al., ff. Tr. 28416, at

-3 .' This is at the upper range of safety factors commonly accepted in the Euro-ps:n diesel industry (1.15-1.3) and, coupled with the 10E7 loading cycle con-firmstory test, demonstrates that the crankshafts have infinite life at 3300 KW.

Pischinger et al.,. ff. Tr. 28416, at 5.

C. DEMA C-7. Although no party presented evidence at the reopened hearing con-carning whether the crankshafts meet DEMA at 3300 KW, as this was not contested by Suffolk County, it can be concluded that DEMA is met at 3300 KW from the

, r: cord in the reopened proceeding and the prior proceeding.2/ Using 1/ For a further explanation of Kritzer-Stahl and its methodology, see LILCO's Proposed Findings of Fact filed herein on November 5,- 1984, Findings 82-89 at 31-33.

2/ While Professor Sarsten's calculations in the prior proceeding showed the DEMA limits were exceeded at 3500 KW at overspeed and underspeed, he used all crd:rs (24) instead of major orders as DEMA. requires and had had no prior expe-ri:nce with DEMA. See LILCO's Proposed Findings of Fact, dated November 5, '

1984, Findings 23-25 at 10-11. Professor Sarsten also calculated by interpola-

- ti:n that at 3300 KW, DEMA was met at; synchronous and underspeed conditions, but indicated that these calculations were only approximate. Tr. 23378-79 (S:rsten).

- t,alculations typically performed by the American diesel engine industry to veri-fy a crankshaft's ability to withstand torsional stresses, it has been shown that the Shoreham replacement crankshafts meet DEMA at 3500 KW. McCarthy e_t 6, ff. Tr. 22610, at 27-30; see also LILCO Ex. C-17, at 2 2-5, 4-1. The m22sured stresses on the crankshaft between 1750 KW and 3800 KW show that they

. vary linearly with load. See LILCO Ex. C-16, Figure B-4. The stresses at 3300 KW and 3500 KW are within approximately 5% of each other. Tr. 28455 (Pischinger). Therefore, since DEMA is met at 3500 KW, a fortiori, it is met at 3300 KW since the stresses at this load are lower.

Ill. Reliability of Crankshafts For Loads Above 3300 KW A. Loads Between 3300 and 3500 KW C-8. Loads between 3300 KW and 3500 KW3/ are of no concern given Dr.

Pischinger's calculations under Kritzer-Stahl which demo'nstrate that the crank-shafts have a factor of safety of 1.248 and unlimited life at 3'500 KW.

Youngling and Pischinger, ff. Tr. 22610, at 5; Pischinger et al., ff. Tr. 28416, at 4, 13. Dr. Pischinger also performed a cumulative damage analysis based on tho 745-hour confirmatory test and the loads experienced during those hours, which demonstrates that the crankshafts have experienced the equivalent of t: sting. to 10E7 cycles at 3505 KW and thus have unlimited life at this load.

Tr. 28419-20 (Pischinger). This analysis was performed by Dr. Pischinger under tha generally accepted riethodology of Miner-Palmgren-Haibach. Tr. 28420 (Pischinger). Although this methodology does not take into account load scquencing which can have an effect on the analysis, it-does take into account cycl s below the conservatively estimated endurance limit, which ;s thought to c:mpensate for the influence of load sequencing. Tr. 28420-21 (Pischinger).

i 3/ For a discussion of postulated loads and their likelihood, see Findings L L-28, and L-36 ' - L-53.

i

{

- Additional conservatism is provided by the fact that Dr. Pischinger did not take crzdit for more than '500 hours0.00579 days 0.139 hours 8.267196e-4 weeks 1.9025e-4 months of operation at or below 3300 KW. If credit were taksn for these hours, Dr. Pischinger would have calculated an unlimited life lord figure above 3505 KW. Tr. 28426, 28453 (Pischinger) .

C-9. Dr. Bush also. performed a cumulative damage analysis using the M:nson approach to Palmgren-Miner and determined initially that the endurance limit for the replacement crankshafts was at least 3430 KW a!!owing for a 70 KW instrument error. Bush and Henriksen, ff. Tr. 28503, at 16-17. In subsequent tsstimony, Dr. Bush indicated that the endurance limit would be approximately 3480 KW allowing for a 20 KW instrument error. Tr. 28532 (Bush).4/ Dr. Bush

= modsled various combinations of load sequencing which Manson utilizes and deter-mintd there was no significant difference caused by load sequencing. Thus, Dr.

Bush concludes that this indicates the endurance limit for the eplacement crankshafts is around 3500 KW or possibly higher. Tr. 28660 (Bush). Finally, Dr. Bush also indicated that based on experimental data in reference literature, mat:1 such as that in the Shoreham replacement crankshafts initiates its fatigue limit near 10E6 loading cycles. Bush and Henriksen, ff. Tr. 28503, at 17. Be-causa LILCO's nondestructive examination of the crankshaft following the 10E7 Inding cycle test demonstrates that cracks' had not initiated during the initial 3 x 10E6 cycles at or above 3500 KW, and because stresses at 3500 KW and 3300 KW ara not substantially different, subsequent operation for 7 x 10E6 cycles at 3300 KW would have propagated any cracks caused by operation at 3500 KW. M.

H nce Dr. Bush concludes that this is further confirmation that the crankshaft cndurance limit _is at or above 3500 KW less any assumed instrument error. M.

4/ For a further discussion of the + 70 KW watt meter; accuracy and the + 20 KW digital test loop accuracy,- see Load Findings L L-41.

, - - ~

C-10. The adequacy ~ of the replacement crankshafts to sustain loads be-

'twssn 3300 KW and 3500 KW is also confirmed by FaAA's fatigue analysis which re-sultcd in a factor of safety of.1.48 at 3500 KW utilizing the actual experience

from the failed original crankshafts as well as experimental data from the re-placcment, cran kshafts.5/- McCarthy et al., ff. Tr. 22610, at 32-39; LILCO Ex.

C-17, at 3-11.

C-il . Dr. Chen also is of the opinion, based upon his DEMA calculations and his. years of experience. that the replacement crankshafts are safe and reli-abla for operation at 3500 KW. McCarthy et al. , ff. Tr. 22610, at 27-30; Tr.

23020 (Chen); see also LILCO Ex. C-18, at 2-4.

.B. Loads ' Above 3500 KW

,C-12. '

i Loads between 3500 KW and 3900 KWQ/ are not of concern since the raplacement crankshafts have been analyzed by Dr. Pischinger under Kritzer-Stahl l

to have unlimited life at 3900 KW when the inherent safety factor of Kritzer-Stahl is considered. Tr. 28427-28 (Pischinger). Moreover, without taking into -

account this inherent conservative feature of Kritzer-Stahl, the replacemer.t crankshafts would have a predicted lifetime of 1200 hours0.0139 days 0.333 hours 0.00198 weeks 4.566e-4 months at 3900 KW.- Tr. 28428 (Pischinger). . See LILCO's Proposed Findings 'of FactLdated November 5,1984,

. Finding 90 at 33-34.

C-13. Dr. Chen's analysis further confirms-that the crankshafts are safe

-and reliable at 3900 KW, as does FaAA's fatigue analysis. Tr. 23020-21, 73080

-(Chsq); see also. LlLCO Exs. C-17, at 4-1, and C-18, at 3; McCarthy et al. , . ff.

Tr. 22610, at 38.

5/ For a further description of this' analysis see LILCO's Proposed Findings cf Fact dated November 5,1984, Findings 49-81,. at 18-31.

' .L-56.

!/ .For a discussion of postulated loads and their likelihood, .see, Finding

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

i e C214. Dr. -Bush's testimony concludes that the replacement crankshafts can sustain, without concern, any loads above 3500 KW postulated by the Staff. Bush and Henriksen, ff. Tr.' 28503, at 14, 21-22. This would include the ability to sustain' a load as high as 3900 KW for up to one hour without crack initiation.

l M. .

j-l C-15. .None of the analyses set out herein have made any allowance for the .

I bsnsficial effects Lof shotpeening. There would certainly be a gain in fatigue p rssissance due to shotpeening. Tr. 23155-56 (Bush). This gain could be from 10% to 30%. Wells et al., ff. Tr. 23122, at 22; Tr. 23158 (Wells). Thus, not allowing for shotpeening adds an additional factor of conservatism to the vari-j ous analyses of the crankshafts.

j IV. Future Testing and Inspections C-16. Surveillance testing of the Shoreham EDGs should be conducted at a-1 maan value of 3300 KW.' It is unnecessary to test the EDGs at loads above this l sysn though they are capable of sustaining higher-_ loads without concern. This would be excessive and not accomplish anything. Tr. 28732-37 (Bush).

J C-17. Additional assurance is provided by the fact that LILCO has com-mitsd to the Staff to perform certain inspections on the crankshafts at sched -

ulsd times, all 'of which has -been agreed to by the- Staff. . A copy of thosa com-l mitm:nts is attached to these findings in accordance with the Board's prior

[ instructions (Attachment 1).

! ;V.' Conclusions o q C-18. ' The Shoreham replacement crankshafts have.been demonstrated to be '

[

b adequate and reliable at a qualified load of 3300 KW by both testing and analy- '

s:s. Findings C-1 C-3, :C C-7. Therefore the requirements of GDC 17 are m:t.

t C-19. The Shoreham replacement' crankshafts- have been shown' by analysis '

l-

[ :and testing to be adequate"and reliable for unlimited life at 3500 KW and by.

crnsarvative analysis-to have unlimited life at 3900;KW. Findings' C C-18. -

i s: - _ _

+

C-20. Given 'the demonstration of unlimited life at 3500 KW and 3900 KW, .

i' the replacement crankshafts are adequate and reliable for any postulated loads above the 3300 KW qualified load, up to 3900 KW. In addition, should LILCO seek to license the diesel generators at the r design ratings at a subsequent date,

- tha~ crank' s hafts are qualified for a continuous rating of 3500 KW and a short-tarm rating of 3900 KW.

i 4

i t

i 4

i d

l

)

-m , , .--_, - -_ - ,

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

ATTACHMENT 1 CRANKSHAFT INSPECTIONS

1. LILCO Commitments

1. At each refueling cutage, LILCO will measure and record hot and cold web deflection readings on each of the diesels.

2. At the first refueling outage, LILCO will inspect the crankpin journals numbers 5, 6 and 7 and associated oil holes in these journals, using LP !

and ET as appropriate. These inspections will only be performed on EDG l 101 and EDG 102.

3. During the second and subsequent refueling outages, LILCO will inspect two

- of the three crankpin journals subject to the highest stresses (Numbers 5, 6 and 7) and associated oil holes in these journals, using LP and ET as appropriate. These inspections will be performed on EDG 101,102 and 103.

4. At intervals of.every 3 refueling outages, LILCO will inspect the main bearing journals and associated oil holes, between crankpin journals num-bers 5,' 6 and 7, using LP and ET as appropriate. These inspections will be performed on EDG 101,102 and 103. Based on the results of this first inspection, LILCO may request that such inspections be terminated.

I ll. NRC Staff Recommendations

1. The foregoing LILCO commitments satisfy NRC Staff recommendations with re-spect to crankshaft inspections. Thus, there are no NRC Staff recommenda-tions not accepted by LILCO. As opposed to the intervals discussed in paragraph I.4 above, the current SER recommends that inspection intervals for the main bearing journals on EDG 101 and 102 be at the first and all subsequent refueling outages, and for EDG 103, the second and all subse-quent,. refueling outages. The Staff no longer considers this -necessary and intends to issue a revised SER to reflect the changes in inspection inter-vais to those shown in paragraph I.4 above.

2. It is also agreed by and between the NRC staff and LILCO that at the con-clusion of the first, 3 refueling outage interval, the necessity for fur-ther inspections in accordance with paragraph I.4 above, if any, will be re-evaluated.

s 1

l 1

LILCO, November 5, 1984 4

i UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION Before the Atomic Safety and Licensing Board In the Matter of )

)

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

)

(Shoreha:n Nuclear Power Station, )

Unit 1) )

f j

LONG ISLAND LIGHTING COMPANY'S PROPOSED FINDINGS OF FACT HUNTON & WILLIAMS Post Office Box 109

' Raleigh, North Carolina 27602 I

LI . Background

.l. The crankshaft in a diesel engine such as the Emer-p gency Diesel Generators (EDGs) at Shoreham converts the I reciprocating motion of the pistons and connecting rods into rotary motion. In this process the crankshaft converts the in-ertial and gas pressure firing forces into torque. The torque l

from the crankshaft. drives the electrical generator to provide

! emergency power. (McCarthy et al, ff Tr. 22,610 at 7).1/

i 2.- The original crankshafts provided by TDI had a 13-inch main journal and an ll-inch crankpin and the fillet re-gions were not shotpeened. On August 12, 1983, the original l

j 13-inch by ll-inch crankshaft on EDG 102 fractured through the f

crankpin and rear web under cylinder No. 7. Subsequent inves-

{

tigation revealed that the crankshaf t on EDG 101 was cracked at the No. 5 and No. 7 crankpins and the crankshaft on EDG 103 was cracked at the No. 6 crankpin. (Montgomery, ff Tr. 22,610 at-1

! 7-9).

1

3. The f ailure of the original crankshaf ts was caused ,

by high cycle vibratory f atigue. The crankshafts were unable 1/ The profiled testimony of Messrs. McCarthy, Johnston, Montgomery and Chen and Messrs. Pischinger and= Youngling,were t not numbered.when bound into the transcript. Both sets of tes-timony may be found following page 22,610 of the transcript.

I t

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

- - . . ...-m . m , . ,

i <<

to withstand the torsional stresses imposed upon them during operation of the engine. (McCarthy, Johnston, ff Tr. 22,610 at 7-8).

4. The original crankshafts did not comply with the Diesel Engine Manuf actuers Association (DEMA) recommendations for allowable crankshaft vibratory stress. (Tr. 22,840 Johnston; Tr. 22,841 Chen).

5. LILCO replaced the crankshafts in all three engines with crankshaf ts of a different design. The replacement crank-I shafts have a 13-inch main journal and a 12-inch crankpin. The 4 crankpin-to-web fillet radii of the replacement crankshafts 4,

have a larger radius of curvature than the fillet radii of the original crankshafts, and the fillet regions of the replacement

! crankshaft have been shotpeened. (Montgomery, f f Tr. 22,610 at i

8-9).

6. The minimum ultimate tensile strength of the re-placement-crankshafts is over 100,000 psi. The average ulti- .

mate tensile strength of the original crankshaf ts was approxi-

, mately 93,500 psi. .(Montgomery, f f Tr . 22,610 at 9 ) .

7. The replacement crankshafts have undergone exten-sive engineering analyses and testing to determine their ade-quacy for service in the Shoreham EDGs. Failure Analysis Asso-i i ciates (FaAA), Power and Energy International (PEI) and FEV have analyzed the replacement crankshafts, and all concluded 1

4

- . -~ - - -

- - - . _. _ _ = _ _

4 that the crankshafts are adequate for service in the Shoreham EDGs at 3500 KW and 3900 KW. FaAA reviewed TDI's torsional analysis and the results of torsiograph tasts to determine that the crankshafts comply with DEMA. FaAA also conducted a full

< scale fatigue analysis based on testing to determine a true factor of safety for the crankshafts. FaAA found the factor of safety was 1.48. PEI performed a torsional analysis typically used by the diesel engine industry to determine whether a 4

crankshaft complies with DEMA. The nominal torsional stresses in the replacement crankshafts are well below the DEMA limits.

J 4 FEV calculated a factor of safety for the replacement crank-l shafts under the Kritzer-Shahl criteria. The crankshafts have a margin of safety of 24%. (Montgomery, f f Tr. 22,610 at 8-9; Chen, ff Tr. 22,610 at 21; Tr. 23,004 Pischinger).

i II. The Crankshaf ts are Only Required to Comply with DEMA. The Crankshafts do not Have to Comply with the Requirements of any Other Design Society 1

8. The purchase specifications for the Shoreham EDGs, Spec. No. SH1-89, Revision 2, January 26, 1983, required that the replacement crankshafts meet the DEMA recommendations for allowable vibratory stress. (Montgomery, f f Tr. 22,610 at 10; LILCO Diese] Exhibit C-2).

! l l

i L ._

e l l

i l !

