ML20112A768
| ML20112A768 | |
| Person / Time | |
|---|---|
| Site: | Shoreham File:Long Island Lighting Company icon.png |
| Issue date: | 03/12/1985 |
| From: | Bush S, Henriksen A ADAM J. HENRIKSEN, INC., REVIEW & SYNTHESIS ASSOCIATES |
| To: | |
| References | |
| OL-S-D-014, OL-S-D-14, NUDOCS 8503180448 | |
| Download: ML20112A768 (34) | |
Text
-
5- 3lQglr i
Febri;aggg.s W I4 AH:07 fitBR {g} it,h ANC 'T
- UNITED STATES OF AMERICA NUCLEAR REGULATORY COMMISSION BEFORE THE ATOMIC SAFETY AND LICENSING BOARD e
In the Matter of )
)
LONG ISLAND LIGHTING COMPANY ) Docket No. 50-322-0L
)
(Shoreham Nuclear Power Station, )
Unit 1) )
JOINT TESTIMONY
- AND of SPENCER H. BUSH ADAM J. HENRIKSEN, A"O PROTC000R ART;"JR SAraTC" 3
on LOAD CONTENTIONS CONCERNING TDI EMERGENCY DIESEL GENERATORS at the SHOREHAM NUCLEAR POWER STATION >
IlUCLEAR REGULATORY CCMMIS$10ll onw no. 8322'/ omeu Es.no.0 A- ' J' !Y is the man., of 2 /C 66
! :. NT!T;ID
!!2?f t:CUVLD V
A;;tcant - _ _
A!JLCTED tetornnot Cut's Oft's C<ntractor DAfg OA-VC ,
Mtases
,U #
Cther
/
[- , , ,
Reporter 8503180448 PDR PDR850312ADOCK 05000322 C
hd /t'
CONTENTS INTRODUCTION OF WITNESSES ............................................... 1 2
SCOPE OF TESTIMONY ......................................................
Sul tMARY OF TE ST I MON Y . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ' 3 FATIGUE LIFE OF CRANKSHAFTS IN THE SHOREHAM E0GS.................... 3 CYLINDER BLOCKS .................................................... 6 TESTIMONY ON CONTENTIONS ................................................ 9 o
I - CRANKSHAFT ..................................................... 9 Conclusions That May Be Drawn From Confirmatory Testing ....... 9 Fatigue Li fe of Crankshaf ts in the Shoreham EDGs ... . .. . . .. . . . . 13 I I - C Y L I N D ER B LOCKS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Moni to ri ng of Cam Gall e ry Crack s i n EDGs 101 and 102 . . . . . . . . . . 26 Comments on Testing Performed by Walter C. McCrone Associates, Inc. ......................................... 27 Assumptions and Conclusions Regarding Origin and .,
Charac teristics of Cam Gall ery Cracks . . . . . . . . . . . . . . . . . . . . 27 Conclusions Regarding the Need for Monitoring Cam G a l l e ry C r a c k s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Stud-to-Stud Cracks in the Cylinder Bl ock Top . . . . . . . . . . . . . . . . . 31 l
l l
iii
INTRODUCTION OF WITNESSES Q. Please state your names, your business addresses, and your professiona1 qualifications.
A. (Bush) liy name is Spencer H. Bush. I am sel f-employed, under the firm name of Review and Synthesis Associates, Richland, Washington. A sunnary of my professional qualifications and experience was submitted as Attachment 2 to Volume 1 of the joint testimony filed by the NRC staff in August 1984.
A. (Henriksen) My name is Adam J. Henriksen. I am self-employed, under the finn name of Adam J. Henriksen, Inc., Fox Point, Wisconsin. A sumary of my professional qualifications and experience was submitted as Attachment 3 of the joint testinony referenced above.
(Scr: ten) "y n=: S ."-th r Scr: ten. I r : "r:ft::Or Of Interan Combustion Engines at ian Institute of y, Trondheim, Norway. A summary of m onal quali ic d experience was s ed as Attachment 5 of the icint testimony referenced above.
1 l
l s -
SCOPE OF TESTIMONY Q. What is the scope of your testimony?
A. (All) Our testimony addresses the following parts of Suffolk County's load contention as admitted by the Atomic Safety and Licensing Board:
Contrary to the requirements of 10 C.F.R. Part 50, Appendix A, Gener71 Design Criterion 17 -- Electric Power Systems, the emergency diese' generators at Shoreham ("EDGs") with a maximum " qualified" load of 3300 kW do not provide sufficient capacity and capability to assure that the requirements of clauses (1) and (2) of the first paragraph of GDC 17 will be met, in that (a) LILCO's proposed " qualified load" of 3300 kW is the maximum load at which the EDGs may be operated, but is inadequate to handle the maximum load that may be imposed on the EDGs because:
(1) intermittent and cyclic loads are excluded; (ii) diesel load meter instrument error was not considered; (iii) operators are permitted to maintain diesel load at 3300 kW *100 kW; and (iv) operators may erroneously start additional equipment.
(c) The EDG qualification test run performed by LILC0 was inadequate to assure that EDGs are capable of reliable operation at 3300 kW because:
(i) DG 103 block was not subjected to the entire 740 hours0.00856 days <br />0.206 hours <br />0.00122 weeks <br />2.8157e-4 months <br /> of testing; (ii) the test results on the DG 103 block are not transferable to the DG 101 and 102 blocks; (iii) operators were permitted to control the diesel generators at 3300 kW *100 kW during the test; and (iv) instrument accuracy was not considered.
2 L ._
l
SUMMARY
OF TESTIMONY 0 Please summarize your testimony on these contentions.
