ML20086S222

Investigation of Types AF & Ae Piston Skirts

ML20086S222
Person / Time
Site:	Grand Gulf, 05000000, Shoreham
Issue date:	02/27/1984
From:	FAILURE ANALYSIS ASSOCIATES, INC.
To:
Shared Package
ML20086R808	List: ML20086S222
References
FAAA-84-2-14, FME-R-6-7396, NUDOCS 8403010446
Download: ML20086S222 (83)

Text

-, s r-d' a .$

Failure ENGINEERING AND METALLURGICAL CONSULTANTS 2225 EAST BAYSHORE ROAD, PO. BOX 51470.

PALO ALTO, CALIFORNIA 94303 (41G1856-9400 TELEX 704216 FaAA-84-2-14 FME-R-6/7396 INVESTIGATION OF TYPES AF AND AE PISTON SKIkTS TDI Diesel Generator Owners Group The report is final pending confirmatory review:;

required by FaAA's OA operating procedures.

Prepared by Failure Analysis Associates Feburary 27, 1984 ok$ .P PALO ALTO + LOS ANGELES

• HOU,STON e PHOENIX e OETROIT

• BOSTON

( s TABLE OF CONTENTS 4

Page EXECUTIVE

SUMMARY

.....................................................i1 1.0 INTR 000CTION................................................... 1-1 2.0 LABORATORY EVALUATION OF CRACKED (MODIFIED AF) PISTON fKIRTS... 2-1 2.1 Nondest ruct i ve Exami nat i ca. . . . . . . . . . . . . . . . . . . . . . . . . .

2-1 2.2 Destructive Examination...................................

2.3 El ect ron Mi c ro s c opy . . . . . . . . . . . . . . . . . . . . . . , . . . . . .. . . . . . . . . . 2-2 2,A Opti cal Met al l og ra phy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-3 2.5 Chemi ca l Compos i t i on . . . . . . . . . . . . . . . . . . . . . . . . . . .... . . .2-4

. . . . . . . 2-4 2.6 Hardnes?....................................

............. 2-5 2.7. Tensile Properties.......................... ............. 2-6 2.8 Resi dual St ress Measu rements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-6 3.0 EXPER IMENTAL STRESS ANALYS I S. . . . . . , . . . . . c . . . . . . . . . . . . . . . . . . . . . . 3-1 3.1 Comparative Description of Piston Skirts.................. 3-1 3.2 Test Set-Up............................................. 3-2 3.3 Stress-Coat Test.......................................... ...

3.4 S t ra i n Ga g e Te s t . . . . . . . . .' . . . . . . . . . . . . . . . . . . . . . . . . .3-4 ........3-3 3.4.1 Instrumentation................................ ... 3-4 3.4.2 Test Results................................... ... 3-6 4.0 FINITE ELEMENT STRESS ANALYSIS................................. 4-1 4.1 Load-Considerations.................................. ,... 4-1 l,

l

-4.2 Piston Skirt Analysis................................ .... 4-3

! 4.2.1 AE Skirt................................... 4-4 4.2.2 AF Skirt........................................... ........ 4-7

'5.0 FATIGUE AND FRACTURE ANALYSIS.................................. 5-1 5.1 Ma t e ri al Prope rti es . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . 5- 1 5.1.1. Fatigue............................................ 5-1 5.1.2 Fracture Mechanics................................. 5-2 5.2 Fat i gue Crack - Ini t i a t i on Ana lys i s . . . . . . . . . . . . . . . . . . . . . . . . . .

s -5.3 Fati gue Crcck Growth Ana lysi s. . . . . . . . . . . . . . . . . . . . . . . . . . 5-8 . . . 5-5 6.0 AE PISTON SKIRT INSPECTIONS.................................... 6-1

7.0 CONCLUSION

S.................................................... 7-1 1

I

- .< .a y

EXECUTIVE SUWWUtY

,:~

PISTON REPORT This report addresses the structural integrity of the Transamerica Delaval, Inc. (TDI) types AF and AE piston skirts. The report was prepared for. Long Island .Lightir.g Co. (LILCO) on behalf of the TDI Diesel Generator Owners Group as ~ one of a series of reports on generic components of those diesel engines in nuclear installations, the generically termed Phase I

. components.