I l (Reg. Guide l

9. NRC Regulatory Guide 1.9, Revision 2,

, 1.9) addresses the design requirements for diesel generator units used as standby electric power systems at nuclear power plants.. Reg. Guide 1.9 provides that conformance with the re-quirements of.IEEE Std 387-1977 is acceptable for meeting the design criteria and qualification testing of diesel generator 1 units used-as onsite electric power systems for nuclear power plants. -IEEE Std 387-1977 provides that diesel generators should comply with the standards of DEMA's Standard Practices for Low and Medium' Speed Stationary Diesel and Gas-Engines.

(Montgomery, f f Tr. 22,610 at 11-12; LILCO Diesel Exhibits C-3, C-4; Henriksen, Sarsten, ff Tr. 23,126 at 10-11).2/

10. DEMA is not a design code in the sense that DEMA does not provide detailed rules that tell an engine manuf actur-er how to design a crankshaft. However, DEMA does . provide spe-cific stress limits to measure the adequacy of a crankshaft. ,

Engine manufacturers have used DEMA for years on stationary diesel generator installations to determine whether a crank-shaft is adequate for its intended service. (Chen, ff Tr.

22,610 at 13-14).

2/ The prefiled

.Henrikson,

. testimony of Messes. Berlinger,- Bush, Laity and Sarsten was not numbered when bound into l

the transcript. The testimony may be found following page l

l 23,126 of the-transcript.

4 4- e - - - . - - . . . ._ m -. r ~m. _ _--.

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

l l

< ll.

.Suffolk County asserts that the crankshafts do not ,

comply with the maximum horsepower requirements of Lloyd's Reg-

'ister of Shipping (Lloyd's), do not meet the torsional or web thickness requirements of the American Bureau of Shipping (ABS), do not comply with the minimum safety factor under CIMAC, and do not comply with the criteria used by FEV.

The (Christensen, Eley, ff Tr. 23,826 at 114-21, 123-32).

record demonstrates that the crankshafts have unlimited life at 3500 KW and at least 1200 hours0.0139 days 0.333 hours 0.00198 weeks 4.566e-4 months of life at 3900 KW under the i

Kritzer-Stahl criteria.used by FEV. (Pischinger, ff Tr. 22,610 at 5).

A CIMAC calculation performed by ABS shows that the

' crankshafts exceed the minimum CIMAC factor of safety of 1.15.

(Suffolk County Diesel Exhibit 43 at 29). ABS has determined that the crankshafts meet ABS torsional requirements. (LILCO Diesel Exhibit C-13)..' ABS has approved the sizing of the webs (Montgomery, f f Tr. 22,610 at 17) and the NRC Staff's consul-tants believe the webs meet the ABS requirements. (Henriksen, Sarsten, ff Tr. 23,126 at 11; Staff Diesel Exhibit 1).

12. The rules, standards and design methodologies of marine classifications societies vary widely and, in fact, pro-l 4 vide differing acceptance criteria for the same crankshaft de-

[ sign parameters (e.g., journal / pin sizing, allowable horsepow-A crankshaft may er, allowable torsional stress levels, etc. .

not meet the criteria of certain codes and be perfectly adequate under other codes. (Chen, ff Tr. 22,610 at 14).

- . i I'

l

l l

l 1

13. A crankshaft may be entirely adequate for its in-tended service and not comply with the rules of ABS, Lloyd's or CIMAC. While compliance with one of the codes generally pro-vides assurance.that a crankshaft is adequate, noncompliance i

does not necessarily mean a crankshaft is inadequate. Rather, noncompliance merely means a crankshaft does not meet the de-sign requirements of a particular code. If a crankshaft is not required to meet that code by specification or other require-

ment (e.g., insurance purposes, licensing requirements, etc.),

and.there is assurance from other sources that the crankshaft is adequate, noncompliance is not significant. (Chen, ff Tr.

22,610 at 15-16).

14. Suffolk County witnesses argue that the crank-shafts should be required to comply with the design criteria of all major classification societies. (Christensen, Eley, ff Tr.

23,826 at 113-14). There is, however, no regulatory require-ment that the crankshafts comply with any standard other than DEMA. (Montgomery, f f Tr. 22,610 at 11-12, Henriksen, Sarsten, ff Tr. 23,126 at 10-11)

15. Good design practice does not require that diesel generators in nuclear standby. service meet any of the rules or requirements established by various marine classification soci-eties. The rules of the classification societies are for en-gines designed to operate in marine applications. Marine l

l l i

l

I l

sngines are exposed to conditions far different from those for l (Henriksen, Sarsten, l standby . engines at nuclear power plants.

ff Tr. 23,126 at 11; Chen, ff Tr. 22,610 at 15-16; Tr. 22,708

- Pischinger).

III. The Crankshafts Comply with DEMA i

16. The DEMA recommendations for allowable crankshaft vibratory stress provide:

In the case of constant speed units, such as generator sets, the objective is to insure that no harmful torsional vibratory stresses occur within five percent above and below rated speed.

For crankshafts, connecting shafts,

flange or coupling components, etc., made of conventional materials, torsional vibratory conditions shall generally be. considered safe when they induce a superimposed stress of less than 5000 psi, created by a single order of vibration, or a superimposed-stress of less than 7000 psi, created by the summation of the major orders of vibration which might come into phase periodically.

l DEMA's Standard Practices for Low and Medium Speed Stationary Diesel and Gas Engines was last revised in 1972. .The allowable i limits for torsional vibratory stresses,-however, have not been revised since at least 1958. (LILCO Diesel Exhibit C-14; Tr.

22,710-12 Chen) .

~

17. The DEMA allowable stress' limits are-based on the 4 assumption that the crankshaft is manufactured from

-- - w - , - - - , , -- ,--w--., a . .~w,- - - .. -,

I conventional materials. At the time the DEMA allowables were established, the average ultimate tensile strength of conven-tional material used in crankshafts was between 60,000 and (Tr. 22,711 Chen) . In contrast, the minimum ulti-70,000 psi.

mate tensile strength of the replacement crankshafts is over 100,000 psi. (Montgomery, f f Tr. 22,610 at 9).

18. In a four-stroke engine such as the Shoreham EDGs, harmonics of the order 0.5, 1.0, 1.5, 2.0, 2.5, 3.0 . . . .

1 These orders continue infinitely. (Johnston, Chen, ff exist.

Tr. 22,610 at 23; Tr. 23,048 Johnston). At the time the DEMA stress allowables were first adopted, the only available method of calculating combined stresses involved laborious, i time-consuming hand calculations. (Tr. 22,742 Pischinger).

1 i

Not until the development of sophisticated, high speed digital computers in the mid-1960s was it possible as a practical mat -

ter to calculate the combined stresses of more than six orders.

Current computational techniques permit the summation of 24 or-4 ders. (Tr. 22,989-90 Pischinger). Stresses are typical.ly not I calculated for orders higher than number 12 (a total of 24, 4

including half orders) because the stresses' caused by the I

higher orders are insignificant. (Tr. 23,253 Sarsten).

19. .With the development of sophisticated computers

' that permit the calculation of combined nominal stresses of 24 orders in all modes,:and the development of better quality.

i j i

l cteel, the trend among classification societies has been to be-l come less conservative in their rules for allowable stress.

(Tr. 22,995-96 Pischinger). DEMA, on the other hand, has not l

(Tr. 22,710-12 revised its allowable stresses since the 1950s.

Chen). This indicates that DEMA has a large built-in margin of cafety,

20. The DEMA allowables were established as a result of experience with many crankshafts. The allowables have to be correlated with the analytical techniques that were used to an-alyze the stresses in those crankshafts. Since the allowable limits are based on stress values that were calculated by a certain technique (i.e. , Holzer forced vibration calculations),

it is appropriate to perform that type of calculation to deter-It is not ap-mine whether a crankshaf t meets the DEMA limits.

propriate to use a technique that did not exist at the time the limits were established. (Tr. 22,851-53 Johnston). The Holzer forced vibration calculations performed by TDI are standard techniques used to determine whether a crankshaf t complies with DEMA. (Tr. 22,755-56 Johnston). TDI's single order Holzer l

calculations show the replacement crankshafts comply with DEMA. f (Johnston, ff Tr. 22,610 at 24).

21. The standard practice in the diesel engine industry is to sum four or six orders for purposes of comparison with the DEMA allowables. (Tr. 22,729-30, 22,832 Chen).

In certain

. . s instances, as few as two orders may be summed. For example, .

ABS summed only two orders when reviewing the stresses in the l

replacement crankshafts. (Tr. 22,738 Johnston). The TORVAP-C computer l program used by PEI, which is a common domain software program widely .used by the diesel engine industry to calculate nominal stress, is designed to sum only six orders. (Tr.

22,745, 22,747 Chen).

22.. The calculation PEI performed using the TORVAP C f computer program is a modal superposition calculation. (Tr.

22,716-18 Chen). FaAA and Professor Sarsten also calculated i nominal torsional stresses using modal superposition or harmon-i ic synthesis analyses. The distinction between PEI's calcula-tions and the calculations performed by FaAA and~ Professor

! Sarsten lies primarily in the number of orders summed. (Tr.

23,034-35 Chen). FaAA and Professor Sarsten did not sum major orders. They summed all (24) orders. PEI performed two calcu-lations, one in which six orders were summed and one in which

^

twelve orders were summed. The calculation that is most appro-I priate to compare with the DEMA allowables'is the PEI'calcula-

] tion that sums six orders. (Tr. 22,729 Chen). FaAA and Pro-fessor Sarsten summed many more orders than are included in the i

term major orders. . -(Tr. 22,734 Johnston) .

23. The NRC Staff asserts that the crankshafts do not comply with DEMA.. Professor Sarsten's calculations show that i

i

l l

l the combined nominal stresses in the crankshafts slightly ex-coed 7000 psi at full load and at overspeed and underspeed.

j Professor Sarsten, however, made no attempt to sum the major orders. ~Rather, he summed all 24 orders in his calculations.

(Sarsten, ff Tr. 23,126 at 13-14; Staff Diesel Exhibit 2).

-24. Professor Sarsten has no experience with the ap-plication of DEMA. He has never used DEMA prior to his in-i volvement in this case. (Tr. 23,255 Sarsten).

25. DEMA does not require that the combined stress of

.all (i.e., 24) orders be less than 7000 psi. Rather, DEMA only requires that the combined stress of the major orders of vibra-i tion be less than 7000 psi. DEMA does not define major orders, because they are different for every engine and depend on a number of variables. The determination of what the major or- .

j dets are for each crankshaft is left to sound engineering judg-ment. (Tr. 22,741, 22,832 Chen).

! 26. The major orders in the replacement crankshafts at

' Shoreham are the 4.0, 5.5, 4.5 and 2.5 orders. (Tr. 22,739, 22,747-48 Johnston; Tr. 22,741 Chen; LILCO Diesel Exhibit C-17

! at 3-14; LILCO Diesel Exhibit C-18 at 11). The major orders

! are those that give the largest free end vibrational ampli-tudes. (Tr. 22,741 Chen) . ,

~

! 27. PEI's torsional calcu1ations were performed in two j i

! parts. First, PEI calculated the single order stresses for the )

I l

I S

l

)

l first 20 orders (1 through 10, including half orders) using the TORVAP R program. TORVAP R is a classical Holzer forced vibra-tion calculation. (Tr. 22,728-31, 22,843 Chen) . Second, PEI selected the six largest orders and calculated the combined t

l stresses using the TORVAP C program. TORVAP C is a modal su-l perposition calculation. As an added measure of conservatism, PEI' designed a-special subroutine for TORVAP C and calculated i the combined stresses for the twelve largest orders. (Tr.

22,745-47 Chen).

28. The six orders used by PEI, in its TORVAP C calcu-lation include the 0.5, 1.5, 2.5, 4.0, 4.5 and 5.5 orders.

(Tr. 22,749 Johnston; LILCO Diesel Exhibit C-18 at 10) .

29. Dr. Chen, the president of PEI, was chairman of the DEMA technical committee from 1971 to 1973 and has worked in

'the American diesel engine industry since 1952. (Chen, ff Tr.

22,610 at 4, 30). The calculations performed by PEI are typi-i cal of a calculations performed by the diesel engine industry to check the adequacy of a crankshaft to withstand torsional l stress. These calculations show that the crankshafts comply j

with DEMA at full load (3500 KW), overload (3900 KW), overspeed 4

(105%) and underspeed (954). (Chen, ff Tr. 22,610 at 28-30).

30. PEI's single order calculations show the crank-shafts comply with DEMA. The nominal ntresses for the number 4

4.0 order ( the ~ largest single order) are well below the DEMA i

allowable of 5000 psi.

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

. ENGINE SPEED LOAD . NOMINAL STRESS '

(RPM) (]91) (PSI) l 450 3500 3455 l

450 3900 3740 427.5 3500 3071 472.5 3500 4010 (Chen, ff Tr. 22,610 at 29).

l 31. PEI's combined stress calculations for the six largest orders show the crankshafts comply with DEMA. The nom-inal stresses for the major orders are well below the DEMA al-lowable of 7000 pai.

f ENGINE SPEED LOAD NOMINAL STRESS (RPM) (KW) (PSI) i j 450 3500 5101

450 3900 5401 427.5 3500 6232 472.5 3500 5673 l

f (Chen, ff Tr. 22,610 at 29-30).

32. As an added measure of conservatism, PEI summed an additional six orders and created a special subroutine for TORVAP C. The combined nominal stress for twelve orders (at I least double the nuraber typically summed for a DEMA calcula- -

t tion) at full load (3500 KW) and rated speed (450 rpm) is 6020 psi, well below the 7000 psi allowable. (LILCO Diesel Exhibit i

! C-18 at 10).

4

33. The nominal stresses in the original 13-inch by ll-inch crankshaf ts significantly exceeded the DEMA limits. '

l t

f 1

i k

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

l L

The single order stress for the number 4.0 order was 6200 psi.

'The combined stress, based on the summation of only four or-ders, was 9000 psi. (Tr. 22,841, 22,969-70 Chen) .

34. One of the inputs to any calculation of nominal torsional stress is the gas pressure tangential effort, or Tn values. PEI used Tn values provided by Lloyd's Register of Shipping in its calculations. Lloyd's Tn values were the high- r l- est Tn values published in an available common domain refer-ence. At the tinae PEI performed ~ its calculations, the FaAA Tn

) values were not available. In addition, PEI did not want to I

use private. data that it had not generated. In determining the i nominal stresses on a crankshaft, it is commonly accepted prac-i

! tice to use common domain Tn values such as Lloyd's. (Tr.

l 22,853-56 Chen). Lloyd's Tn values are higher than the Tn val- '

ues used by TDI for orders above number 4.0 and lower than the k

j i values used by TDI for orders below number 4.0. (LILCO Diesel ,

i Exhibit C-18 at 13) .

! .35. FaAA measured the pressures in the Shoreham EDGs to i

obtain a pressure versus_ time curve. This curve allowed FaAA to develop accurate Tn values for the replacement crankshafts.

(Tr. 22,814, 22,850 Johnston). FaAA's Tn-values are higher !

than Lloyd's Tn values. (LILCO. Diesel Exhibit C-18 at 13; LILCO Diesel Exhibit C-17 at 3-13) . However, even if PEI had

(

used FaAA's Tn values, the nominal stresses would still be i

safely within the DEMA allowable limits. (Tr. 23,035-26 Chen) . i f

i 4 ,

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

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

- 36. Stone & Webster performed torsiograph tests on the ,

replacement crankshaft in EDG 103 in January, 1984. The i

j' .torsiograph tests measured the total torsional vibrations re .

culting from all orders. FaAA converted the torsional vibra-tions into stresses for comparison with DEMA. (Johnston, ff ;

1 Tr. 22,610 at 24; Tr. 22,814 Johnston). '

J

37. The torsiograph provides the angular displacement-i response of the free end of the crankshaft as a function of

1 l

time. This displacement may be decomposed into components cor-l responding to each order. The torsiograph also provides the peak-to-peak response. These responses are used to calculate I

i the nominal stresses. (Johnston, ff Tr. 22,610 at 26).

I 38. In order to convert the amplitude of free end rota-l l tion into nominal shear stresses, those measurements must be J

l multiplied by a f actor of 9562 psi per degree. This number r (9562) is a stress that occurs on the crankshaft by applying a I

i rotational displacement of one degree at the free end of the

! crankshaft, assuming that the shape of the crankshaft is in the

.! first mode of vibration. (Tr. 22,837 Johnston).