A. (All) Our summary testimony is provided under the two subheadings 1
that follow.
FATIGUE LIFE OF CRANKSHAFTS IN THE SHOREHAM EDGs From our review of LILC0's testimony 'and data logs, we believe that EDG 103 was, in fact, operated at a nominal, instrument-indicated load of 3300 kW during that portion of the 1 x 107 -cycle confirmatory test claimed by LILCO to have been conducted at the 3300-kW load level. We understand that the wattmeter may oscillate approximately t100 kW around the value at which the load is set, presumably because this is as close as the load can be controlled without blocking the governor. Based on wattmeter calibration data, the actual load could have differed from the indicated load by abouth kW. In the con-text of the overall test loads included in the 107 cycles and the order in which they occurred, however, we view these deviations from 3300 kW as of no consequence.
In our opinion, EDGs 101,102, and 103 are suitable for nuclear standby service at the " qualified" load of 3300 kW. This opinion is subject to the surveillance and maintenance recommendations documented in the following tech-nical evaluation report, which we assisted in preparing: Review and Evaluation of Transamerica Delaval, Inc., Diesel Engine Reliability and Operability -
Shoreham Nuclear Power Station Unit 1, PNL-5342, dated December 1984. As noted on pages 4.24 through 4.25 of that report, "...the replacement crankshafts for 3
EDG 101, EDG 102, and EDG 103 are acceptable for their intended service, pro-vided that they are not operated during engine tests at loads in excess of the qualified load of 3300 kW." We believe that this restriction is necessary to avoid routine operation of the crankshafts at loads in excess of the load at which one crankshaft has been successfully tested.
Accordingly, we recommend that the permissible load for engine tests, including surveillance tests at the qualified load, be no higher than 3300 kW as read on control room instrumentation. , We understand that the wattmeter may oscillate approximately t100 kilowatts around the value at which the load is set, as discussed above. In our opinion these oscillations during routine
! tests will not be detrimental to engine reliability, provided that the indicated mean load is no higher than 3300 kW.
Loads at which EDG 103 was operated as part of the confirmatory test to 1 x 107 cycles, and the post-test examination that revealed no evidence of damage to the crankshaft or other key engine components, provide a basis for
! drawing conclusions about the capability of the EDGs for emergency operation at loads above the qualified load. EDG 103 sustained over 220 hours0.00255 days <br />0.0611 hours <br />3.637566e-4 weeks <br />8.371e-5 months <br /> (approxi-mately 3 x 106 cycles) at instrument-indicated loads of 3500 kW and above.
With a conservative application of instrument error from calibrations performed by LILC0 preceding and following the time the higher-load testing was per-formed, we estimate that the actual load during this period was at least 3430 kW. If cracks had initiated during this testing, it is likely that they would have propagated during subsequent operation at approximately 3300 kW for the time necessary to bring the total cycles to 1 x 10 7. But no cracks were found in the post-test inspection of the crankshaf t.
4
}
In light of these results, and taking into consideration the small but inevitable differences in the properties of the three crankshafts, it is our opinion that it would be within the demonstrated capability of the engines to operate at Toads to 3430 kW for an hour or so if the engines were needed to carry such loads under emergency conditions. This comment does not apply for routine operation of the engines, including engine testing, for which we recommend a load limit of 3300 kW as discussed earlier in this summary.
The testing performed on EDG 103 does, not provide an adequate basis for drawing conclusions about the effects on the EDGs of loads higher than 3430 kW.
" = ver, Or dditten:1 05:Orvatten :y Sc m:d: 5:: d er other centideretient.
It is gene accepted in the technical literature on fatigue umulative damage in metals that m ry overloads, even th pproaching the ultimate tensile strength of the metal, can b ained without failure. This litera-ture provides a basis f idence that brief ex ions (less than 1 minute) of the Sho engines to loads as high as 3900 kW under emerg conditions
=1 d not c mprert:0 engin: Oper:5fli ty .
If an engine were operated at high overload for a longer period during an emergency, its capability to meet the load profile throughout the emergency would depend on whether or not a crack would initiate in the crankshaft during the overload and propagate to failure before the engine was no longer needed.
The available information does not provide a basis for us to comment with con-fidence on this scenario. However, overloads to 3900 kW for up to 1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br /> under emergency conditions followed by much lower loads in accordance with LILCO's predicted LOOP /LOCA profile are believed to be sustainable. Any crankshaft 5
0 6
that is subjected to more than a momentary overload approaching this level should receive a thorough nondestructive. examination before it is returned to service.
CYLINDER BLOCKS The replacement EDG 103 block was not subjected to the entire qualifica-tion test performed on the EDG 103 engine. Nevertheless, the absence of any reportable indications in the block top a,fter more than 500 hours0.00579 days <br />0.139 hours <br />8.267196e-4 weeks <br />1.9025e-4 months <br /> of operation at or above 3300 kW provides significant evidence that the replacement block is suitable for service at the qualified load. If further operation beyond the most recent inspection does not exceed the FaAA-recommended inspection interval before the end of the first fuel cycle, the top of the replacement block need not be reinspected until the first shutdown for refueling. It is also unneces-sary, in our opinion, to monitor cam gallery cracks in the replacement block.
The known cam gallery cracks in this block have not been repair-welded, and, therefore, residual stress fields that may be associated with repair welds havi-not been introduced into the block material.