. The piston skirt types evaluated are common to the LILC0 Shoreham Nuclear Power -Station DSR-48 engines and for the Mississippi Power and Light Co.,

Grand Gulf Nuclear _ Power . Station, RV16 engines. These engines are rated at the same speed, bmep, and horsepower per cylinder and incorporate essentially P int'erchangable cylinder liners, pistons, _and cylinder heads. Both plants experienced linear indications in toeir modified AF piston skirts and both plants have replacad the AF skirts with type AE.

A ' future report will evaluate the other TDI piston . types (AN and AH) in

-nuclear service.

All of the type AF piston skirts - originally installed - in the DSR-48 diesel engines at Shoreham were found to exhibit _ linear indications during dye l penetrant inspection in one or t ore of ~ the crown-to-ski rt stud attachment

. bosses. . Tne ' indications were -found to be fatigue cracks upon - destructive

' metallographic examinations. The cracks were ' determined not to result from material or fabrication defects.

The AF piston skirts had. been factory modified by TDI, to replace the original. spherical' washer ' sets used in the stud attachments by two stacks of

~

lBelleville washars. These skirts were replaced by a later version, type AE,

'which incorporates increased stud attachment thicknesses and eliminates one of the Belleville washer stacks.

11

c .

4 1

These Ai skirts._ have now been operated for over 300 hrs. in one of the SNPS engines including '100 hrs, of full power operation. Other AE skirts have

. accumulated.over 6000 hrs. .in a stationery generating plant, and over 600 hrs.

'in? an ; advanced _ development engine. Inspection of ~ these skirts (one after 6000 hrs. - two after 600 hrs., and four after 300 hrs.) with a high-resolution eddy current procedure disclosed no cracking.

Comparative. analyses of the modified AF -and the AE skirts under identical

-loading.. conditions demonstrated that stresses were significantly reduced in the .AE type. 'The absolute stress levels calculated did not include thermal

~

stress or the distortion of; the piston crown and wrist pin, but were neverthe-

less useful for comparing the two skirt types. The calculations showed that

under the. sade operation conditions in which fatigua cracks could grow to the

depths observed 1.n the modified- AF skirt, cracks would not grow at all in the

. typeJ AE skirt, even ' if they were . to occur. Further, based on strain gage tests and available. inspection data, it appears unlikely that cracks will form in the AE skirt.

E{oerimental stress- analysis - of s a -type AE skirt was conducted under hydrostatic' loading ..of the piston . crown. - The maximum stress measured was below the yield '. strength of . the' materia'l and' corresponds - to cyclic loading below the value required to produce a crack.

.It was concluded that, ' based on both analysis and test results, the type E. AE skirt' attachment ~would not fail in fatigue.-

+ .

L L-  !

jjj.

l

[ .

i

1.0 INTRODUCTION

Inspection of 24 piston skirts from the Transamerica Delaval, Inc. (TDI),

DSR-48 engines of the Shoreham Nuclear Power Station disclosed linear indications in one or nera of -the skirt-to-crown stud attachment bosses in each of 23 of the skirts. The one exception proved to be a later version of the skirt, designated by TDI as AN-type. With this one exception, the piston skirts were the AF' type, factory modified according to a report issued bj TDI under 10CFR ' Part 21 on November - 5,1981. This modification, performed by TDI

' late in 1981, consisted of spot-facing each of the four bosses through which _

studs extend to secure the piston crown and replacing the o,iginally supplied spherical washer set with two stacks of Belleville washers. The spot-facing reduced the height of the stud attachment bosses from 2 inches to approximately 0.25 inch.

The AF piston skirts are no longer in use at Shoreham. LILC0 has replaced all 24 piston skirts with the latest IDI production type, designated AE. :The AE design restores half the original height of the attachment boss and incorporates.one stack of Belleville washers instead of two. In addition, the boss is wider and is blended' more smoothly to the skirt wall. These changes provide additional material for support of the loads on the top of the

~

skirt. This _ redistribution of material provides improved stiffness and strength in the AE skirt. The major differences in dimensions are illustrated in Figure 3-1. Thus qualitative review of the AE design indicates that it has improved strength and crack resistance over the AF design.

In - order. to provide a quantitative ccmparison of the two designs, FaAA undertook a comparative stress analysis of the two skirt configurations. This consisted of finite element analyses of the AE and AF skirts and experimental stress analysis ofLthe,AE skirt. The same boundary conditions were applied in the. finite element analyses of both skirt configurations. The boundary condi-tions were simplified in order to have a finite element model that could be analyzed. with the. capab.lities of current computers. The finite element stress results are presented in Section 4, and the experimental results in

-Section .3. - The finite element calculations showed that the peak stresses were 1-1 I.

h 9 - . . _ , , - - -

t significantly lower in the AE skirt. Strain gage measurements :nade under hydrostatic pressurization of the piston crown showed that stresses were lower than those predicted by finite element analyses. Thus the model was shown to be conservative relative to the test conditions.