! 39. In converting torsiograph data into nominal stress, j Lt is customary to assume a single mode of response. In con-1

verting the measurements, FaAA assumed a first mode of re-sponse. This type of approach is found in many common text-

! books and is the approach used by A88. (Tr. 22,838, 22,850 ;

! Johnston).

i l !

i i

_ _ __ . _ _ _ _. ... _--. _ --._ -_. _ .. .____...--_-_._..___._.D

T i

i

! 40. The conversion factor of 9562 psi was calculated '

using.the first mode of response. It would be possible to cal-culate a similar conversion factor using the second, third or ,

any other mode of response. The replacement crankshafts, hcw-

{

ever, vibrate primarily in the first mode. The conversion fac-tor based on the first mode of response was used because it represents a customary way of reducing torsiograph test data.

i -

i 1

t (Tr. 22,838-39 Johnston). t

$ 41. While the principle of using the first mode of re-

't

[ sponse to reduce torsiograph data is common, the principle of l i 1 using a half peak-to-peak is a very conservative approach.to l reducing torsiograph data. Much of the data in the past has

I i

been reduced based on the square root' of the sum of the squares-

! (SRSS) of individual orders. If the SRSS method had been used I to convert the torsiograph data'into stresses the result'would

}

l have been in the range of 4000 psi. (Tr. 22,839 Johnston).

I

42. The torsiograph data showed that the single order At I and combined stresses are below the DEMA allowable limits.

1004 load (3500 KW) the fourth order stress is 3108 psi and the combined stress is 6626 psi. At overload (3900 KW) the fourth f

i

! order stress is 3287 psi and the combined stress is 6958 psi.

l (Johnston, f f Tr. 22,610 at 26; LILCO Diesel Exhibit C-17 at i

i 2-11).

I i i

{

r

l I

43. Suffolk County presented no evidence to show whether or not the replacement crankshafts complied with DEMA.

Professor Christensen did not perform any forced torsional vi-l bratory calculations. (Tr. 23,966 Christensen). Mr. Eley is not capable of performing forced torsional vibratory calcula-I tions. (Tr. 23,968 Eley). Neither Professor Christensen nor Mr. Eley have had any experience with DEMA prior to this case.

(Tr. 23,975-76 Christensen, Eley).

IV. The Crankshafts Have Been Approved by ABS

44. ABS has certified that the material properties of the crankshaf t conform to the requirements for ABS grade 4 ,

steel. (LILCO Diesel Exhibit C-12) .

45. ABS has approved the dimensional sizing for the diameter of the pins and the journals of the replacement crank-chaft and has approved the proportions of the crankshaft webs.

(Montgomery, ff Tr. 22,610 at 17). The NRC Staff also agrees that the sizing of the pins, journals and webs comply with ABS

, rcquirements. (Henriksen, Sarsten, ff Tr. 23,126 at 11; Staff Diesci Exhibit 1).

46. Neither Suffolk County nor the NRC Staff believe

, the replacement crankshafts comply with ABS torsional require-COnts. ABS, however, has determined that the crankshafts com-ply with ABS torsional requirements. (LILCO Diesel Exhibit C-13).

1 17-

_ _m __, _ m s _ _. , , . _ _ ,

l i

47. TDI submitted information to ABS concerning the !

torsional critical speed arrangment of the replacement crank-chaf ts at Shoreham. On the basis of reviewing the information submitted by TDI and performing its own check calculations, ABS I

) cpproved the critical speed arrangement of the crankshaft, flywheel and generator at Shoreham. (Montgomery, f f Tr. 22,610 at 17-18; LILCO Diesel Exhibit C-13).

48. ABS summed only two orders when it performed its check calculations for torsional stress. ABS used the SRSS 1 method to calculate torsional stress. (Tr. 22,738 Johnston; Suffolk County Diesel Exhibit 43 at 24-28).

i V. Fatique Analysis

49. FaAA performed a fatigue analysis to determine the true margin of safety of the replacement crankshafts. This analysis was independent from the design criteria specified by any code and shows that the crankshafts have an adequate factor of safety. (McCarthy, Johnston, ff Tr. 22,610 at 31).

50. In order to determine a factor of safety for the replacement crankshafts it is necessary to know two things:

(1) the maximum stress the crankshaft will see in service; and (2) the endurance limit of the crankshaft material. FaAA's fa-tigue analysis established both of these factors. (McCarthy, Johnston, ff Tr. 22,610 at 32).

1 . - . _ _ _

l-l

l

)' -

i

51. The maximum stress on the crankshaft occurs in the

! fillet regions. The actual maximum stresses were measured dur- !

ing strain gauge testing on the original crankshaf t in EDG 101 l j in September, 1983, and on the replacement crankshaft in EDG j 103 in January, 1984. (Tr. 22,814, 22,889 Johnston).

j 52. The crankshafts were instrumented with strain j

' gauges at the locations of maximum stress. The location of 4

f maximum stress was determined analytically by FaAA. First, t

]

FaAA performed a dynamic torsional analysis of the crankshaft ,

! to determine the true range of torque at each crank tnrow. ,

L

! Second, using the results of the dynamic torsional analysis, a I !

finite element model of a one quarter crank throw was used to l ,

l compute the magnitude and location of peak stresses in the fil-1 I let region. The calculated peak stresses corresponded closely f with the measured peak stresses. (Tr. 22,889-93 Johnston; f

4 McCarthy, Johnston ff Tr. 22,610 at-32).

53. The fatigue endurance limit for the replacement crankshafts was established by first obtaining the endurance

! limit for. the failed crankshaf ts, and then increasing that i

i i

limit to reflect the difference in ultimate tensile strength ~

1 l

l between the failed and replacement crankshafts. (McCarthy, j l

! Johnston, if Tr. 22,610 at 32). f I

} $4. FaAA developed a dynamic torsional model of the i ?

I crankshaft to determine the total torque at each crank throw. !

i i

l i ;

n

i i

, e i, _

_.- _ ,~_,_,_ _ __.m.

. _ _ _ _ _ . . _ , ~ , _ _ , - , . _ - , _ ' l. _ . _ - - . _ . , - . _ , - _

t

The total torque is calculated by a summation of the torque -

l i

produced by each order and mode. The analytical method used by j FaAA (modal superpositon) computes the phase relationship be-tween the various orders and modes, which permits this summa- l tion. (Johnston, ff Tr. 22,610 at 33).

4

55. One of the inputs into FaAA's dynamic torsional l cnalysis was the Tn values based on pressure measurements taken ,

by a quartz transducer placed in the air start valve of cylin-der No. 7 on EDG 103. Cylinder-No. 7 was chosen for the pres-1 '

oure measurement because strain gauges were placed on crankpins f

i j No. 5 and 7 and FaAA wanted a pressure measurement on a .corre-sponding cylinder. Number 7 was chosen over No. 5 because i

pressure diagrams are typically more accurate the closer the <

]

j cylinder is to the location where the top dead center marker i f was measured at the flywheel, and the nearest cylinder that was a

f: strain gauged was No. 7. (Tr. 22,866-67 Johnston).

I

56. Cylinder No. 7 was not chosen for pressure measure- -

f ment because of any prediction that the prousure would be the 1

i highest in that cylinder. The engines are typically balanced so that the cylinder pressures are approximately equal j

throughout all of them. FaAA sought neither to find the high-est nor the lowest pressure measurement. (Tr. 22,868 1 Johnston).

j ,

I l

t 6

l

57. The type of pressure measurement that is of concern o

for a torsional' analysis is not a peak pressure. What is im-portant is an entire pressure curve. It is important that the i pressure curve be typical, because vibrations do not respond to

( one individual peak pressure, but rather to an accumulation of

\

a series.of loadings. That is what causes vibrations to build l ,

above a certain level, and is the reason for performing a dy-f i namic rather than a static analysis. The measurement from cyl-inder No. 7 was taken over many cycles and then averaged in j

order to calculate an appropriate pressure curve. (Tr.

22,869-70 Johnston).

58. Professor Sarsten agrees that FaAA's Tn values are accurate and used them in his harmonic sythesis calculations.

[ (Sarsten,.f f Tr. 23,126 at 13).. FEV-derived its Tn values from i

the-same pressure data used by FaAA. (Tr. 22,810-12 I

! Pischinger).

i

59. The total torque calculated by the dynamic tor-sional model was'used as input to the finite element model to >

determine the actual maximum value and location of stress in the fillet regions. (Johnston, ff Tr. 22,610 at 33).

j

60. The nominal stresses calculated from the dynamic model are considerably less than the ac#tual. maximum stresses,'in d

the crankshaft. These nominal values would prevail if the i

j crankshaft were a long circular cylinder. Stresses in the real i

i !

l -

t r

, - , - . - , , , . - - - - . , - , - ~ - - , rw,- -r--, ,-.,e-n , -~.-----.--..,,,w.,-a,,-w.--,.,-.c,, -,.-n..,,v.- -.--,-, . , + - -r.n-- --- ,.

l I

l crankshaft are greatly influenced by its complex geometry and 1 by stress concentrations, especially at the fillet radii be-In tween the main journal and web, and the crankpin and web.

codition, a crankshaft throw is subjected to loads of two basic l types: (1) torque transmitted through the throw, which is in-I fluenced by the output power level and by the torsional vibra-j tion of the crankshaft; and (2) connecting rod forces applied (Johnston, ff to the crankpin and reacted at bearing supports. '

l Tr. 22,610 at 33-34).

} 61. FaAA used a finite element model of a one quarter j

i crank throw, considering stresses due to torsio'nal loading and t

stresses due to gas pressure loading, to compute the actual maximum value and location of stress in the crankpin fillet i~

l area. The strain gauges used during dynamic testing were

! placed at the location of maximum. stress calculated by the fi-h nite element model. (Johnston, ff Tr. 22,610 at 34).

62. FaAA's finite element analysis predicted that the 1

j-highest stresses would occur in the fillet regions of crankpin i

Nos. 5 and 7. Two different sets of boundary conditions were used in the finite element analysis. For a determination of the stresses in the crankpin fillet area due to torsional stresses alone, the stresses calculated at the two boundary i

conditions would be expected to bracket the measured stresses.

The stresses measured at crankpin No. 5 are bracketed by the ,

i l 1 I

{

i

'r

two finite element models. This is to be expected because the l stresses on crankpin No. 5 are due almost exclusively to tor-sion. (Tr. 22,891 Johnston; LILCO Diesel Exhibit C-17 at 3-17). In regard to crankpin No. 7, the range of principal

! stress is bracketed by the two boundary conditions, although the range of equivalent stress f alls outside the bracket by ap-proximately one and a half percent. (LILCO Diesel Exhibit C-17 at 3-18). This is due to the fact that on crankpin No. 7 there is a small amount of bending stress. The discrepancy is so small, however, that it was not necessary to perform additional analyses using boundary conditions suitable for a bending anal-ysis. (Tr. 22,891-92 Johnston).

63. FaAA performed calculations to compute the maximum 4 bending stresses in the crankshaft. The maximum stress in any crankpin due to bending is 15.5 ksi. The point at which the maximum bending stress occurs is in a different location than the point of maximum torsional stress. The location of maximum bending stress is at the bottom of the crankpin when the pin is at top dead center. The location of maximum torsional stress occurs approximately 45-50* around the crankpin away from the

< bottom of the pin. The maximum bending stress also occurs at a dif ferent time than the maximum torsional stress. The result is that the maximum stress that occurs on the crankshaft, which

is the stress of concern when determining a factor of safety, occurs on crankpin No. 5. (Tr. 22,893-94 Johnston).

M i 64. On crankpin No. 7, there is a small overlap in time l

l between.the occurrence of the bending stress and the occurrence of the secondary peak of torsional stress, which causes the range of equivalent stress to be 44.5 kai. This is the number that f alls slightly outside the two finite element models.

This number is, however, significantly lower than the range of cquivalent stress of 49.3 kai on crankpin No. 5. (Tr.

22,894-95 Johnston; . LILCO Diesel Exhibit C-17 at 3-17 and 3-18).

65. The amplitude of equivalent stress is half the range of equivalent stress. The highest range of equivalent l

stress was on crankpin No. 5 (49.3 kai). The amplitude of equivalent stresa ;ts 24.6 ksi. This amplitude is the value used in the Goodman diagram for purposes of comparison with the endurance limit. (LILCO Diesel ' Exhibit C-17 at 3-9, 3-32) .

66. FaAA's finite element analysis was merely a step in calculating 'a factor of safety for the replacement crankshafts.

The factor of safety was calculated from the stresses measured in the replacement crankshafts. The finite element calcula-tions were performed to demonstrate the location where the strain gauges should be placed on the crankshafts. (Tr. 22,892 Johnston).

67. The strain gauges were placed in the locations of maximum stress that were indicated both around the

circumference of the pin and within the fillet. While the dis- .

l tribution of principal stresses varies by a considerable amount l

l between the two bounding finite element cases, the location of l I

maximum stress is the same under both conditions. It is only the location of the maximum stress that was used as input to the strain gauge tests. (Tr. 22,892-93 Johnston; LILCO Diesel r

i Exhibit C-17 at 3-27 through 3-30) .

68. The typical procedure for running a strain gauge test on a crankshaft is to bring the engine to the load of in-i terest and maintain it at that load for approximately ten i

minutes to stabilize the engine. Data is then taken for ap-proximately two minutes. It is not necessary to stabilize the load for more than ten minutes when taking measurements on a crankshaft because the torsional vibration condition stabilizes very rapidly. The torsional vibration conditions are not de-pendent on temperature transients or other phenomena that might take a long time to stabilize. This test procedure is also typically followed in taking torsiograph data. (Tr. 22,976 Johnston).

69. The next step in FaAA's fatigue analysis was to compare the measured stresses with the fatigue endurance limit of the replacement crankshafts. The fatigue endurance limit of the replacement crankshaft was established by first obtaining i the endurance limit of the failed crankshaft. Since the i !

cndurance limit scales linearly with ultimate tensile strength, the endurance limit of the replacement crankshaft was increased l to reflect the difference in ultimate tensile strength between l the failed and replacement crankshaft. (Johnston, ff Tr.

22,610 at 36).

70. The original 13-inch by ll-inch crankshaf t on EDG 101 was instrumented with strain gauges in the fillet location i

of crankpin No. 5. This fillet had previously experienced a fatigue crack during testing. After the test, the three-dimensional finite element model of a quarter section of 2

a crank throw showed that the strain gauges were placed close The measured stress range to the location of maximum stress.

was used to establish the endurance limit in this analysis as a conservative assumption, although the actual maximum stress range was revealed by the finite element model to be about 15%

higher at a nearby location. The original crankshaf t on EDG 102 had experienced 273 hours0.00316 days 0.0758 hours 4.513889e-4 weeks 1.038765e-4 months at equal to or greater than 100%

l load, or about 4,000,000 cycles. By using linear cumulative i

' damage techniques, FaAA determined that the endurance limit for the original crankshafts was 36.5 kai. (Johnston, ff Tr.

{ 22,610 at 37).

71. The fatigue endurance for the. replacement crank-shafts is 39.2 kai. This is higher than the fatigue endurance j limit for the original crankshaf ts because the ultimate tensile '

f 26- l 4

. . . , . _ . _ -. ,,__._-_,._,___c_,.,_m_. , -

, - - , . , ..r m_,. ... _ - .

)

i

\

strength of the replacement crankshafts exceeds the ultimate tensile strength of the original crankshaf ts, and the fatigue endurance limit is directly proportional to ultimate tensile strength. (Johnston, ff Tr. 22,610 at 37).

72. FaAA calculated the factor of safety against fa-tigue f ailure by plotting the amplitude of equivalent stress (24.6 kai) on a Goodman diagram constructed using the fatigue endurance limit and the ultimate tensile strength values for

) the replacement crankshafts. The factor of safety against fa-tigue failure is 1.48, without taking into account any benefi-cial effect of shot peening the fillet regions. (Johnston, ff Tr. 22,610 at 38: LILCO Diesel Exhibit C-17 at 3-32) .

73. A factor of safety of 1.48 provides sufficient as-surance that the replacement crankshafts are adequate for their I intended service in the Shoreham EDGs. (McCarthy, ff 22,610 at 38).