The replacement EDG 103 block was more suitable than either the EDG 101 block or the EDG 102 block for the tests that LILC0 conducted to obtain data on compressive and alternating stresses in the camshaft gallery. Use of either of the latter two blocks for the cam gallery tests would have involved the instal-lation of strain gages over repair welds rather than over base metal . However, the test of EDG 103 at qualified load did not contribute to resolution of ques-tions concerning the ligament cracks in the top surfaces of the EDG 101 and 102 6
. ~-
l blocks, the potential for developing stud-to-stud or stud-to-end cracks in those blocks, or the circumferential cracks reported in the original EDG 103 block.
Our coiiciusions expressed previously in written testimony regarding the EDG 101 and 102 blocks remain unchanged. In our opinion, the 101 and 102 blocks are adequate for service subject to certain caveats on surveillance of known cracks. Following any period of operation of EDG 101 or EDG 102 at or above 50% of qualified load, visual (with ,the naked eye) and eddy-current inspections should be performed on those portions of the block top that are accessible between cylinder heads. The purpose of these inspections is to verify the continued absence of detectable cracks between studs of adjacent cylinders. In addition, the behavior of several representative cracks in the can, shaft galleries of the EDG 101 and 102 blocks should be monitored. If no changes indicative of crack growth are observed over the first fuel cycle, the need for continued monitoring of the cam gallery cracks should be reconsidered by the flRC staff.
Our opinion expressed in previous testimony is also unchanged regarding circumferential cracks of the type found in a cylinder liner counterbore of the original EDG 103 block. If such cracks were to develop in any of the three blocks currently in service, it is highly unlikely that they would represent a hazard to EDG reliability. They would be expected to propagate only a short distance into a region of compressive stress and stop. At any time a liner is removed from any of the three engines, however, it would be prudent to perform an appropriate nondestructive examination of the landing of the block. If a cirecmferential indication is found, an attempt should be made to characterize l
7
~
\
l the depth and length of the indication through appropriate nondestructive tests. However, we do not advocate removal of cylinder liners for the sole purpose of this inspection.
I h
e l
L 8
a l .
r -
TESTIMONY ON CONTENTIONS ;
Q1. How is your testimony organized? ,
A1. ( 11) The testimony is presented in two general parts c,oncerning
- 1) the crankshaft and 2) the cylinder block.
1 - CRANKSHAFT Q2. What issues are addressed in this part of your testimony?
A2. (All) This part of the testimony deals with 1) conclusions that may be drawn from the qualification tests, and 2) the fatigue life of the crank-shaf ts currently installed in the Shoreham TDI diesel engines, designated as EDGs 101, 102, and 103. Item 1 is relevant to the contentions (c)(1) through (iv') and item 2 is relevant to contentions (a)(1) through (iv).
Conclusions that May be Drawn From Confirmatory Testing Q3. Can you comment on the purpose of the confirmatory tests done by LILCO to accumulate 107 operating cycles on EDG 103?
A3. (All) It is our understanding that these tests were conducted by LILCO primarily te provide unequivocal evidence that the high-cycle fatigue endurance limit of the crankshaf t used in EDGs 101, 102, and 103 is at or above 3300 kW. The tests also included strain gage measurements to determine if the stress field in the cam gallery region of the block is compressive. These cam gallery tests are discussed in a later section of this testimony.
9
. ~
Q4. Have you reviewed the procedures and results pertaining to the con-fimatory tests done by LILC0 to accumulate 10 7operating cycles on EDG 103?
A4. ( All) Yes. Our review of the test results has been provided to the Board in two reports, namely Post-Test Examination of Transamerica Delaval, Inc. Emergency Diesel Generator 103 at Shoreham Nuclear Power Station for U.S.
Nuclear Regulatory Commission Staff, by A. J. Henriksen, B. J. Kirkwood, W. W.
Laity, P. J. Louzecky, J. F. Nesbitt, and L. G. Van Fleet, dated December 3, 1984, and Post-Test Examination of the Transamerica Delaval, Inc. Emergency Diesel Generator 103 Piston Skirts and Related Components at Sharaham Nuclear Power Station for U.S. Nuclear Regulitory Commission Staff, by A. J. Henriksen,.
~
B. J. Kirkwood, W. W. Laity, P. J. Louzecky, J. F. Nesbitt, and L. G. Van Fleet, dated December 14, 1984. Our review of the procedures is based on LILC0's letter to NRC (Harold Denton) dated October 18, 1984, concerning the confirmatory test, and information provided in test data sheets and supporting procedures regarding the calibration of electrical switchboard instruments.
.5 Y. Why WG3 it ggt PC33iblC 10 drGW CGGCl'd3 IGG 3 TC5CrdiC3 INC accep lity of the crankshafts from calculations alone?
AS. (Sar nkshaf t can.uhtbi involve um .Mi es arising from the complh geoutY7% track 6Mti ared the vgrieffen; fn torque, bend- )
ing icads, and nttu
\
Merant
/
inp;f%% #11rF W . of safety must be employed to accommodate these unc inties. t appears to me that the analy-tical evidence alone doe ot provide a sufficient ba ' for concluding that the crankshafts e adequate for the qualifled load of 3300 k . unequivocal answer n be supplied only by an engine test for a sufficient time to
= u! te 107 cpc atW q2.
10 i
__9
Q6. Regarding the tests conducted by LILCO at a nominal 3300 kW, do you believe that they can be proven to have been at that value?
A6. (All) No. We noted several points that could affect the certainty of the teste'd value:
- 1. There was uncertainty with respect to whether operators had the flexibility during the confirmatory tests to operate at 3300
- 100 kW.
o
- 2. Instrument uncertainties could have introduced an error of up to 2.5% of full-scale power readings.
- 3. LILC0 reported that 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br /> were run at loads in the range of 3250 to
. 3300 kW and that 81 hours9.375e-4 days <br />0.0225 hours <br />1.339286e-4 weeks <br />3.08205e-5 months <br /> were run at leads between 3300 and 3400 kW.
- 07. Have you resolved these questions?
Busa A7. (Henriksen) I believe so. The points just identified have been addressed. First, based en a review of the testimony and the data logs pro-vided, I believe LILC0 operators did operate most of the time with the watt-meter indicating a load of 3300 kW. This is based on my belief that the flexibility provided by NRC in conducting surveillance tests at 3300 kW *100 kW does not really nean that the load will be set at 100 kW above or below 3300 kW during that test. Rather, as I understand it, when set at 3300 kW, due to the mode ci operation described in LILCO's testimony, the wattmeter oscillates between 3200 and 3400 kW. This is probably as close as the load can be controlled unless the governor load limit is blocked.
11 t -
I have also reviewed the level of possible errors involved in the load measuring system. According to LILCO's testimony, the wattmeter instrument error could be as much as 12% of full-scale or 1112 kW. An additional error of 10.5% or *28 kW in the remainder of the instrument loop could result in a total of 12.5% or 1140 kW error in measuring the load. However, the calibration data furnished for the wattmeter, dated November 10, 1983, October 1, 1984, and January 4,1985, indicated that the error in the meter never exceeded 40 kW in the 3000 to 4000 kW load range. Thus, inc1uding the possible 28 kW error in the remainder of the loop, the total instrument error appears to not have exceededh25% orh0 kW during any period of operation of this particular engine since November 10, 1983.
The 20 hours2.314815e-4 days <br />0.00556 hours <br />3.306878e-5 weeks <br />7.61e-6 months <br /> of operation reported to be below 3300 kW is considered to be sufficiently few that they are of little or no significance to the question of the tested load, especially since there were 81 hours9.375e-4 days <br />0.0225 hours <br />1.339286e-4 weeks <br />3.08205e-5 months <br /> of operation above 3300 kW.
- 08. Does the possibility that due to instrument errors the confirmation test may have been conducted at a load as low as 3230 kW mean that the endur-ance limits for the crankshafts cannot be confirmed to meet or exceed 3300 kW7 A8. (Bush) No. I believe the crankshaft is qualified for its intend *4 service even though some of the confirmatory test data may have been accumu-lated at loads slightly below 3300 kW. As I will testify in a later section, I am convinced from my analysis of engine load data that EDG 103 has operated at l or above an instrumented-indicated load of 3500 kW for about 3 x 106 cycles i with no evidence of damage to the crankshaf t. This strongly suggests that the l endurance limit is at or above 3430 kW, accounting for instrument error.
12
~ . )
l 6
Additional testing of 7 x 10 cycles at engine loads near 3300 kW would have been sufficient to propagate any cracks that may have been present because the crankshaft stresses at 3300 kW are quite close to those ac 3500 kW. Therefore, I do not consider it significant that some of the confirmatory testing may have occurred at loads somewhat below 3300 kW.
Fatigue Life of Crankshaf ts in the Shoreham EDGs Q9. Have you reviewed the testimony ,of the County and LILCO regarding the load profiles that the Shoreham EDGs will be required to provide?
A9. (Bush, C;c;t:n, Henriksen) Yes. Generally we understand the engines may be subjected to loads in the following categories:
- 1. Lc:' rp ket egefvalent te 3?00 ku det te cequenced sterti g ef large i
reelia; pumps fer the 'i et 30 to 50 ::::nd: Of a LOOP / LOC? : nt.
- 2. Short time intermittent and cyclic loads for a few minutes that may exceed by a few percent the " qualified load", taken here as 3300 kW.
- 3. LOOP /LOCA loads, assumed to be at or below 3300 kW after the first few minutes.
- 4. Loads that may result from operator error during the first hour of a LOOP /LOCA event, taken as 3800 to 3900 kW for times of 40 to 60 minutes.
- 5. Periodic testing loads of 3300 kW to meet NRC Regulations.
13
Q10. Oc you bcifc;c th; engfecc (EDC: 101, 102, and 103) ca- !"ste!"
loa of Category 1 as described above?
)
1 A10. sh) Short-term loads as high as 3900 kW for less than a ute )
under emergency c ditions are not considered to be a problem. A st all i texts related to fatig and to cumulative damage in metals te the effects of momentary overloads. An exa le is Collins Failure o aterials in Mechanical Design (1981). Figure 1, taken fr Collins (1 , p. 293, Figure 8.27),
illustrates the prestressing effect of ntary overloads on existing cracks and their subsequent delay.in prop tion.
Short-term high loads, en those approaching the ltimate tensile strength, do not gene- ly produce cracks and may, in fact, ovide a plastic zone around any xisting crack that retards its growth. The prece ng condi-tion mar y exceeds the short-term achievable overloads of these EDGs. It is my nclusion, therefore, that loads such as those identified in Category 1 a net-ef cencer .
Q11. Do you believe the Shoreham TDI EDG crankshaf ts can sustain loads identified in Category 2 as described above?
All. (Bush) I would like to offer some background information prior to answering this question. I have carefully reviewed the operating history of the Shoreham EDGs, particularly noting the operating time at engine loads at and above 3500 kW. In the case of EDG 103, which has undergone extensive post-test examination showing no damage to the engine (particularly the crankshaft),
I note that the engine has sustained over 3 x 106 cycles at loads at or exceed-ing 3430 kW when conservative assumptions regarding instrument error are included as discussed earlier.
14 2
r, - . -
~
3 ,
a, i
- Noj
/ D ELE.TED .