74. A factor of safety is an additional margin of f

strength, in either the fatigue strength (endurance limit),

yield strength, or ultimate strength, that is added to a me-chanical design to compensate for uncertainties. There is sig-nificant confusion often generated by a failure to identify whether a stated factor of safety is with regard to fatigue or endurance limit, yield, or ultimate strength. The factor of safety with regard to these three different failure modes will e

,- - - -- m--

1' i

(McCarthy, generally be different for the same design or part.

! .ff Tr. 22,610 at 38-39).

75. A factor'of safety in endurance limit is the factor i

of strength the part or design has over that required for the part to be expected to exhibit infinite life, or a life of some specified number of cycles in repeated or cyclic loading. A .

I f actor of safety in yield is the factor the yield strength of the part is greater than the expected service load. Similarly J

the factor of safety in ultimate strength or overload failure is the factor the breaking strength of the part is greater than l

the expected service load. In older design references it is ,

not uncommon to see a very large factor of safety in overload 1

i recommended, and no mention of a factor of safety in endurance limit or fatigue strength, for parts that were cyclically loaded and could fail in fatigue. This was before fatigue and j: stress concentration effects were as well understood as they i are now. (McCarthy, ff Tr. 22,610 at 39).

76. A factor of safety is an allowance for uncer-I 1

I tainties as to service load, material properties, stress con-centration factors, lifetime, etc., which are directly related i

b to the amount of testing, analysis, and understanding a design-er has of a particular part and its service environment.

! (McCarthy, ff Tr. 22,610 at 39-40).

1 1

\

l l -2s-l i

l

1 l

77. An acceptable factor of safety is determined by the

' degree of uncertainty and the difficulty or panalties of adding cdditional strength to the design. Where the design envelope end the nature of the fabricated part are c'easonably under-ctood, a f actor of safety in fatigue or cyclic loading of 1.3 to 2.0 is generally recommended. When the uncertainty of de-cign f actors is greater, higher values will be recommended.

Some design texts will recommend that, if the designer is 'seri-

- ously considering a factor of safety of greater than two, he chould devote additional time to analyzing the design, rather than accepting the ignorance which is causing him to select a higher factor of safety. A factor of safety of 1.48 in fatigue or endurance limit will produce a much higher factor of safety

~

with regard to yielding or overload failure. (McCarthy, ff Tr.

22,610 at 40).

! 78. The design of the replacement crankshaf ts is under-stood extremely well. Information has been gained from the

! failure of the original crankshafts, full scale instrumented i

tests of the actual service loading, material strength tests for the individual parts, torsiograph testing, and extensive a

three dimensional analytical modeling of the structure. The crankshaft is being run in a temperature controlled, oil filled ,

environment. It is completely guarded from accidental and unanticipated impact by foreign objects by the engine block.

.Usually a designer t.as far less information to work with when .

Essessing a design. This results in uncertainities in the de-l cign being reduced substantially. (McCarthy, ff Tr. 22,610 at 41).

I 79. For well understood designs operating in environ-ments that are noe severe, a factor of safety in fatigue or en-durance limit of 1.3 to 1.5 is generally accepted. For the re-4

! placement crankshaf ts the degree of understanding permits the

use of a safety factor at the lower end of this range, when in fact the actual safety factor is at the high end. (McCarthy, ff Tr. 22610 at 41).

80. The factor of safety of 1.48 is more than accept-I able because of the extensive knowledge concerning the design of the replacement crankshafts. All three of the original crankshaf ts failed at the point they were predicted to fail by i

I FaAA's analytical model. In addition to the analytical model, f there is a dynamic model, which permits prediction of the vi-I beations and deflections of the moving crankshaft. The dynamic i

model has been verified by torsiograph measurements on the 4

l crankshaft. There is a finite element model of the crank throw for both the old and new crankshaft, which has been verified by i

I measurements on the old and new crankshafts in operation. In

! addition, the analytical model predicts the original crankshaf t f ailure and the replacement crankshaf t survival by a wide l

margin. (Tr. 23,027-28 McCarthy).

1 l

1

I

81. Factors of safety are based on a comparison of the l

knowledge of the design and the uncertainty about the expected i cervice. In the case of the replacement crankshafts, the mar-gin of safety is 1.48.and there is extremely detailed knowledge tbout the expected service. This knowledge allows a confident conclusion that the crankshafts are capable of unlimited op-eration at 3500 KW and 3900 KW. (Tr. 23,030-31 McCarthy).

VI. Kritzer-Stahl l

4 i 82. The Kritzer-Stahl design criteria is a method for evaluating the adequacy of a crankshaft by calculating a factor of safety. The method compares calculated stresses with calcu-lated endurance limits, which permits the calculation of a fac-tor of safety. The crankshaf t design and engine operating con-l ditions are used as inputs to calculate the stress levels.

(Tr. 22,767 Pischinger).

83. The' original research on which the Kritzer-Stahl j criteria is based was performed prior to 1961. The criteria l has been updated periodically. Additional research that has i

been conducted since 1961 indicates that the Kritzer-Scahl criteria is very conservative. For example, the stress concen-j

.tration factors calculated by the more recent Lejkin method are lower than the stress concentration factors calculated by the Kritzet-Stahl method. (Tr. 22,769, 22,772, 22,775 Pischinger).

F

,, , , e - . , . . _ - . . _ - , ,r.-_. -,,, . _ . . _ . . , . . _ . .

84. There is a built-in factor of safety in the Kritzer-Stahl criteria of approximately 22%. This is demon-strated by the' fact that the Kritzer-Stahl criteria predicts that the original 13-inch by ll-inch crankshaft should have failed after 150 hours0.00174 days 0.0417 hours 2.480159e-4 weeks 5.7075e-5 months at full load (about 2,000,000 cycles).

In fact, the original crankshafts did not fail until 273 hours0.00316 days 0.0758 hours 4.513889e-4 weeks 1.038765e-4 months of o'peration at_ full load (about 4,000,000 cycles). The origi-nal crankshafts actually had twice the life predicted by Kritzer-Stahl. (Tr. 22,776-77 Pischinger).

85. The calculated endurance limit for the replacement crankshafts under the Kritzer-Stahl criteria is 25.4 ksi. The calculated maximum stresses in the fillet radius using the-Kritzer-Stahl criteria is 24.9 ksi. (Tr. 22,790-94 ,

Pischinger).

86. There is a very close correlation between the Kritzer-Stahl predicted maximum stresses (24.9 ksi) and the maximum stresses measured during dynamic testing (24.6 ksi).

(LILCO Diesel Exhibit C-17 at 3-9). The endurance limit calcu-lated by the Kritzer-Stahl method (25.4 ksi) is extremely- con-

servative when compared to the actual- endurance limit deter-mined by FaAA (39.2 ksi). (Johnston, ff Tr.'22,610 at 37).

87. The ratio of the endurance limit calculated by-the Kritzer-Stahl method to the maximum stress calculated by"the~

Kritzer-Stahl method is 1.02. ._ Combining that margin with the s

i "1 l I

~ ' ~

built-in 22% margin in Kritzer-Stahl provides a factor of safe-(Tr.

ty of 24% at full load under the Kritzer-Stahl criteria.

.23,004 Pischinger).

88. The Kritzer-Stahl criteria is highly accurate for predicting maximum stresses. The criteria predicts extremely conservative endurance limits, however. This is demonstrated

~

by the f act that the original crankshaf ts had twice the life predicted'by Kritzer-Stahl. This is the main conservatism of the Kritzer-Stahl criteria. (Tr. 23,006-07, 23,045-46, 22,776-77 Pischinger).

89. The range of acceptable factors of safety in con-temporary European diesel industry practice is from 1.15 (15%)

to 1.30-(30%). What is acceptable for a factor of safety depends on how much information is available about the crank-shaft. If there have bee 7 actual test measurements from the crankshaft and there-is information from previously. failed crankshafts, a lower factor of safety is acceptable; The less information that is available, the higher the f actor of safety should be. (Tr. 23,012-13 Pischinger). ;

90. The replacement crankshafts have a 24% margin of l

safety at full load and are capable of unlimited operation at 3500 KW. In addition, the replacement. crankshafts are capable of operating for at least 1200 hours0.0139 days 0.333 hours 0.00198 weeks 4.566e-4 months at 3900 KW.. The 1200 hour0.0139 days 0.333 hours 0.00198 weeks 4.566e-4 months figure does not include any allowance for the inherent safety e

l l

l 1

1

k t, l

factor in Kritzer-Stahl of 22%. If the inherent conservatism is considered, the crankshaf ts have a safety margin of 15% for operation at 3900 KW. (Pischinger, ff Tr. 22,610 at 5; Tr.

i 22,792-93, 23,037-38 Pischinger).

91. The crankshafts will never operate at 3900 KW dur-ing a LOOP LOCA. The engines are expected to operate for no more than 60 hours6.944444e-4 days 0.0167 hours 9.920635e-5 weeks 2.283e-5 months at 3900 KW during testing over the 40 year life of the plant. This is only one twentieth (1/20) of-the crankshafts' minimum life at 3900 KW. (Youngling, ff Tr.

22,610 at 5).

4 VII. Shot Peening

92. Shotpeening is a surface cold-working process that

- is used primarily to lengthen fatigue life and prevent cracking-of metal parts. Shotpeening is also used to shape parts, over-come porosity, work harden surf aces, protect against stress corrosion or corrosion fatigue and for many other purposes. A l crack will not initiate in, nor propagate through a compressed layer. As nearly all fatigue, stress corrosion and corrosion f atigue failures originate at the surf ace of a part, . the layer of compressive stress induced by: shotpeening produces a signif-icant' increase in the endurance limit.- The maximum compressive-residual stress produced at or near the surface is at least as great as one-half (1/2) the ultimate tensile. strength of the l

l l

l ;

, - v ,_- , - - - - -

material. Shotpeening is used to eliminate failures in exist- ,

ing designs, or to allow the use of higher stress levels.

(Cimino et al, ff Tr. 23,122 at 6-7).3/

93. The replacement crankshaft fillet areas were l

l shotpeened to reduce the mean surface tensile stresses and place the fillet surfaces in compression. Shotpeening makes the surf ace less susceptible to handling damage, such as the score mark where cracking initiated on the original crankshaf t in EDG 102. Additionally, shotpeening eliminates machine im-perfections, which prevents initiation of cracks on the ma-chined fillet surface. Shotpeening provides a higher endurance (Wells, Seaman, limit for the fillet area and the crankshaft.

ff Tr. 23,122 at 5-6).

94. Two of the replacement crankshafts were initially shotpeened by TDI. Examination revealed that the shotpeening was inadequate. There were areas where coverage was only 80%

to 90% and not all peening intensity tests (Almen strips) were accounted for. There was no concern that the TDI shotpeening had damaged the crankshafts. (Wells, Seaman ff Tr. 23,122 at 7).

3/ The prefiled testimony of Messrs. Wells, Johnson, Wachob, Seaman, Cimino-and Burrell was not numbered when bound into the transcript. The testimony may be found following page 23,122 of the transcript.

95. Metal Improvements Company (MIC) was hired to l

re-peen the crankshafts. The crankshafts were placed on pedes-

[ )

f tals or stands which allowed rotation of the crankshafts so that all fillet areas could be completely saturated with. shot.

The crankshafts were washed with a chemical solution to remove ,

all traces of oil or other preservatives and the areas on both sides'of the fillets were taped. A tent was set up over each of the crankshafts so that shot could be contained within the tent. In addition, Almen strips were set up for measuring shotpeening intensity. Almen strips are flat pieces of metal which are clamped to a solid block and exposed to a stream o f 4

shot. Upon removal from the block the Almen strip will be curved. The curvature will be convex on the. peened side and the height of the curved arc is measured on a special Almen

-gauge which serves as a measure of the intensity. A .008 .010 C strip was utilized for the Shoreham replacement crankshafts 1- which provides surf ace compression to a depth of .027" .034" on ASTM A-668E metal such as the replacement crankshafts. While MIL Spec. No. 13165B required intensity to be checked by Almen

i. strips every eight hours of peening, MIC, in fact, checked peening intensity every four hours of actual peening. In addi-tion, the shot was screened and examined under a microscope

~

'every two hours to ensure uniformity of shot size and shape.

(Cimino, ff-Tr. 23,122 at 8-9).

1

96. MIC utilized a patented process called "peenscan,"

approved by USA Military Specification, MIL - 13165-B, Amend-ment 2, to ensure uniformity and full-coverage on the area being shotpeened. The area being shotpeened is coated with a flourescent dye-type liquid prior to the shotpeening and al-lowed to dry. All areas covered with dye will show a green glow under a blacklight. After shotpeening is completed the area is placed under blacklight to see if any green glow re-If any glow remains the coverage is not 100%. In this mains.

case all fillet areas were checked for any green glow and peened until all traces of the dye were completely gone.

(Cimino, ff Tr. 23,122 at 10).

97. The shotpeening MIC performed on the two replace-ment crankshafts was in accordance with MIL Spec. No. 13165B and placed the surface stresses in the fillet area of the crankshaft in compression. (Cimino, ff Tr. 23,122 at 11).

98. The two crankshafts shotpeened by TDI were sub-l jected to magnetic particle testing af ter machining by the man-In addition, ufacturer and no relevant indications were found.

when the two crankshaf ts were received at Shoreham, both shaf ts were subjected to visual examination, magnetic particle testing and liquid penetrant testing. This examination and testing re-(Wells, Sea-

vealed no relevant surface cracks or indications ~.

man, ff Tr. 23,122 at 12; Bush, ff Tr.-23,126~at 19). Thus, I

_37_

r,. ,, - ,, , . , , , - - -

- - - - - - -- n

- s the as-received surface condition of the replacement crank-shafts shotpeened by TDI was acceptable. (Bush ff Tr. 23,126 at 19).

99. The photographs of the re-peened fillet areas that

! were reviewed by Franklin Research Center and referred to in its report dated April 6, 1984 are representative of all crankpin and. main journal fillet shotpeening. As a result of MIC's work, the peening is uniform, equally dimpled, and the shotpeening at all fillet areas looks exactly as it does in these photographs. (Wells, Seamen, ff Tr. 23,122 at 12).

100. The nondestructive examination of the TDI-peened fillet areas revealed no surface indications or deficiencies which could reasonably be expected to cause a " stress nuclea-tion site." Even if there had been surface " stress nucleation sites," proper repeening of the fillet areas would correct or eliminate any such problem. (Burrell, ff Tr. 23,122 at 13).

101. The re-peening by MIC would have corrected or eliminated any " stress nucleation sites" that may have existed rather.than masking them. Any surface'" stress nucleation site" small enough to escape detection by magnetic. particle testing l and/or liquid penetrant testing-would be eliminated as a result s

of the plastic flow of the surface metal caused.by the re-peening. (Wells, ff Tr. 23,122 at 13-14).

e

- - - . _ . - . ,I

102. The surface stresses in the fillet areas of the l

Shoreham replacement crankshafts have been placed in compres-l sion and any cut, scratch, flaw, machine mark, etc. no deeper than the compression area itself, will not be the initiation point of a fatigue crack. Any undesirable effects of the pre-vious shotpeening have been corrected. (Burrell, Wells, Wachob, ff Tr. 23,122 at 14-15).

103. The possibility that shotpeening increased the likelihood of subsurf ace fatigue cracking is quite remote. In almost all instances, f atigue cracks such as occurred in the original crankshaf ts initiate at external surface areas.

Subsurf ace f atigue cracking is very unusual and requires the presence of a significant void or inclusion at a given stress state for initiation of the fatigue crack. There is always the possibility that any cast or forged piece of metal may contain a subsurface. inclusion or void. The only protection against this risk or possibility is the manufacturer's quality control procedures for the melting, casting and forging ~ processes and its quality assurance procedures during and after the manufacturing process. The replacement crankshafts for the EDG's were manufactured by Krupp, and the forging and machining of these crankshafts were certified by ABS. Additionally, Krupp's ultrasonic testing and magnetic par'ticle testing, as

,well as LILCO's ultrasonic testing, magnetic particle testing l

t

- . - - . . =_

l and liquid penetrant testing, revealed no relevant inclusions or voids. This provides as much assurance as is possible that l

l no subsurface voids or inclusions of sufficient size to initi-ate a subsurface fatigue crack are present in the replacement crankshafts. (Burrell, Wells, Wachob, ff Tr. 23,122 at 15-16).

Subsurface fatigue cracking as a result of shotpeening would Such a not occur unless a large embedded flaw was present.

i flaw would have been detected by the extensive ultrasonic testing during fabrication. (Bush, ff Tr. 23,126 at_19-20).

104. In addition, after 300 hours0.00347 days 0.0833 hours 4.960317e-4 weeks 1.1415e-4 months of operation, of I which 100 hours0.00116 days 0.0278 hours 1.653439e-4 weeks 3.805e-5 months were at 3500 KW or above, the eight crankpin fillet areas of highest torsional stress on each of the three crankshafts were subjected to high resolution eddy current !

testing. The eddy current test was designed to detect cracks larger than 1/32" long by 1/64" deep. No cracks were found.

In addition, the eight crankpin fillet areas of highest tor-l sional stress were subjected to liquid penetrant testing after 300 hours0.00347 days 0.0833 hours 4.960317e-4 weeks 1.1415e-4 months of operation. No relevant indications were found.

f (Johnson, Wells, Seaman f f Tr . 23,122 at - 19) .

105. The crankshafts were subjected to more than one million torsional peak stress reversals during this period of Any " stress nucleation site" that had not been de-operation.

i tected by previous nondestructive testing would have initiated a fatigue crack that would have been detected by the high i

i~

e S

- r - - . _ _ , . _ _ . . . _ , . _ , , ,, . - - - -

resolution eddy current testing and/or liquid penetrant

! testing. (Wells, ff Tr. 23,122 at 19-20).

106. The benefits of shotpeening are attributed to the residual compressive surface stress. This region, although small in respect to the crankshaf t diameter, is significant l

with regard to preventing the initiation of a fatigue crack in the surface. Given the residual compressive stresses and the actual operating stresses in the fillet region, a fatigue crack will neither initiate in the fillet area nor will any flaw or defect contained within the shotpeened volume propagate. There is no equation between the hardened depth of shotpeening and its effective depth. (Burrell, Wells, Wachob, ff Tr. 23,122 at 20-21).

107. The cathode-anode electrochemical principal is 4

not operative upon the replacement crankshafts in the Shoreham EDG's because it requires a driving energy that is not present, the presence of electrolytes, which do not exist within the crankcase of the Shoreham diesels. (Burrell, Wells, Wachob, ff Tr. 23,122 at.18-19; Tr. 23,182-83 Wells, Wachob).

108. In order for heat to appreciably affect. residual stresses caused by shotpeening,-temperature levels of at least 500* F must be attained. This temperature is completely unattainable within the normal operating limits of the Shoreham diesels. The crankshaft temperature is between 200* - 240' r i

- under normal operating conditions. Under unusual circumstances (Burrell, Wells, the temperature may go as high as 260* F.

Wachob, ff Tr. 23,122 at 21).

109. The shotpeening of the replacement crankshaft fil-let areas has resulted in increasing their fatigue endurance limits by a minimum of ten percent (10%) and could conceivably have increased the fatigue endurance limits by as much as l

twenty percent (20%). (Burrell, ff Tr. 23,122 at 22; Tr.

l 23,158 Wells).

Respectfully Submitted,

' LONG ISLAND LIGHTING COMPANY E Af:lk f%.dsv. 12C' L uy.

Counsel' '

j E. Milton Farley, III

~

Hunton & Williams P.O. Box 19230 2000 Pennsylvania Ave., N.W.

Washington, D.C. 20036 T.S. Ellis, III i Hunton & Williams 707 East Main Street I

P.O. Box 1535

' Richmond, Virginia 23212 Odes L. Stroupe, Jr.

David Dreifus Hunton & Williams 333 Fayetteville Street P.O. Box 109 Raleigh, North Carolina 27602 DATED: November 5, 1984 4

4,-, , _ . _ . . -- . _ _ _ m___

a LILCO, November 5,1984 CERTIFICATE OF SERVICE In the Matter of LONG ISLAND LIGHTING COMPANY (Shoreham Nuclear Power Station, Unit 1)

Docket No. 50-322 (OL) 1 I hereby certify that copies of LILCO's PROPOSED FIND-INGS OF FACT were served this date upon the following by j first-class mail, postage prepaid, or (as indicated by aster-

isk) by hand'.

Judge Lawrence Brenner, Esq.* Atomic Safety and Licensing Appeal Chairman Board Panel Atomic' Safety and Licensing U.S. Nuclear Regulatory Commission Board, United States Nuclear Regulatory Commission Washington, D.C. 20555 4350 East-West Highway l Fourth Floor 1(North Tower) Martin Bradley Ashare, Esq.

! Bethesda, Maryland 20814 Att: Patricia A. Dempsey, Esq.

{

County Attorney Dr. Peter A. Morris

Suf folk County Dept. of Law Administrative Judge Veterans Memorial Highway.

Atomic Safety and Licensing Hauppauge, New York 11787 Board,' United States Nuclear Regulatory Commission Richard J. Goddard, Esq.*

4350 East-West Highway Office of the Executive Fourth Floor (North Tower) Legal Director Bethesda, Maryland 20814 U.S. Nuclear Regulatory J

Commission l Dr. George A. Ferguson* 7735 Old Georgetown Road j Administrative Judge Be thesda, Maryland 20814 i Atomic Safety and Licensing i Board Panel Alan Roy Dynner, Esq.*

j School of Engineering Kirkpatrick, Lockhart,

Howard University Hill, Christopher &

2300 6th Street, N.W. Phillips i Washington, D.C. 20059 1900 M. Street, N.W.

8th Floor '

Secretary of Commission

Washington, D.C. 20036 O.S. Nuclear Reguitory Commission Mr. Marc W. Goldsmith Washington, D.C. 20555 Energy Research Group 4001 Totten Pond Road

Waltham, Massachusetts 02154 l

- - . - . , , . .,.--,--,,.v--. --- -n,

Fabian G. Palomino, Esq.

MHB Technical Associates Special Counsel to the 1723 Hamilton Avenue Governor Suite K Executive Chamber, Rm 229 San Jose, California- 95125 State Capitol Mr. Jay Dunkleberger Albany, New York 12224 New York State Energy Office Agency Building 2 Jonathan D. Feinberg, Esq.

Empire State Building New York State Albany, New York 12223 Depart. of Public Service Three Empire State Plaza Stephen B. Latham, Esq. Albany, New York 12223

Twomey, Latham & Shea 33 West Second Street Robert E. Smith, Esq.

P.O. Box 398 Guggenheimer & Untermyer Riverhead, New York 11901 80 Pine Street New York, New York 10005 James B. Dougher ty, Esq.

3045 Porter Street Howard L. Blau Washington, D.C. 20008 217 Newbridge Road Hicksville, New York 11001 4

Ralph Shapiro, Esq.

Cammer and Shapiro, P.C.

9 East 40th Street New York, New York 10016

a. A ,

Hunton & Williams P.O. Box 109

- Raleigh, North Carolina 27602 DATED: November 5, 1984 t

~

-- ,----mm .-- . , - - -, , , - m-- - - .. - - - -,,+ -

LILCO, December 3, 1984 t

i UNITED STATES OF AMERICA j NUCLEAR REGULATORY COMMISSION Before the Atomic Safety and Licensing Board In the Matter of )

)

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

)

(Shoreham Nuclear Power Station )

Unit 1) )

LILCO'S REPLY TO SUFFOLK COUNTY AND STATE OF NEW YORK PROPOSED FINDINGS OF FACT HUNTON & WILLIAMS P. O. Box 109 Raleigh, North Carolina 27602 l

l i

, - _ , - ,__._---_m.. _ _ - - - . - --

I l

l TABLE OF CONTENTS 1

I. INTRODUCTION....................................... 1 II. DISCUSSION............ ............................ 2 A. Compliance wit the Rules of Classification Societies................................... 2

1. Relevan,ce of Classification Societies' Rules.................................... 2

2. Compliance with the Rules of Classification Societies................................ 4 4

(a) Lloyd's............................. 4 1

.(b) IACS................................ 9 (c) ABS Rules on Torsional Vibration... 12 4

(d) ABS web Dimensions................. 16 (e) The Crankshafts are Adequate under the Kritzer-Stahl Criteria............. 17 B. DEMA....................................... 20

1. DEMA is an Appropriate Standard......... 20

2. The Crankshafts Comply with DEMA........ 27 j C.- FaAA 's Fa tigue Analys is . . . . . . . . . . . . . . . . . . . . 33 D. Shot Peening............................... 38-4 i

A l

l l

, , .c4 , . . . . , _ . . _ ~ -.. . ~ - , . - . . _ _ - - , ~.

l

-l l

I. INTRODUCTION

1. These reply findings address the findings proposed by Suffolk County and New York State (Joint Proposed Findings).

! For convenience, this reply follows the format of the Joint Proposed Findings.1/ LILCO does not propose that the Board 1

coopt these reply findings. Rather, the reply findings demon-

' otrate that the findings proposed by the Intervenors should be l rejected. A surprising number of the findings are wholly unsupported by citations to the evidentiary record. (See, 0.q., Joint Proposed Findings 18, 28, 29, 34, 39, 42, 46, 50, 51, 55, 77, 86, 96, and 104-106). Others misrepresent or omit pertinent testimony. The Joint Proposed Findings also largely f

i ignore the cross-examination of witnesses for the County.and the Staf f and the direct testimony. of LILCO's witnesses.

2. Joint Proposed Findings 1-17 are not proposed find-l ings of fact but rather are summary conclusions in the form of ,

4 a proposed decision and are not supported by citations to the l

record. LILCO disag'rees with the conclusions in Section I of l ,

l 1/- LILCO disagrees with some of the findings proposed by' the

~

j Rhc Staff, although LILCO agrees with many of the Staff's Pro-posed Findings. To avoid repetition, LILCO does not separately discuss the Staff Proposed Findings with which it disagrees, l oxcept in certain isolated instances.

f

i i

the Joint Proposed Findings. However, because this Section merely contains unsupported conclusions, LILCO will not specif-ically address each paragraph of Section I of the Joint Pro-L posed Findings.

l l

l II. DISCUSSION 4

! A. Compliance with the Rules of Classification Societies 4

1. Relevance of Classification Societies' Rules

3. Joint Proposed Findings 19-27 argue that the rules of various marine classification societies (Lloyd's, ABS and IACS) are relevant in determining whether the replacement I crankshafts in the Shoreham EDGs are adequate. These proposed findings are argumentative and conclusory and are often unsupported by any citation to the transcript. In addition, t

they ignore the fact that current Nuclear Regulatory Guides \

i l

! provide that standby emergency diesel generators need only com-l ply with DEMA. (LILCO Proposed Finding 9; Staff Proposed Find-ing 17).

i j 4. Joint Proposed Finding 23 concludes that if a )

i crankshaft in a diesel generator at a nuclear' plant does not i l

l

, l

! I i

i 1

-w- , r-,,-,---n.- ---,--,-,-w,- - - - - , . . , , ---,-n -

e,, - - - - , --- --w- , - ----.-e-- ,---- --

l I

comply with the rules of the classification societies, the l

! crankshaft does not satisfy NRC regulations. There is no tran-script citation for this conclusion nor is there any indication which NRC regulat'icns the Intervenors claim the crankshaf ts fail to meet.

5. Joint Proposed Findings 24-27 discuss the' alleged i similarities between marine diesels and diesels in nuclear standby service. These proposed findings are misleading for several reasons. First, the findings ignore testimony by LILCO '

and the Staff that operating conditions for marine diesels are more severe than operating conditions for nuclear standby die-

sels. (LILCO Proposed Finding 15) . Second, even if marine diesel generators (as opposed to main propulsion diesel en- ,

gines).are not subjected to more severe operating conditions than land based diesels, the rules of the classification soci-eties for designing crankshafts make no distinction between main propulsion diesel engines and diesel engines used for-electrical generation. (Joint Proposed Finding 27). While the service seen by a marine diesel generator may be similar to-the service seen by-a land based diesel, marine diesel genera-l '

tors are. designed to the same standards as main propulsion die-sel engines. The County's witnesses concede that main propul-l sion engines are subject to more severe stresses than-land i

. l t

t-

+ ,,, - , - - - - - - ,, , , - , , -. ., , , - - - - - - . - - - , - a- , ,--,,, - , -------g- - +-,n,-- - -

l l

based diesels. (Joint Proposed Finding 26; Tr. 23,984-85 i Eley). It is clear that the design standards of the classifi-cation societies are intended to apply to engines subject to 1

core severe operating conditions than the Shoreham EDGs. It is not appropriate to evaluate a land based diesel according to I

these standards. The fact that marine generators may be sub-i ject to similar operating conditions as land based diesels is -

irrelevant.

i j 2. Compliance with the Rules of Classification Societies i

(a) Lloyd's i 6. Joint Proposed Findings 29-33 discuss the County's

! horsepower calculations under Lloyd's rules. These proposed i

findings completely ignore the extensive cross-examination of the County's witnesses. Proposed Finding 29, which is i

unsupported by any transcript citation, states that LILCO did not contest the accuracy or validity of the County's Lloyd's

! calculations. Contrary to this assertion, LILCO extensively i

cross-examined the County concerning the Lloyd's calculations.

(Tr. 23,976-24,056). Furthermore, there is no requirement in Lloyd's for an overload ' calculation. (Lilco Reply Finding 10).

i 7. Little if any weight.should be given to the Coun-ty's Lloyd's calculations. Professor Christensen described the

. , - - , ~ ~ . - . , r,- - -

..--,e , - -

,.-..m .. -

l i

Lloyd's horsepower calculations as a very, very simple " chug and plug" formula. (Tr. 24,011 Christensen). This type of formula is commonly used in the initial design of a crankshaft.

(Id.) It is not appropriate to conclude, however, that an ex-isting crankshaft is inadequate because it does not comply with ,

this simple horsepower formula. -

As Dr. Pischinger noted, de-i . i l sign foruulas (such as Lloyd's), which do not require any thought during the design process, have been criticized for not being in complete accord with physical laws. (Tr. 22,789

Pischinger).

8. Notwithstanding that the County's witnesses de-scribed the.Lloyd's formula as very simple, they were unable to i

state whether their calculated horsepower rating at full load I

was approximately 54 less than the actital horsepower of the l

EDGs. (Tr. 23,977-79 Christensen). They were also unable to state what effect an increase of 25% in the Zed factor in the i

calculation would have on the allowable horsepower. (Tr.

24,040-49 Christensen, Eley).

9. In addition, when asked if the allowable horsepower

,. et 3300 kW could be obtained by ratioing the calculation that was made at 3500 kW, Professor Christensen stated that the

chug and plug" horsepower formula was " complicated." He'was j unable to say whether the allowable horsepower at 3300 kW could !

l i l l

l I

f

- . - . - _ _ - . -i

be obtained by ratioing without " spending a lot of time

. thinking about it." (Tr. 24,054-56 Christensen, Eley).

10. Joint Proposed Finding 31 discusses the County's horsepower calculation at 3900 kW. There is no requirement in Lloyd's for an overload calculation. (Tr. 24,005-06 Chris-tensen, Eley). According to the County's calculations, the re-i* placement crankshafts would have to be capable of operation at 4290 kW in order to meet Lloyd's requirements. (Tr. 24,016-20 Christensen, Eley).

1 '

11. Joint Proposed Finding 31 also incorrectly states i

that the acutal measured peak firing pressure in the EDGs is j 1720 psi at full load. This representation is misleading be-cause 1720 psi is the single highest peak firing pressure ever measured in any cylinder on the Shoreham EDGs at full load.

(Tr. 22,081 Pischinger). It was measured from one cylinder at one point in time and is by no meanc a representative number.

The peak value for one individual cycle is not the important value in performing a fatigue analysis. The significant value is the peak firing pressure that is representative of the mil-lions of fatigue cycles to be experienced by the engine, or in s

other words, the average peak for a large number of cycles.

(Tr. 22,075 22,081-82 McCarthy). The significant value is ,

neither the highest nor the lowest pressure measured at any one i

1 l

! [

l time and, therefore, is certainly not the 1720 psi represented ;

-by the Intervenors.