AKs o
No a Del cycles Number of cycles FIGURE . Delay in Crack Growth Following the Application of Single Overload S rce: J. A. Collins, Failure of Materials in Mechanical Design - Analysis, Prediction, Prevention,1981, p. 293, Figure 8.27.
.?
i 15
- i
The loads and corresponding hours at which EDG 103 is reported to have operated are as follows:(a)
Load Hours Approxima.te . hours at 3500 kW 119 Approxi.aate hours at loads greeter than 3500 kW 101 Approximate hours at 3900 kW 7 Any of several approaches may be used to predict cumulative fatigue damage from these loads. Miner's rule, more correctly termed the Palmgren-Miner cyclic-ratio summation theory, has been used for many years to predict the i
fatigue (endurance) limit of materials. An alternative method that provides better correlation with experimental data is the Manson approach, which takes into ac:ount the loading sequence. The predicted fatigue limit using the latter approach for the EDG 103 crankshaft would vary markedly depending on the sequence of application of the loads noted in the preceding summary. We are unaware from available information what the actual sequence was. 3 A conservative view is to assume that the beginning of the high-cycle fatigue limit is less than 3 x 106 cycles, and to define the lower bound of the fatigue limit as that associated with the lowest load at which EDG 103 was operated during the first 3 x 106 cycles. This would set the lower-bound value from the EDG 103 test at 3430 kW, based on an assumed instrument error of h70 kW applied to the indicated load of 3500 kW.
(a) Pacific Northwest Laboratory, Review and Evaluation of Transamerica Delaval, Inc., Diesel Engine Reliability and Operability - Shoreham Nuclear Power Station Unit 1, PNL-5342, December 1984 (p 4.22).
16
Table 1 is a summary of data from six references on the high-cycle fatigue limit for several ferrite steels. A significant message from this data is that the onset of the fatigue limit is close to 1 x 106 cycles, regardless of the ferritic al]oy, heat treatment, or surface hardening treatment. Note that several of the values are for aircraft or automobile crankshafts.
As illustrated in Figure 2, the fatigue limit of ferrite steels is essentially constant as a function of the number of cycles above the onset of high-cycle fatigue. This is unlike nonferrous metals, which have no clearly defined fatigue limit with time.
The steel used in the EDG 103 crankshaft is ABS Grade 4S, which corre-sponds roughly to an AISI-5050 steel in composition. The tensile strength is
_ about 100 ksi and the yield strength about 60 ksi . The mechanical properties would correspond to some of the 4000 series steels cited in Table 1, and, therefore, one would anticipate similar initiation of the fatigue limit near 1 x 106 cycles. *
=
LILCO's nondestructive examinations of the EDG 103 crankshaft following the 107 -cycle test provide evidence that cracks had not initiated in the 6
crankshaft during the initial 3 x 10 cycles at loads at or above 3500 kW as read on the wattmeter. Because crankshaft stresses at 3500 kW are not sub-Basso Opo.4 My Exnamanm or DNrA stantially different from stresses at 3300 kW,(2: df tcur:0d " re pente te tw Fa. A A- 84 K questica 12), subsequent operation at the latter load to bring the total cycles to 107would have been sufficient to cause propagation of cracks formed at the higher load. This is further confirmation that the high-cycle fatigue limit is at or above the value corresponding to 3500 kW minus known instrument error, or 3430 kW.
17
TABLE 1. Location of the Initiation of High-Cycle Fatigue (Endurance)
Limit for Several Ferrite Steels l
Beginning of g
Fatigue Limit Material Comments t
Reference x 10 Cycles (1) 1.0 1047 Steel (2) ~3.0 4340 Vacuum melted - longitudinal specimens
~3.0 4340 Vacuum melted - transverse specimens
~0.9 4340 Air melted - longitudinal specimens (3) ~1.5 4340 Completely reverse S-N curve (4) ~0.3 3130 Temper embrittled
~0.8 3130 Non-temper embrittled (5) 2.0 0.78% C Spheroidized 2.5 0.78% C Pearlitic (5) 1.5 4140 Quenched and tempered 2.0 4140 Shotpeened 2.5 4140 Nitrided (5) 0.7 (4140,x4340, VCM)((a)Quenched and tempered Shot-peened 1.0 (4140,x4340, VCM) a) 1.5 (4140,x4340,VCM)(a) Nitrided, polished nitrided
~3.0 (4140,x4340,VCM)(a) Nitrided (5) 0.8 4340 Automobile crankshaft -
normal heat treatment 0.7 4340 Automobile crankshaft - shot-peened
~2.0 4340 Automobile crankshaft -
nitrided l (5) 1.5 4340 Transverse specimens from crankshaft !
0.2 1.20% C Quenched and tempered l l
(a) Above are torsional fatigue results on aircraft engine crankshaf ts j including 4140 series, i
18
r.
TABLE 1. (contd)
Beginning of Fatigue Limit Material Comments Reference x 10 Cycles (6) : 0.9 3420 Quenched and tempered
, 1.0 , 1050 Quenched and tempered
- 1.0 4130 Normalized 1.5 Structural steel -
1.5 Alloy struc. steel -
~2.0 Cast iron -
(1) Hayden, H. W. , et al . 1965. "Mechani, cal Behavior". Volume III in
! The Structure and Properties of Materials. John Wiley & Sons; New York, New York.
f (2) Reed-Hill, R. F. 1964. Physical Metallurgy Principles. Van Nostrand,
t (3) Collins, J. A. 1981. Failure of Materials in Mechanical Design - Analysis
j (4) Hollomon, J. H., and L. D. Jaffee. 1974. Ferrous Metallurgical Design.
John Wiley & Sons, New York, New York.
! (5) American Society of Metals. 1961. " Properties and Selection of Metals".
' Volume 1 in ASM Metals Handbook. Novelty, Ohio.
i (6)' Marks, L. S. 1941. Mechanical Engineers' Handbook. 4th ed. McGraw-Hill, New York, New York.
19 l l
l
80 70 -
60 -
50
- y;:
Fatigue (Endurance) =
[ Umit gg _
E
=
30 20 -
Note: .
Fatigue limit. the horizontal section.
10 - begins about 108 cycles and cantinues indefinitely I I I ' ' I 0 gos ,,to 103 10' 10' 10s 107 too Numtwr of Cycles FIGURE 2. Typical High-Cycle Fatigue Curve for a Ferritic Steel (1050 AISI) i l
l 1
l i
20 l
l l
The point of the background discussion is now clear. In my opinion, the Category 2 engine loads that may result from intermittent and cyclic demands in the vicinity of 3350 to 3400 kW for times up to one hour or so are below the probable high-cycle fatigue ~11mit. Therefore, loads in Category 2 are not of Concern.
Q12. Can you quantify the rcletive strcsscs et 3200 k nd 2500 k"?
A12. ten) If one takes the bending stresses as em and inter-preted by Det Norske Veri for the Shore' ham cr af ts in their report 84-0099A of September 17, 1984, and imum firing pressures as read from TDI test curves dated 19, 1976, for a Shoreha, ine, then the relative calculat nding stresses are 20,450 psi and 21,120 psi for 3 W and 500 '", c:pect9fely.
Q13. Do you believe the EDGs can sustain the loads identified in Category 3 above?
A13. (Bush) As defined in the response to Question 9, all loads in Category 3 are at or below 3300 kW. I believe the endurance limit for these crankshafts is above this value. Hence, the Category 3 loads are not of Concern.
Q14. The engine loads that may result from operator error (e.g.,
Category 4) could exceed the high-cycle fatigue limit. Do you believe the crankshafts will sustain these loads for periods up to an hour and still have the ability to meet the succeeding load challenge of a LOOP /LOCA?
A14. (Bush) I believe the crankshaft can survive up to an hour of overload to about 3900 kW without crack initiation, but the probability of 21
crack initiation cannot be quantified. It is a function of parameters such as previous load history and metallurgical properties. The question then is, if a crack initiates during a LOOP /LOCA, will it propagate to the point of engine shutdown before the engine'is no longer needed? My engineering judgment is that the combination of a Category 4 transient operation followed by time at lower load / time profiles such as the LOOP /LOCA demand profile should not lead to crankshaft failure. The only way to quantify this judgment would be to conduct a three-dimensional finite elemen,t analysis combining the LOOP or LOOP /LOCA load histories that were imposed on a crankshaft having an initial crack and determine the final crack size.
I feel that any crankshaf t that is subjected to a sustained overload approaching Category 4 should be given careful surface and volumetric non-de,structive examination prior to returning it to service.
Q15. What LOOP /LOCA load profile did you consider in evaluating the ability of the crankshaft to sustain the . assumed operator error load?
A15. (Bush) I assumed the following LOOP /LOCA load profile based on data provided in LILCO's testimony dated January 15, 1985, and the Shoreham Final Safety Analysis Report (FSAR), Tables 0.3.1-1A and 8.3.1-2: ,
Time Load (kW)
Less than 1 minute 3900 1 minute to 3 minutes 3331 3 minute to 12 mirates 3266 12 minutes to 30 minutes 3265 30 minutes to 60 minutes 3253 Longer than 60 minutes 2617 22
Q16. Do you believe the Shoreham EDGs can sustain t;' P required monthly and refueling-outage testing at the qualified lotd of 3300 kW, identified in the response to Question 3 as Category 5 loads?
A16. (Bush, % r:t:n, " r" t:r) Yes. These Category 5 testing loads are considered to be below the endurance fatigue limit for these crankshafts. As stated earlier, this limit is believed to be at or above 3430 kW, based on the results of the testing up through the first 3 x 106 cycles, and is certainly confirmed to be at or above 3300 kW, based on the confirmatory tests that brought the total testing cycles to over 1 x 10 .7Detailed comments regarding these confirmatory tests, including our views on the uncertainties with watt-i meter readings, are provided earlier in this testimony.
In view of the fact that the endurance limit can be established with l
certainty as being only at or above 3300 kW, we feel that it would be prudent to limit surveillance testing to this value. The reason for this is that surveillance tests can add over 3 x 107cycles during the assumed 40-year life -
of the Shoreham Nuclear Power Station.
II - CYLINDER BLOCKS Q17. What is the purpose of this testimony?
A17. (Bush) This testimony addresses parts c(i) and c(ii) of the conten-tion concerning testing of the EDG 103 block, and also addresses metallurgical considerations related to my conclusion that existing cracks in the cam gallery region of the EDG 101 and 102 blocks should be monitored.
23 i
Q18. Have you reviewed the testimonies filed by the County and by LILC0 concerning the test involving the EDG 103 block, the suitability of the cylinder blocks in EDGs 101 and 102 for service at 3300 kW, and whether there is a need to monitor the cam gallery cracks in the EDG 101 and 102 blocks?
A18. (Bush) Yes.
Q13. Please summarize your conclusions on these issues.
A19. (Sush) My conclusions are as follows:
First, as I have stated previously in written testimony (file'd on October 12,1984), the replacement EDG 103 b1c:k was more suitable than either the EDG 101 block or the EDG 102 block for the tests that LILC0 conducted to obtain data on compressive and alternating stresses in the camshaft gallery.