12. In addition to the fact that the 1720 psi is the 4

highest peak firing pressure ever recorded on the Shoreham

EDGs, this measurement was recorded by a Kiene guage, which, in and of itself, is designed to record "the maximum of the maxi- ,

num values of different cycles." (Tr. 22,562-63 Pischinger).

l The Kiene guage overemphasizes the higher peak firing pressures

[ and is not as accurate as other measurement devices that re-

! flect a more representative peak firing pre,ssure. (Tr.

4 4

22,113-15, 22,562-63 Pischinger). ,

1

13. A more accurate measurement of the average or rep-I resentative peak firing presure can be obtained from a quartz Piezo electric transducer. (Tr. 22,113 Pischinger). FaAA mea-suced the peak firing pressures from cylinder number seven in the Shoreham EDG 103 with a Piezo electric transducer. The In-tervenors totally fail to give these measurements the proper

! credit. The measurements using the quartz Piezo electric i

tranducer gave the whole pressure trace (Tr. 22,113-15

, Pischinger), reporting pressures from 800 individual cycles of

! the engine as opposed to just the maximum pressure of 1720 psi l l from one cycle measured by the Kiene guage. (Tr. 22,537

( Swanger).- The highest peak firing pressure obtained from the i

(

t

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

l l

l measurements (adjusted for the turbocharger boost) for 250 sec- >

l onds using the more accurate quartz Piezo electric transducer at 1004 load was 1668 pai, and the average value was 1604 psi.

(Tr. 22,461-62; 22,534-35 McCarthy). The Piezo tranducer mea-r surements are more accurate than the Kiene guage measurements ,

and nothing in the record contradicts this. The County at-i tempted to discredit the accuracy of the Piezo electric tranducer measurements, but the entire record, and not just the i portions extracted by the Intervenors, does not support their I attempt 4

! 14. In sum, little weight should be given to the Coun-I i ty's Lloyd's calculations. These formulas are intended to be I

j used as initial design guidelines. The fact that the crank-shafts may not comply with these formulas does not mean the crankshafts are inadequate. The horsepower calculations were l

developed at a time when engineers did not know how to calcu-

! late torsional stresses. (Tr. 24,203-04 Christensen).

4

) Cross-examination established that the County's witnesses did not fully understand even the simple Lloyd's horsepower calcu-lations, one of the few they actually performed. l l

!' 15. An additional reason cited by the County for their i

belief that compliance with Lloyd's is important is the fact that Lloyd's has a staff of experienced surveyors who inspect 4

.g.

4 l

l l

cnd evaluate engines after they have been in operation. The purpose of these inspections is to ensure continued safe op-oration. As an example of the type of inspections Lloyd's sur-veyors are capable of performing, the County's witnesses cited the testing of bottom-end bolts. The testing procedure is to remove the bolts, hang them on a piece of string, give them a

! blow with a hammer and se'e if they ring. If the ring is doubt-

! ful, the bolt may be tested further. There is, however, no ;

cpecific rule covering bolt ringing in Lloyd's. (Tr. 24,215-17

> Christensen, Eley). Obviously, this type of testing procedure l could provide no assurance of the adequacy of any component in the Shoreham EDGs and provides no support for the County's con-tention.

(b) IACS

16. Joint Proposed Findings 34-38 discuss the crank-chafts' alleged non-compliance with the CIMAC rules. These i findings should not be adopted because the record demonstrates i ;

that the County's witnesses had little or no understanding of j the CIMAC rules and their testimony should be entitled to lit-s tle weight. :

i l l

17. Joint Proposed Finding 35. correctly states at l

footnote 10 that the CINAC rules are based on the conservative assumption that the maximum bending and torsional stresses l

l i

\

3 HV

+

occur sitaultaneously and at the same location. However, when

asked if this is what the CIMAC rules said, Professor Chris-tensen said no. When he was shown a copy of the CIMAC rules 4

(County Exhibit 38), he said the translation was poor. (Tr.

}

24,106-07 Christensen). Furthermore, Professor Christensen did f

not know "without thinking about it" whether the actual maximum bending and torsional stresses in the Shoreham crankshafts oc-g f cured simultaneously and at the same location. (Tr. 24,109 k Christensen). The uncontradicted evidence showed that the max-

=

imum bending and torsional stresses do not occur simultaneously

[

[ or at the same location. (LILCO Proposed Finding 63) .

18. Joint Proposed Finding 36 states that the County 1

reviewed TDI's CIMAC calculations and determined that the re-L f placement crankshafts do not comply with the CIMAC rules. This i statement is misleading. Cross-examination of the County's

?-

witnesses established that they did little more than check 1

TDI's math. (Tr. 24,138-39 Eley). One of the important inputs

[_

c to a CIMAC calculation is torsional stress. The County's wit-c nesses made no attempt to check the accuracy of TDI's torsional calculation. (Tr. 24,137-39 Christensen, Eley). Indeed, Mr.

7 Eley freely admitted that he was incapable of calculating tor-P

E sional vibratory stresses. (Tr. 23,968 Eley). Professor Christensen claimed to be capable of performing forced P

l i

l

- torsional vibratory calculations, but admitted he had not per- l l formed such calculations for this case. (Tr. 23,965-66 Chris-tensen). When questioned further, Professor Christensen re-vealed that he did not know how to sum orders or how to calculate the phase relationship between two orders. Indeed,

all Professor Christensen knew how to do was a Holzer tabu-lation. (Tr. 23,965-70 Christensen). The opinion of these witnesses concerning the accuracy of TDI's CIMAC calculation should be given no weight.

19. Joint Proposed Finding 36 also states that the firing pressure of 1720 psi should have been used in the CIMAC calculation, rather than the 1650 psi used by TDI. There is no basis for this statement. For purposes of calculating tor-j sional stress, a representative rather than a-maximum peak fir- {

ing pressure is desired. (LILCO Proposed Finding 57). 1720 psi is not a representative peak firing pressure. (LILCO Reply l Findings 11-13). The TDI test logs, which are part of County Exhibit 46, show average peak firing pressures at full load no 1

! higher than 1647 psi. The figure of 1650 psi used by TDI was a conservative average.

20. . Joint Proposed Findings 37 and 38 discuss an ABS calculation that shows compliance with the CIMAC rules. The~

! Intervenors assert this calculation is insufficient evidence to I

l I

support LILCO Proposed Finding 11. The Intervenors' assertion in this. regard is entitled to no weight. First, the calcula-tion in question was part of an exhibit offered into evidence by the County. (County Exhibit 43 at 29). Second, the fact that the County's witnesses had not previously reviewed the

! calculation does not support Joint Proposed Finding 38. In-stead, this fact supports a finding that the County's witnesses i ,

were ill-prepared, unfamiliar with even their own exhibits and i i generally uninformed about the areas in which they were offered as expert witnesses.

(c) ABS Rules on Torsional Vibration J

21. Joint Proposed " Findings 39-50 urge the Board to 1 find that the crankshafts do not comply with the ABS rules on torsional vibration. The Intervenors essentially ask the Board to overrule ABS's approval of the crankshaft and to determine f that ABS is incapable of performing its own-job. The fact re-mains, however, that ABS approved the ~ torsional critical speed t

arrangement of the crankshafts (LILCO Exhibit C-13) and, de-spite an invitation by counsel for Suffolk County to reconsider (County Exhibit 46), ABS.has not withdrawn its approval of the crankshafts.- ABS' approval on this issue is dispositive.

22. Joint Proposed Finding 40 and 41 state that the Staff and the County both conclude that the torsional stresses l

.n. _ _: . . - . , . . _ _ _ , _ _ . _ _ - - . - . -. _ -- - ., - , -

l oxceed ABS limits. There is no basis for a finding that the i

! ~ torsional stresses in the crankshafts exceed ABS limits.

First, the County's witnesses have no ability to calculate tor-sional stresses (LILCO Reply Finding 18) and were merely inap-propriately comparing FaAA's calculations to the ABS limits.

3 .

Second, both the Staff and County compared a calculation that summed 24 orders. There is no evidence whatsoever that ABS re-quires or even contemplates a 24 order calculation. Indeed, 4

the only evidence in the record is that ABS sums two orders.

(Joint Proposed Finding 41, n. 12). There is no basis for a finding that ABS requires more information concerning torsional stress that was submitted by TDI. The statements by Professor Sarsten in footnote 12 of the Joint Proposed Findings are sheer speculation on his part.

23. Witnesses for both the Staff and County argued that the crankshafts should be evaluated by the-rules of vari-

< ous clasification societies. The classification soc.ieties are in the best position to interpret their own rules. ABS inter--

- preted its rules to approve the torsional stress-levels in the

~

, replacement crankshafts. Furthermore, the last sentence in footnote 12 is just plain wrong. The ABS calculations showed that the stresses, as determined by ABS, clearly met the allow -

able'. limits of'the 1984 rules. (County Exhibit-43 at 26-28).

L

_13_

4 l

l l

The stresses did not meet the limits of the 1983 rules. The limits were increased by approximately 400 psi in 1984. The Intervenors' statement-in footnote 12 is blatantly misleading.

24. Joint Proposed Findings 42-50 argue that the ABS approval is entitled to no weight because TDI submitted inaccu-rate information to ABS concerning shotpeening. The Interve-nors again distort the record in an attempt to provide a com-pletely one-sided view of the ABS approval. The Intervenors suggest that ABS approved the crankshafts only because TDI attributed a 20% increase in the fatigue life to shotpeening.

d (See, e.g. Joint Proposed Findings 44, 49). There is however, nothing in the record to support this assertion. The informa-tion TDI submitted to ABS is contained in County Exhibit 45.

The information includes a torsional analysis by TDI, data con-cerning the mass elastic system for a number of DSR-48 engines,

- results of the torsiograph tests on EDG 103 in January, 1984, information on the physical properties of the crankshafts, re-sults of the strain guage tests on EDG 103 and a log of op-erating hours'for other DSR-48 engines. The Intervenors would have the Board believe, incorrectly, that the sole factor that lead to ABS approval was shotpeening. The record supports no such finding.

re , - - , . - - .-- ,

25. The-Intervenors also suggest that ABS would not have attributed a 20% increase in fatigue life to shotpeening were it'not for TDI's representation. The Intervenors again fail to reveal. facts' that do not support their version of

events. For example, County Exhibit 43 contains, among others, pages 93, 98 and 199 of the ABS deposition. Page 93 of the dep-

.osition contains the statement that ABS accepted TDI's repre-sentations concerning shotpeening. The County intentionally a

excluded pages 94-97 of the ABS deposition from County Exhibit

43. This is not surprising.- On page 94 the ABS representa-tives-stated that they had no reason to disagree with the 20%

value because that value was established by the literature and ABS had seen that value quoted most often. Indeed, they said-i come people actually claimed credit for a much higher increase in the fatigue limit as a result of shot peening. In addition, at page 97 the ABS representatives said they had reviewed lit-erature on shotpeening and for that reason did not question the 20% figure reported by TDI. Therefore,-contrary to the Inter-venors' mischaracterization, ABS gave credit for the l

shotpeening because ABS believed 20% was a reasonable figure, l

'. not because ABS was mislead by TDI.2/

b l 2j/ LILCO recognizes that pages 94-97 of the ABS ' deposition cre not part of the record and could not be - the basis of a 1

(Continued) 4 15-t

, v., ,.g,- ,- .,, ,w., - ~.- - , , - , , , , , - , , .,-~e.,..-e ,e,-.. - ~ e , a

26. It would be entirely inappropriate for the Board to overrule the ABS approval of the replacement crankshafts.

l The Intervenors argue they are in a better position to inter-i pret and apply the ABS rules than ABS. The record is clear that according to ABS the crankshafts comply with the ABS rules. No further inquiry is necessary or appropriate.

(d) ABS Web Dimensions

27. Joint Proposed Findings 51-55 argue that the re-placement crankshafts do not comply with the ABS rules on web dimensions. The record does not support these proposed find-ing.

28. Joint Proposed Finding 52 discusses Professor Christensen's web calculations. Even his calculation shows compliance with ABS rules at full load. Professor Chris-tensen's web calculations show that the maximum allowable fir-ing pressure at full load is 1746 psi. Even assuming that the correct peak firing pressure at full load is 1720 psi, the j crankshafts pass.

s (Continued from Previous Page)

. - finding by the Board. LILCO merely brings this information to the Board's attention to assist in fairly evaluating the Joint

Proposed Findings relating to ABS.

29. Joint Proposed Finding 55 states that there is in-sufficient evidence to support a finding that ABS has approved the web dimensions of the replacement crankshafts. The fact I

remains, however, f that the testimony of Mr. Montgomery was uncontradicted concerning ABS approval of the web dimensions.

(LILCO Proposed Finding 45) . Despite the fact that Professor Christensen thinks the webs do not meet ABS rules, ABS has ap-i proved the webs. .

(e) The Crankshafts are Adequate Under the Kritzer-Stahl Criteria

30. Joint Proposed Findings 56-63 attempt to discredit f

the analysis performed by Dr. Pischinger under the Kritzer-Stahl criteria. These proposed findings are based on a tortured reading of the record and distort Dr. Pischinger's testimony.

31. Joint Proposed Finding 58 states that the replace-j ment crankshafts are "just on the boundary" of the Kritzer-Stahl' criteria at full load. The support -for this statement'is a reference to Dr. Pischinger's deposition on June 21, 1984. This proposed finding completely ignores Dr.

! Pischinger's testimony at the hearings. Dr. Pischinger per-formed additional work after his deposition which allowed him to state unequivocally that the replacement crankshafts have a !

l 1

1 l

factor of safety at full load of 1.24. (Tr. 23,026). The In-tervenors simply ignore the further analyses Dr. Pischinger i

performed. There is no' basis for a finding that the crank-shafts are "just on'the boundary" at full load.

32. Footnote 17 to Joint Proposed Finding 58 and Joint e

Proposed Finding 63 draw a totally unsupported conclusion con-cerning the' significance 5f the ultimate tensile strength (U.T.S.) values used by Dr. Pischinger in his calculations.

Dr. Pischinger used a U.T.S. value of 700 Newtons per square millimeter in his calculations. The actual U.T.S. for two of the replacement crankshafts is 695 Newtons per square millime-ter. Dr. Pischinger contacted Krupp, the manufacturer of the crankshafts, and asked them what U.T.S. value he should use.

Krupp supplied 700 Newtons as the appropriate figure. (Tr.

22,993 Pischinger). In addition, the difference between 700 Newtons and 695 Newtons is less than11% and is not significant.

(Tr. 22,993 Pischinger). The Intervenors argue this is unsettling because minor refinements to Dr. Pischinger's calcu-lations can substantially change the results. Dr. Pischinger 1 1

testified without contradiction, however, that the difference in U.T.S. ' values was of no significance. l

33. Joint Proposed Findings 59 and 60 are merely con-clusions.that are unsupported by the record.

i _ - __ _ _ -

I I

! 34. Coint Proporti Finding 62 distorts the record in en attempt to discredit Dr. Pischinger's calculations. The In-i tervenors argue that if failure is defined as the time a crack l

'i nitiates, rather than the time a crankshaft severs, Dr.

Pischinger's calculations accurately predict the lifetime of the original crankshafts. The results of any calculation may be altered if the assumptions on which the calculation is based

. are changed. It is clear, however, that all of Dr.

{

Pischinger's calculations, as well as the S-N curve he used, l were based on hours of operation until the crankshafts actually i

eevered. (Tr. 22,778, 23,007-08 Pischinger). Dr. Pischinger's calculation of the lifetime of the original crankshaft predicts 150 hours0.00174 days 0.0417 hours 2.480159e-4 weeks 5.7075e-5 months of life at full load until severance. He did not predict 150 hours0.00174 days 0.0417 hours 2.480159e-4 weeks 5.7075e-5 months of life until crack initiation. The original crankshaf t on EDG 102 actually operated for 273 hours0.00316 days 0.0758 hours 4.513889e-4 weeks 1.038765e-4 months at full load before it severed, almost twice the life predicted by Dr.

Pischinger. (LILCO Proposed Finding 84). The discussion of when failure occurs-was totally theoretical. (Tr. 23,008-09

- Pischinger). Only by ignoring what Dr. Pischinger actually did and making an entirely unfounded comparison can the Intervenors cegue that his calculations do not~ provide assurance of the ad-cquacy of'the replacement crankshafts.

i 4

19_

i .