Use of either of the latter two blocks for the cam gallery tests would have indolved the installation of strain gages over repair welds rather than over base metal. However, the selection of EDG 103 for the test at qualified load .
did not contribute to resolution of questions concerning the ligament cracks in -
the top surfaces of the EDG 101 and 102 blocks, the potential for developing stud-to-stud or stud-to-end cracks in those blocks, or the circumferential cracks reported in the original EDG 103 block.
Second, operation of the replacement EDG 103 block for more than 500 hours0.00579 days <br />0.139 hours <br />8.267196e-4 weeks <br />1.9025e-4 months <br /> at or above 3300 kW based on the meter reading, followed by LILCO's nondestructive examinations that revealed no reportable indications in the block top, provides significant evidence that the replacement block is suitable for service at the qualified load of 3300 kW. Based on the known performance of the block through the qualification test, I concur with the conclusion of 24
i Dr. Rau and Dr. Wachob(a) that it would be appropriate to reinspect the replacement block top at intervals determined through FaAA's cumulative camage analysis.(b) This means that if further operation beyond the most recent inspection does not exceed the FaAA-recommended interval before the end of the first fuel cycle, the top of the replacement block will not have to be reinspected until the first shutdown for refueling.
Third, the conclusions I expressed in previous written testimony regarding the EDG 101 and 102 blocks are not affecte,d by the qualification test performed with EDG 103. As I previously testified, I believe that the 101 and 102 blocks are adequate for service subject to certain caveats on surveillance of known cracks. Following any period of operation of EDG 101 or EDG 102 at or above 50% of qualified load, visual and eddy current inspections should be performed on those portions of the block top that are accessible between cylinder heads.
The purpose of these inspections is to verify the continued absence of detect-able cracks between studs of adjacent cylinders. Ir :dditten, the behavfer of ,
rever:1 rep-etentative crack: '- th ; shaft gallerie: cf the ECC 101 and 102 M eck theuld be ~^riter^d. !# nc ch:nge: 'nd!::tive Of crack greefth a e ebrerved ever the '* ret fuel cycle, the need fer centinued erf ter a; ef the i
ce= p'!e y cracks ceuld be recentidered by th '"' 0 .
Fourth, I have previously expressed the opinion based on engineering judgment that circumferential cracks of the type found in a cylinder liner (a) Additional Cylinder Block Testimony of Dr. Duane P. Johnson, Dr. Charles A. Rau, Jr., Milford H. Schuster, Dr. Harry F. Wachob and Edward J. Youngling on Behalf of Long Island Lighting Company, January 15, 1985, at 10.
(b) This analysis is presented in the FaAA report Design Review of TDI R-4 and RV-4 Series Emeraency Diesel Generator Cylinder Blocks, the most recent revision of which is FaAA-84-9-11.1 dated December 1984.
25
i .
4 ,
d l counterbore of the original EDG 103 block do not represent a hazard to EDG reliability. - My opinion on that issue remains unchanged. Similar cracks may also occur in the EDG 101 and 102 blocks because of the high stress concen-tration associated with the geometry of the cylinder liner landing. They may occur even in the replacement EDG 103 block, although the stress concentration i in the replacement block appears to be less severe. At any time a liner is
- removed from any of the three engines, it would be prudent to perform an i-appropriate nondestructive examination of ,the landing in the block. If a circumferential indication is found, an attempt should be made to characterize ,
I the depth and' length through appropriate nondestructive tests. However, I do -
q not advocate removal of cylinder liners for the sole purpose of this l inspection.
- x "cht t
- A; cf C:- C:1kry Cr:d; i- C"C; 101 cr.d 102 -
}
. How is your testimony organized on this topic? .
A20. (Bu I first will comment on the examination (a) perfo ed by Walter C. McCrone As i ates, Inc. of a cam gallery crack s imen removed from i
i the original EDG 103 block. will next briefly su ze my assumptions and I conclusions regarding the origin a character ics of the cam gallery cracks.
Finally, I will present my c'onclusions* r rding the need for monitoring cam gallery cracks in the blocks of s 101 and 10 and my reasons for those 4 conclusions.
i i
i (a) Wa r C. McCrone Associates, Inc., Cast Iron Analysis re LILCO vs Su ik
! meany (sic), MA number 13747, dated January 11, 1985.
26 L
, , ~ - , , ,- - ,-~n,-, -
,v--- ,-,--,e.,- ,.~n- --,,r~-r,- ,n,, . , + , -,,.,,,n,- m, , .r , , , , , . ,- , -.,,,.,-,-n--
s Comments on Testing Performed by Walter C. McCrone Associates, Inc.
The test results reported by McCrone provide unequivocal evidence at the predominant oxide in the samples removed from the crack surface was ag-netite. ThE x-ray diffraction patterns are unambiguous and can b readily interpreted by an analyst who is trained in the field of x-ray diffraction.
The McCrone laboratories are well known at the Pacific Nort' west Laboratory as having competence in conducting quantitative iron-oxide easurements of the type requested by the County. ,
1)ELETED Assumptions and Conclusions Regarding Ori 'n and Characteristics of Cam Gallery Cracks Based on the above-mentioned test r suits, I have concluded that the crack examined in. the sample removed from t e original EDG 103 cylinder block was formed during cooling of the casti g. There was no evidence of an oxide film formed at low temperatures, whi could have been indicative of" crack propaga-tion after the block was p1 ed in service. The absence of the latter oxide film tends to confirm th the crack is in a compressive stress field as deter-mined analytically an experimentally by FaAA.