35. Footnote 20 to Joint Proposed Finding 63 com-l - pletely distorts Dr. Pischinger's testimony concerning the ade-quacy of the webs of the replacement crankshafts. While the statements attributed to Dr. Pischinger in footnote 20 are cor-rect, the Intervenors fail to mention one critical fact. Al-though Dr. Pischinger said he might make the webs 1/2 inch thicker if he were designing the crankshafts, this fact had nothing to do with his assessment of the crankshafts under the Kritzer-Stahl criteria or his conclusion that the crankshafts have an adequate safety factor. (Tr. 23,024-25 Pischinger).

Web dimensions are merely an input to the Kritzer-Stahl calcu-lations. (Tr. 22,783 Pischinger).

B. DEMA'

1. DEMA is an Appropriate Standard

36. Joint Proposed Findings 64-76 present the County's view that DEMA is an inadequate standard by which to judge the l l adequacy of the replacement crankshafts. For example, Joint.

Proposed Findings 64, 67, 68, 70 and 76 all conclude, without any. transcript citation, that LILCO has failed to establish l , l 1

i that DEMA constitutes an acceptable criteria to judge. the ade-

quacy of the replacement crankshafts. These proposed findings ignore one critical fact: the adequacy of DEMA as a standard 1

I

,. -% ,,r..# -,r .--....m,- ,. - - , - , _ . - - , + - ..v,c--,ev.m., -- - * - . - - . ,..---+m --- -_

is~not within the scope of the admitted contention. (See Joint-

. Proposed Finding 3). ' The Intervenors were we.'.1 aware that the purchase specifications required the diesels l':a comply with DEMA. (LILCO Proposed Finding 8) . At no poine, however, did

?

the Intervenors propose a contention that cha.-l'enged the ade-l ;

quacy of DEMA as a standard. Therefore, LILCO did not have the l burden of proving that DEMA was an appropriate standard. LILCO merely had to prove'the crankshafts met the DEMA recommenda-tions. The Joint Proposed Findings also ignore the fact that i

DEMA is the only standard referenced in NRC Reg. Guide 1.9, Rev. 2, which addresses the design of standby diesel generators I at nuclear power plants. (See LILCO Proposed Finding 9; Staff Proposed Findings 15-21).

! 37.- Joint Proposed Finding 66 cites only a portion of Dr. Chen's testimony and does not accurately reflect the i record.3/ Dr. Chen specifically, stated that while he thought i' the DEMA rules were older rules and conservative, he did not i think they were out-of-date or obsolete. (Tr. 23,015 Chen).

In addition, Dr. Chen considers DEMA a valid and reasonable method for evaluating torsional stresses on a crankshaft. (Tr.

[ 23,015-16 Chen).

L 3/ The selective use of the record by the Intervenors is par- l l

ticularly inappropriate lin'this instance. While Dr. Chen's early testimony after arriving late on September 17, 1984, was l somewhat' unclear, he later corrected himself. 'The Intervenors have chosen to ignore Dr. Chen's corrections.

l w w ,--s-,, - - - - - ~ ~ + , e- , ,-e -ene w-----A , , a---,, .- --,-e ,--g .- ,-s-- we

i

. l

38. Joint Proposed Finding 67 makes the unsupported leap from the fact that DEMA is not a design code, which LILCO does not dispute (LILCO Proposed Finding 10), to the conclusion

'that DEMA.does not provide standards to measure the adequacy of a crankshaft. Suffolk County, and to some extent the Staff,

- have lost sight of the fact that the issue in this proceeding is not which organization or society provides the best rules for designing crankshafts. Rather, the issue is whether the i

existing replacement crankshafts are adequate for their intend-cd service. Therefore, the fact that DEMA cannot be used to design a crankshaft is irrelevant. DEMA can plainly be used to evaluate a crankshaft. (Tr. 23,015 Chen).

39. Joint Proposed Findings 68-75 deal with the ques-tion of how the DEMA recommendations should be interpreted.

These findings urge, as do Staff Findings 25-28, that the DEMA ellowable stresses should be calculated according to the most sophisticated computational techniques, which permit the summa-tion of 24 orders of vibration. These proposed findings are based exclusively on the testimony of Professor Sarsten. The County's witnesses were unable to perform torsional vibration s

calculations, had no prior experience with DEMA and were able to shed no light whatsoever on this question. (LILCO Reply Finding 18; Tr. 23,975-76 Christensen, Eley).

l

40. Both ch'e County and the Staff urge the Board to cdopt ?rofessor Sarsten's interpretation of DEMA and reject Dr.

Chen's interpretation. Professor Sarsten's interpretation of DEMA is entitled to little, if any, weight. Professor Sarsten freely admitted that he had no way of knowing what methodology DEMA intended to be used for the calculation of vibratory otresses. (Tr. 23,239 Sarsten). He had no knowledge of how i American diesel manufacturers sum orders for purposes of DEMA ccmpliance or how those firms interpret DEMA. (Tr. 23,246-47, 23,250 Sarsten). Professor Sarsten had no experience with DEMA prior to his invovivement in this case (Tr. 23,255 Sarsten), he orde no inquiry of DEMA members concerning the number of orders typically summed for a DEMA calculation (Tr. 23,254 Sarsten),

and he based his interpretation of DEMA solely on his own reading of DEMA. (Tr. 23,247 Sarsten).

41. Dr. Chen, on the other hand, has worked in the 1

Anerican diesel industry since 1952, worked for Fairbanks-Morse l l

(which is a member of DEMA) from 1969-1973, and was chairman of the DEMA Technical Committee from 1971-1973. (Chen, ff Tr.

s 22,610 at 4, 30;'Tr. 22,695). Dr. Chen's testimony was uncontradicted that it is standard practice'in the American diesel engine industry to sum only the four or six largest or-ders of vibration for purposes of comparing stresses against ,

4 the DEMA allowables. (LILCO Proposed Finding 21) .

/

42. Joint Proposed Finding 68 implies that DEMA is in-cdequate because it does not specify any method to be used for ecleulating torsional vibratory stresses. However, it should b3 noted that of the three classification societies supported by the Intervenors (Lloyd's, ABS and IACS) only the IACS calcu-lotion (CIMAC) specifies the number of orders to be used in calculating torsional stress. (Tr. 23,286 Sarsten). Neither ABS nor Lloyd's specifies the number of orders to be used in calculating torsional stress, but rely on the calculations sub-eitted by'the manufacturer. (Tr. 23,286 Sarsten).

43. Joint Proposed Finding 69 is riddled with internal inconsistencies. The Intervenors first criticize LILCO for not caeking an interpretation from DEMA. They then criticize DEMA for having no procedure for obtaining an interpretation of the recommendations. If no procedure exists for obtaining an in-terpretation of DEMA, LILCO can hardly be criticized for l failing to seek such an interpretation. Finally, they recog-nize that Dr. Chen did contact several members of DEMA in for- l tulating his opinions.

44. Joint Proposed Finding 71 and 72 state that calcu-lational methods such as modal superposition have been conven-tionally used since 1972. This is only partially accurate.

Both Professor Sarsten and Dr. Pischinger testified that such 6

methods were conventional in Europe. (Tr. 23,283-84 Sarstent i Tr. 22,989-90 Pischinger). There was no evidence that such cathods were or are conventionally used in the United States. (

45. Joint Proposed Finding 72 concludes, totally with-cut support from the record, that DEMA would not continue to publish its recommendations if it did not believe current ana-lytical techniques should be used. There is no basis for this conclusion. Indeed, the only fair conclusion that may be drawn from the record is that analytical techniques similar to those cvailable at the time the DEMA recommandations were adopted chould be used for comparison with the DEMA allowables. Wit-nesses for LILCO, the Staff and the County all testified that prior to the development of digital computers in the mid-1960s it was very difficult to sum more than two or three orders.

(Tr. 22,742 Pischinger: Tr. 23,282 Sarsten: Tr. 24,200-04 Christensen). It is uncontradicted that at the time the DEMA torsional vibration limits were established in 1958, the summa-tion of 24 ordern was not possible. It is further i uncontradicted that DEMA only calls for the summation of

' major" orders and that the DEMA stress levels have not changed since 1958. (LILCO Proposed Finding 19). Professor Sarsten's testimony on this matter is entitled to little weight because has has no experience with DEMA. (LILCO Reply Finding 40) .

4

46. Joint Proposed Findings 73-74, and Staff Findings 27-28, state that Professor Sarsten's interpretation of DEMA chould be accepted because Dr. Pischinger and FaAA also sum 24 Ceders. These proposed findings misinterpret the significance I

of FEV's and FaAA's calculations. First, Joint Proposed Find-ing 74 distorts Dr. Pischinger's testimony. Dr. Pischinger's otatement concerning the number of orders he would sum was made specifically in the context of the calculations he would per-form for the Kritzer-Stahl criteria. It was not a general otatement concerning all calculations of torsional stress.

(Tr. 22,798 Pischinger). Second, both FEV and FaAA calculated torsional stresses using 24 orders to determine an input to a fatigue calculation. Neither FEV nor FaAA were calculating nominal torsional stresses for purposes of comparison to the DEMA limits. Indeed, both Dr. Pischinger and Dr. Johnston stated that they did not believe their 24 order torsional stress calculations should be compared to DEMA. (Tr. 22,801, 22,809 Pischinger; 22,851-53 Johnston). Only Professor Sarsten summed 24 orders solely for the purpose of comparing his re-suits to the DEMA allowables, and even he recognized that his calculation of nominal stresses should more appropriately be ured as an input to a factor of safety calculation. (Tr.

23,384 Sarsten).

i 1

47. Joint Proposed Finding 75 incorrectly asserts that Dr. Chen stated experts could reasonably disagree over which t

orders were major. Nowhere on the page of the transcript cited

) by the Intervenors is there a statement resembling the one attributed to Dr. Chen. In fact, the record reflected that l there was remarkable agreement among the experts about what the t

i torm major orders meant. (Tr. 23,085-87 Chen, Johnston, 1

3 Pischinger; Tr. 22,747-50 Chen, Johnston).

2. The Crankshaf ts comply with DEMA l 48. Joint Proposed Findings 77-85 discuss the tor- l oional calculations performed by various witnesses. The entire l

f discussion in this section is based on the incorrect assumption I

that the appropriate torsional stress calculation for purposes of comparison with DEMA is one that suas 24 orders.

49. Joint Proposed Finding 77 is a conclusion that is unsupported by any transcript citation.

50. Joint Proposed Finding 79 incorrectly states that the analyses by FaAA and Dr. Pischinger show noncompliance with DEMA. Both Dr. Johnston and Dr. Pischinger specifically stated that it was not appropriate to compare the results of their cnalyses with DEMA. (Tr. 22,801, 22,809 Pischinger; 22,851-53 Jchnston). Nominal tersional stresses esist only hypothetical-ly. The stresses may be computed in different ways for

~27-

e .

different purposes. It is appropriate to use a 24 order summa-tion as an input to a fatigue calculation to calculate a safety cargin or a true stress, rather than a nominal stress. It is not appropriate to use a 24 order summation to make a compari-son with DEMA. The appropriate calculation for a DEMA compari-

)

con is a four or six order summation or a reduction of j torsiograph test data. (Tr. 22,851 Johnston).

51. Joint Proposed Findings 80 and 81 state that Pro-fessor Sarsten's calculations are "the most accurate of the cethods used by the expert witnesses in this case." Although the record supports a finding that there are some slight dif-forences in the computational methods and results obtained by the dif ferent experts, the record does not support a finding that Professor Sarsten's calculations are "more accurate."

(Tr. 23,050-54 Johnston, Pischinger). The Intervenors empha-size the fact that Professor Sarsten's prediction of free end cmplitude was in closer. agreement with the free end amplitude ceasured by Stone & Webster than the values calculated by FaAA and Dr. Pischinger. There is no significance to this fact.

The tent measurements are only accurate to within t'St. (LILCO Exhibit C-16 at 7-3; County Exhibit 49). Professor Sarsten's, FaAA's and Dr. Pischinger's predicted free end values are all within 5% of each other. This fact shows a rather remarkable u

l l

\

cgreement among the results and in no manner establishes that one calculation is more accurate than another.

52. The primary basis for the proposed finding that l Professor Sarsten's calculations are "more accurate" is that Professor Sarsten used a larger dynamic magnifier in his calcu- 3 j lations than did FaAA. However, under cross-examination by the 1

Board, Professor Sarsten was unable to explain why the dynamic 4

cagnifier he used (40) more accurately represented the real world condition of the engines than the dynamic magnifier used

! by FaAA (20). (Tr. 23,437-40 Sarsten). Professor Sarsten also i "

l comitted that at the engine speeds in question, the stresses l

li were not very much influerced by the damping. (Tr. 23,440 l Sarsten). The question of which method is "more accurate" is f largely theoretical. (Tr. 23,441 Sarsten). Professor ,

I '

j Sarsten's assertions that in theory his method is more accurate i

! do not support Joint Proposed Findings 80 and 81. Inter-ostingly, the Staff makes no such claims for Professor

]

! Sarsten's calculations. (See Staff Proposed Findings 31-33).

l 53. Footnote 35 in Joint Proposed Finding 81 incor-l

[ . rectly states that FaAA used a one node vibratory form as the ,

! basis for calculating stresses. The record is clear that in !

its modal superposition analysis, which is the calculation being discussed by the Intervenors in Joint Proposed Finding 29-e

- r--- 't e yr--w .-e< ---=-:=w -e * -e----m--*--- r- = ---rur---ww-*--ee -$mvm-r+,eeem-=-- -------+--e-w+,- e--,e-a w---- - * - -----=-a-+-e,r+-

81, FaAA calculated the multi-modal response of the crankshaf t.

(Tr. 23,050-53 Johnston). FaAA did assume a one node vibratory form when it converted the torsiograph data into stresses.

(LILCO Proposed Findings 38-41) . FaAA's torsiograph conversion is not, however, discussed in Joint Proposed Finding 81.

54. Joint Proposed Findings 83-85 state that the otresses calculated by Professor Sarsten, Dr. Pischinger and FaAA are too low, because FaAA's Tn values are too low, because i the pressure measurements taken from cylinder No. 7 on EDG 103 ore too low. The record supports none of these proposed find-l ings. First, nowhere in the record is there testimony that the 0 tresses should be higher, which is evident from the fact that

there is no transcript reference in support of the statement.

S;cond, Professor Sarsten's testimony about alleged

inaccuracies in FaAA's Tn values was, like his discussion of damping, largely theoretical. Professor Sarsten could not say with certainty that there was any error in the Tn values, and to the extent there was any error, it was not large and was not significant enough for him to address in his direct testimony.

(Tr. 23,412 Sarsten). Further, it was not even possible to calculate what error might exist. (Tr. 23,415 Sarsten) . Final-ly, to the extent there was any error, it was not unacceptable end the Tn values were more accurate than tabulated Tn values.

a (Tr. 23,418-19 Sarsten). In sum, there is no evidence to sup-

i

! port.a finding that FaAA's Tn values are too low. Dr. ;

i Pischinger also calculated Tn values from the pressure measure- i I

Eents taken on EDG 103. There was good agreement between Dr. l Pischinger's Tn values and'FaAA's Tn values. (Tr. 22,811-16

Pischinger).

55. Joint Proposed Finding 85 and footnote 36 assert

]

i that the pressure measurements are too low. The accuracy of pressure measurements is a these repeated frequently by the In-tervenors. (See egg,. , Joint Proposed Findings 31 n. 8, 36, 49 1

n. 15 and 52). However, as before, the Intervenors have no ev-idence to support their assertions. The Intervenors represent i

j that the transducer pressure measurements are too low because f the indicated mean effective pressure (IMEP) calculated from l the pressures does not correspond to the break mean effective l presure (BMEP) of the ShorehamLEDGs. The Intervenors cite tes-l timony by Mr. Henriksen to support this idea. Contrary to this I

implication, however, Dr. Sarsten stated unequivocally just two

! pages before those cited by the Intervanors that the difference -

j need not be attributed to an error in the maximum firing pres-

! suces, but probably was due to the shift in the top dead con- :

i ter, which changes the mechanical efficiency. Moreover, at an- i other point in his testimony, Mr. Henriksen agreed with Dr. i i

i

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

Sarsten that the difference between the IMEP and BMEP was more '

likely attributable to the top dead center location. (Tr.

23,727-728 Henriksen). The Intervenors, however, do not quote (

l this portion of Mr. Henriksen's testimony and attempt to convey the impression that he attributed the difference in IMEP and !

BMEP to the fact that the peak firing measurements should be higher. Mr. Henriksen, in fact, made it very clear by his en-

[ tire testimony that this was not the case. In addition, LILCO j

} Exhibit C-16 (the results of the field test on EDG 103 in 2

January, 1984) contains a detailed discussion of the pressure i

j eeasurements taken by the Piezo electric transducers and the i

procedures used to calibrate the transducers. The Intervenors '

{ presented absolutely no evidence to show that the transducers were improperly calibrated or that the pressure measurements j were too low. Indeed, the County's witnesses did not .even know j! how to calibrate the transducers. (Tr. 24,220-24 Christensen, Eley).

{

f j 56. The Intervenors state as a fact in footnote 36 4

l 1 that cylinder number 7 was not developing full power when the I ceasurements were taken. There is absolutely no evidence to :

support this finding. All of Mr. Eley's calculations, which

cre discussed in footnote 36, were based on the asumption that cylinder number 7 was not developing full power. The I

i i

t l

i i l

Intervenors attempt to parly Mr. Eley's assumption into a fact.

There is, however, no evidence to support Mr. Eley's assump-tion.

C. FaAA's Fatique Analysis

57. Joint Proposed Findings 86-95 argue that the 1.48

, factor of safety calculated by FaAA is insufficient proof that the replacement crankshafts are adequate. There is, however, no evidence' in the record that challenges the methods used or the conclusions reached by FaAA in its fatigue analysis. Nei-ther the County nor the Staff presented any evidence challenging the validity of FaAA's fatigue analysis, nor was FaAA's analysis successfully challenged on cross-examination.

The Staf f agrees that FaAA's analysis is proper and that the '

factor of safety is accurate. (Staff Proposed Finding 61).

FaAA's actual factor of safety for the replacement crankshafts of 1.48 stands uncontradicted.

58. The Intervenors rely almost exclusively on the testimony of Professor Sarsten to support Joint Proposed Find-ings 86-95. This is not surprising due to the inability of the County's witnesses to evaluate the adequacy of a crankshaft based on a fatigue analysis. For example, Mr. Eloy admitted that he was incapable of calculating the actual stresses in the crankshaft fillets. (Tr. 23,847-48 Eley). Professor

Christensen stated that he could calculate the stresses in the fillets. He then described his method, which was a simplifica-tion that actually eliminated the fillets. (Tr. 23,848-49 Christensen). In addition, the County's witnesses did not have cufficient training or experience to evaluate the test data taken from EDG 103. Neither of the County's witnesses knew the degree of accuracy of pressure measurements taken by Piezo olectric transducers. (Tr. 24,220-24 Christensen, Eloy). Pro-fessor Christensen believed that a crack of several microns in the replacement crankshaf ts would propagate. However, he was l unable to describe the stress field such a crack would need in '

order to propagate nor had he performed any calculations to show that such a crack would propagate. (Tr. 24,224-25 Chris-tensen).

59. Notwithstanding Professor Sarsten's obvious abili-ty to evaluate FaAA's fatigue analysis, he did not do so. Pro-fessor Sarsten had no opinion about the adequacy of the re-placement crankshafts because he made no attempt to determine whether the crankshafts were adequate. (Tr. 23,352-53, 23,363-84 Sarsten). Indeed, Professor Sarsten stated that his only concern was whether the nominal torsional stresses ex-coeded 7000 psi. (Tr. 23,352-53 Sarsten). He did not review the results of the tests on EDG 103 and he made no calculations D

l I

of'the stresees in the original 13 inch x 11 inch crankshafts for purposes of comparison with the replacement crankshafts.

(Tr. 23,385, 23,389 Sarsten). Professor Sarsten admitted that i

the calculation of torsional stress was merely one of the in-puts required to evaluate the adequacy of the crankshafts.

L (Tr. 23,'384 Sarsten).

i l 60. Joint Proposed Finding 89 states that FaAA's fac- !

tor of safety is not sufficient proof that the replacement ;

]! The record does not support this

! crankshafts are adequate.

proposed finding. Professor Sarsten's opinion (upon which this l proposed fining'is premised) is not based upon any criticism of

! FaAA's methods or conclusions. (See Staff Proposed Finding (

j 61). Rather, it is simply based on the fact that Professor !

j Sarsten wants to see the crankshafts tested for 10 7 cycles, and ;

t t

i feels more comfortable with classification societies' ;

guidelines. (Tr. 23,528-29 Sarsten).

61. The record is uncontradicted, however, that the j degree of knowledge about the replacement crankshaf ts is more ,

j than sufficient to provide confidence in FaAA's fatigue analy- !

i I

{ sia. Dr. McCarthy testified that FaAA had more knowledge about

the design of the replacement crankshafts than any other.part

, i he had ever confronted in his entire professional esperience, cnd he did not espect to have this kind of information again i

for a long, long time. (Tr. 23,027 McCarthy). -

I l !

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

62. Furthermore, Joint Proposed Finding 89 incorrectly suggests that FaAA's approach to analysing the crankshaf t was solely to compare a calculated endurance limit with the sea-suced stresses. First, FaAA did not calculate the endurance limit of the replacement crankshaf ts in the sense suggested by the Intervenors. The actual endurance limit was established by

~

tests and analyses of the original and replacement crankshafts.

(LILCO Proposed Findings 69-71) . Second, FaAA's assessment of the crankshaft was based upon a three-tiered approach of testing, inspections and analyses, all of which confirmed that the replacement crankshafts were adequnte. (See LILCO Proposed Findings 49-81).

63. Joint Proposed Finding 91 incorrectly attempts to compare the actual endurance limit determined by FaAA with the ondurance limit calculated by Dr. Pischinger. The Intervenors l

imply that FaAA should have used Dr'. Fischinger's endurance limit in its fatigue analysis. It would be totally improper, however, to use an endurance limit that was calculated ac-cording to a specific criteria (Kritzer-Stahl) in a fatigue cnalysis that is based upon actual tests and measurements.

~

Dr.

Pischinger specifically noted that the Kritzer-Stahl criteria cas extremely accurate at predicting stresses because the pre-dicted stresses were very close to the actual measured

stresses. (Tr. 23,006 Pischinger). Dr. Pischinger believed, however, that the endurance limit calculated by the Kritzer-Stahl criteria was extremely conservative and was too low. (Tr. 23,005-07, 23,045-46 Pischinger). There is no evi-dance to indicate that the actual endurance limit determined by FaAA is inaccurate.

64. Joint Proposed Findings 92 and 93 discuss the '

forging process of the crankshafts. There is no evidence that the factor of safety determined by FaAA is effected in any man-ner by_the facts discussed in Joint Proposed Findings 92 and 93.

65. Joint Proposed Finding 94 dAccusses the various safety factors that have been calculated for the replacement crankshafts. It is significant to note that of the ten safety factor calculations made for these crankshafts (seven by ABS, one by TDI, one by Dr. Pischinger and one by FaAA), none yield !

a factor of safety of less than 1.0. (County Exhibit 39 at 6; County Exhibit 43 at 29, 32; LILCO Proposed Findings 72 and 87). Of these ten, only two are below 1.1, and only three are !

below 1.15. (County Exhibit 39 at 6; County Exhibit 43 and I 32). Six of the safety factors exceed 1.22 and four exceed !

6.3. (County Exhibit 43 at 32; LILCO Proposed Findings 72 and i

87). Dr. Pischinger testified that an accepted range of safecy 1 i

1 J

factors was between 1.15 and 1.3. (LILCO Proposed Finding 89) .

Professor Sarsten testified that a factor of safety of 1.1 was adequate under certain circumstances. (Tr. 23,509 Sarsten).

66. FaAA determined that the actual factor of safety for the replacement crankshafts was 1.48. This factor of safe-ty was not based upon a predetermined criteria. It was based upon an extensive engineering analysis to determine a true mar-gin of safety. There is more than adequate assurance that the replacement crankshafts are adequate for their intended ser-vice.

D. Shotpoening

67. Joint Proposed Findings96-106 discuss the issue of shotpoening. It is apparent from the proposed findings that the Intervenors have abandoned their contention that the origi-nal shotpeening damaged the crankshaf ts. (See Joint Proposed Findings 3, 106 n. 41). The Intervenors now merely argue that it is impossible to precisely quantify the increase in the fa-tigue endurance limit attributable to shotpoening. Because the Intervenors have abandoned their shotpoening contention, LILCO

, will not specifically address Joint Proposed Findings96-106.

However, this should not be interpreted to mean that LILCO agrees with Joint Proposed Findings95-106. The record sup-ports a finding that shotpoening increased the fatigue

1

_ i a

cndurance limit by ten percent (10%) to twenty percent (20%).

(LILCO Proposed Finding 109) .

Respectfully Submitted, LONG ISLAND LIGHTING COMPANY BY

[ 2A Counsel ~ f

E. Milton Farley, III John Jay Range Hunton & Williams 2000 Pennsylvania Ave., N.W.

P. O. Box 19230 Washington, D.C. 20036 T. S. Ellis, III Anthony F. Earley, Jr.

Hunton & Williams 707 East Main Street P. O. Box 1535 Richmond, Virginia 23212 Odes L. Stroupe, Jr.

David Dreifus Hunton & Williams 333 Fayetteville Street P. O. Box 109 i Roleigh, North Carolina 27602 i

DATED: December 3, 1984

!' l I

LILCO, December 3, 1984 CERTIFICATE OF SERVICE In the Matter of LONG ISLAND LIGHTING COMPANY (Shoreham Nuclear Power Station, Unit 1)

Docket No. 50-322 (OL)

I hereby certify that copies of LILCO'S REPLY TO ,

SUFFOLK COUNTY AND STATE OF NEW YORK' PROPOSED FINDINGS OF FACT were served this date upon the following by first-class mail, t

postage prepaid, or (as indicated by asterisk) by hand.-

i Judge Lawrence Brenner, Esq.* Atomic Safety and Licensing

Chairman Appeal Board Panel

< Atomic Safety and Licensing U.S. Nuclear Regulatory Board, United States Commission :

Nuclear Regulatory Commission Washington, D.C. 20555 l I

4350 East-West Highway

Fourth Floor (North Tower) Martin Bradley Ashare, Esq. !

l Bethesda, Maryland 20814 Att: Patricia A. . Dempsey, Esq.

County Attorney 1 Dr. Peter A. Mo r ris* Suf folk County Dept. of Law

! Administrative Judge Veterans Memorial Highway Atomic Safety and. Licensing Hauppauge, New York 11787 i Board, United States '

Nuclear Regulatory Commission Richard J. Goodard, Esq.*

4350 East-West Highway Office of the Executive Fourth Floor (North Tower) Legal Director l

! Bethesda, Maryland 20814 U.S. Nuclear Regulatory Commission l Dr. George A. Ferguson* 7735 Old Georgetown Road i Administrative Judge Bethesda, Maryland 20814 i Atomic Safety and Licensing

} ,, Board Panel Alan Roy Dynner, Esq.*

l School of Engineering - Kirkpatrick, Lockhart, l Howard University Hill,' Christopher &

2300 6th Street, N.W. Phillips l Washington, D.C. 20059 1900 M.~ Street, N.W.

8th Floor Washington, D.C. 20036 1

1

s

. .' i

\

l

Secretary of Commission

Mr. March W. Goldsmith U.S. Nuclear Regulatory Energy Research Group Commission 4001 Totten Pond Road Washington, D.C. 20555 Waltham, Massachusetts 02154 i

MHS Technical Associates Fabian G. Palomino, Esq.

1723 Hamilton Avenue Special Counsel to the Suite K ' Governor San Jose, California 95125 Executive Chamber, Rm 229 1 State Capitol l Mr. Jay Dunkleberger Albany, New York 12224 l New York State Energy Office Agency Building jt . Jonathan D. Feinberg, Esq.

l Empire State Building New York State i Albany, New York 12223 Depart. of Public Service l Three Empire State Plaza i

Stephen B. Latham, Esq. Albany, New York 12223 I Twomey, Latham & Shea

33 Westssecond Street Robert E. Smith, Esq. ,

P. O. Box 398 Guggenheimer & Untermyer l , Riverhead, New York 11901 80 Pine Street ;

New York, New York 10005 4 James B. Dougherty, Esq.

3045 Porter Street Howard L. Blau

Washington, D.C. 20008 217 Newbridge Road i Hicksville, New York 11801 Ralph Shapiro, Esq.

Cammer and Shapiro, P.C.

l 9 East 40th Street New York, New York 10016 D = D E L .: L -

HUNTON & WILLIAMS P. O. Bo x - 109 Raleigh, North Carolina 27602 .

j DATED: December 3, 1984 .

, 4 l

l l

l i

_,__4, LILCO, April' 4, 1985 CERTIFICATE OF SERVICE

.: !.'EO

~

In the Matter of LONG ISLAND LIGHTING COMPANY (Shoreham Nuclear Power Station, Unit 1) T3 ffR -8 gg5 Docket No. 50-322 V I:Ci C= Sicpi p GCCXU;ldG A SER$#

I hereby certify that copies of LONG ISLAND LIGHTING 3RANcy COMPANY'S PROPOSED FINDINGS OF FACT CONCERNING EMERGENCY DIESEL '

GENERATOR CONTENTIONS were served this date upon the folldwing by first-class mail, postage prepaid, or by hand, as indicated s by an asterisk: .

Lawrence Brenner, Esq.* Atomic Safety and Licensing Atomic Safety and Licensing Appeal' Board Panel Board Panel U.S. Nuclear Regulatory .

U.S. NRC , Commission 4350 East-West. Highway Washington, D.C. -20555 Fourth Floor (West Tower)

Bethesda,- Maryland 20814 Atomic Safety and Licensing Board Panel Dr. Peter A. . Morris

U.S.. Nuclear Regulatory Atomic Safety and Licensing Commission Board Panel. Washington, : D. C. 20555 U.S. NRC 4350 East-West Highway Robert E.-Smith, Esq.

Fourth Floor (West Tower) Rosenman Colin _Freund Lewis Bethesda,. Maryland 20814 &-Cohen 575 Madison Avenue

. Dr. George A. Ferguson* ~New York, N.Y. 10022 Atomic Safety and Licensing _

Board _

Robert G. . Perlis, . Esq.*

School of Engineering U.S. Nuclear Regulatory Howard University Commission 2300 6th Street, N.W. 7735 Old Georgetown Road . ;

Washington, D.C. 20555 Maryland National Bank Bldg. +

Bethesda, Maryland 20814 Secretary of the Commission

.U.S. Nuclear Regulatory ~

-Alan;R. Dynner,'Esq.*

Commission . Kirkpatrick & Lockhart Washington, D.C. -20555' 1900 M Street, N.W.

-Washington, D.C. 20036

Mr.' Marc.W. Goldsmith Energy Research. Group. Stephen _B.: Latham, Esq.

4001 Totten Pond: Road' Twomey,.Latham & Shea Waltham,'. Massachusetts 02154' '33 West Second Street t P. O.. Box 398 o Riverhead,eNew York 11901 J

o. =t c :

N u_ m]

MHB Technical Associates Ralph Shapiro, Esq.

'1723 Hamilton Avenue Cammer and Shapiro, P.C.

Suite K 9 East 40th Street

-San Jose, California 95125 New York, N.Y. 10016 Mr. Jay Dunkleberger James Dougherty, Esq.

New York State Energy Office 3045 Porter Street Agency Building 2 Washington, D.C. 20008 Empire' State Plaza Albany, New York -12223 Martin Bradley Ashare, Esq.

Attn: Patricia A. Dempsey, Esq.

Jonathan D. Feinberg, Esq. County Attorney State.of New-York Suffolk County Department of Law Departm'ent of Public Service Veterans Memorial Highway Three Empire State Plaza Hauppauge, New York 11787~

Albany, New York 12223 Fabian G. Palomino, Esq. Howard L. Blau Special Counsel to the 217 Newbridge Road Governor Hicksville, New York 11801 Executive Chamber, Room 229

' State Capitol Albany, New York 12224 b

. Tay o Wydley, III T. S. lis, III Odes L Stroupe, Jr.

JohnfJay Range Hunton & Williams 707 East Main Street P.O. Box 1535 Richmond, Virginia 23212 DATED: April 4, 1985 L

ML20112J882

Text

Navigation menu

Search