Because the o iginal EDG 103 block exhibited degraded metallurgical pro-perties as conf rmed by the morphology'of the Widmanstaetten structure, it is reasonable t assume the following:
- 1. T e tensile properties of the typical Grade-40 cast iron in the EDG 101 and 102 blocks are superior to those of the degraded Grade-40 cast iron in the original EDG 103 block. The Grade-45 cast iron in j l
27
the replacement EDG 103 block compares even more favorably in thi regard. If one reasonably assumes that the hot tensile proper es of the EDG 101, 102, and replacement 103 blocks would also be tter than those of the original EDG 103 block, the depth of c . gallery cracks in the former would be expected to be shallowe than those in the latter.
- 2. With the evidence that cam gallery cracks i the original EDG 103 block are hot tears that did not propaga , and recognizing the superior materials properties of the G 101, 102, and replacement 103 blocks, it is reasonable to as me that the cracks in the latter blocks are also hot tears and t at these cracks have not grown in service.
D E LE.TED Conclusions Regarding he Need for Monitoring Can Gallery Cracks .
Based on the info ation summarized above, I conclude that the existing can gallery cracks i the EDG 101, 102, and 103 cylinder blocks would not be expected to grow nd'r normal operating conditions. Nevertheless, I believe that monitori of the cam gallery cracks in EDGs 101 and 102 is necessary for the reason listed below. I do not believe it is necessary to monitor cam gallery cracks in EDG 103, because the known cracks in the replacement block have not been repair-welded.
I l 28 u
~
- 1. The inferences and conclusions regarding crack behavior are based on detailed examination of one crack in the original EDG 103 blo .
This is insufficient data on which to draw conclusions with c rtainty regarding the other EDG blocks.
- 2. Associated with the known repair welds in the cam g eries of the EDG 101 and 102 blocks are residual stress fields of an undetermined nature. These stress fields could in, fluence ack propagation.
D ELETED
- 3. Cracks in the can gallery represent a do raded condition. In my opinion the known data on these crac where weld repairs have been made is insufficient to establish hat will or will not happen to these cracks over time. My co ern is related to the possibility of an initial lengthening of t cracks into stress fields of decreasing compression or, possibly tension. ,
- 4. Certain postulated rack growth patterns ultimately could lead to a loss of fu tion of a diesel generator. I recognize this is improbable, rticularly when coupled to the low probability of a LOOP /LOCA However, crack monitoring will provide confirmation as to whether or not the cracks continue to be benign. The action needed to rform the monitoring is straightforward, and I believe that it uld be consistent with good practice for safety-related equipment in nuclear service.
29
In my opinion, the preferred approach for monitoring the cracks ould be to install crack-opening d'isplacement gages at the weld overlays n the second camshaft bearing saddle inboard of each end of the engine. T se saddles are representative, and they are much more accessible than sad es toward the middle of the engine for any servicing of gages that ma be required. The gages should be monitored during monthly engine test . DELE 11ED Other methods of monitoring may also be ac ptable. One alternative approach would be to monitor the depth of te esentative cracks (e.g., at loca-tions described above) with an appropriat surface probe (e.g., a TSI depth gage), and also monitor crack length ( rallel to the longitudinal axis of the engine) using magnetic particle or liquid penetrant examinations. Depth measurements taken in this manne may lack accuracy, but the combination of depth measurements and lengt measurements would probably be sufficient to show any significant changes i crack size. To obtain the desired information in this manner with minin disruption of engine availability (due to the need to -
remove access cover , it would be sufficient to take these measurements every 3 months.
J i Regard 1 ss of the method chosen, it is my opinion that the monitoring should co tinue through the first fuel cycle. A decision should be made by the NRC s ff at the first refueling outage regarding the need to continue with the mo toring.
30
e Stud-to-Stud Cracks in the Cylinder Block Top Q21. Do you consider that the qualification test performed on the EDG 103 engine provides an appropriate basis for predicting the behavior of block top cracks in the EDG 101 and 102 engines?
A21. (Bush) No. Differences in the mechanical properties of the cast iron used in the EDG 101 and 102 blocks from the cast iron used in the replace-ment EDG 103 block and, perhaps more importantly, design changes incorporated into the top of the replacement EDG 103 b1'ock do not permit an extrapolation of test results from the latter block to the blocks of EDGs 101 and 102.
Q22. What are your views on the probability that stud-to-stud cracks could initiate in either EDG 101 or EDG 102 during a LOOP /LOCA and propagate to the extent that either engine would be lost from service?
A22. I consider loss of function of EDGs 101 and 102 under these postu-lated circumstances to be highly improbable for the following reasons: .
- 1. There is no evidence of stud-to-stud cracking in these blocks from previous operation at and above 3500 kW. Such cracks would be more likely to initiate at these higher loads than at the qualified load of 3300 kW. .
- 2. All future surveillance testing is to be accompanied by monitoring of the block tops of EDGs 101 and 102 to verify the continued absence of detectable stud-to-stud cracks.
i e
31 .
l
r
~
l
- 3. Based on extrapolations from the original EDG 103 block, I would not l expect the fatigue crack growth rates in the stud-to-stud area to l be so high that there would be a loss of EDG function during a LOOP /
LOCA, assuming crack initiation occurred shortly after the start of the LOOP /LOCA. This is particularly true at the low power levels--
less. than 3000 kW--characteristic of predicted load profiles through most of a LOOP /LOCA, even if one assumes the improbable situation that the engines would be the only so,urce of emergency power for approximately a week. A quantification of crack initiation and growth to the point of loss of function would require a three-dinensional finite element analysis in which c ;k initiation is assumed. FaAA has conducted such an analysis (FaAA-84-9-11.1, December 1984). My own semi-quantitative assessment is that the cumulative probability of crack initiation and propagation to the point of loss-of-function is quite low. ,
6 32