The analyses demonstrated that the ,AE design has improved fatigue resis-tance over the AF. The experimental results would indicate that crack: will neither initiate nor propagate in the AE skirt. However, a comparison of the AE .and AF skirts performed using finite element results under conditions in which the field observations of cracking in the AF could be predicted revealed e.

that cracks would initiate but not propagate in the AE skirt configuration.

6 y

f l

i 1-2

2.0 LA8 ORATORY EVALUATION OF CRACKED (MODIFIED AF) PIS10N SKIRTS Metallurgical evaluation of three type AF piston skirts from the SNPS emergency diesel generators disclosed fatigue cracks in all but one of the twelve crown-to-stud attachment bosses. None of the cracks had grown to failure. _The chemistry and microstructures appeared consistent with typical ductile iron properties. TDI reported that all SNPS piston skirts were manufactured to ASTM Specification ~ A536-65 for ductile cast iron. Skirt Brinell hardness values were within the specifications attributed to the manu facturer.

According to TDI, the AF skirts were normalized and tempered. They were reportedly austenitized at 1700 F - 1750*F for 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />, then cooled in ambient air.- They were reported to have been subsequently tempered at 1050 F for 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br />, then cooled in ambient air. AE skirts were processed similarly except that they were cooled in reci rculating air after normalizing at 1700 F to 1750'F.

Three of the skirts identified as having potential cracks were shipped to Failure Analysis Associates (FaAA) for confirmation and analysis of the cause of these cracks. These skirts have been designated Nos. 1, 2, and 3. The No.

I skirt was removed from the No. 4 cylinder of EDG #101, The No. 2 piston

. skirt was from the No. 6 cylinder of EDG #102. The No. 3 skirt was from the No. 7 cylinder of EDG #103. The Nc._3 piston skirt was previously rejected because of prior scuffing damage.

2.1- Nondestructive Examination Figure 2-1.is a photograph of one of the modified AF-type piston skirts after disassembly from the crown. The interior of a piston skirt is shown in Figure 2-2. - The white color is from the dye-penetrant test. The spot-faced b*.sses are shown in Figures 2-3 and '2-4. A linear indication of red dye from tha dye penetrant test is seeping out of a suspected crack as shown in Figure-2-4. All' of the cracks found were similarly located and oriented on a sp;t-faced boss.

2-1

2.2 ' Destructive' Examination Three AF-type piston skirts were examined by eddy-current to confirm the presence' of cracks; eleven of the twelve bosses contained one or two cracks.

The cracks appeared on the 'inside of the- skirt in the vertical ridges remain-ing after spot-facing. of the bosses. A total of four linear indications, revealed by dye penetrant testing, were subsequently saw-cut and fractured out

.of the three skirts. These indications were opened by fracturing and proved to be pre-existing cracks.- All the ' cracks were located at the base of the vartical. ridges ~ approximately 0.1 inch above the spot-faced surface. The ,

~ crack size varied from 0.3-inch deep by 0.9-inch long to 0.2-inch deep by

~

0.~3-inch long. Location of these _ cracks is shown in Figures 2-3 and 2-4.

Macrophotographs of the four opened cracks are given in Figures 2-5 to 2-9.

These pre'-existent cracks appeared to be fatigue cracks. Some of them had faint. beach ~ marks. - In addition, no evidence of ductile dimples or cleav-age Jwas . observed .ir the flat' fatigue crack region. The exact origin was

impossible to locate for three' of the cracks. However, one crack appeared to have. its origin at the tip of the ridge that remained after the bosses had

. been spot-faced. The overloaded portions of the fractures . produced in opening the. cracks were rougher than the smooth, pre-existing cracks. A notable fea-ture, visible using 'a iower power stereomicroscope, was the many graphite

. nodules intersected by the fracture surface.

- These nodules were smoothly l - fractured inf the fatigue ciacked areas but were more disturbed in the over-r

[ Icaded regions of the fra'ctures. Several of the pre-existing cracks had The' distinctly d1fferent appearance of the pre-

~

L faintly visible beach marks.

l ' existing fracture surfacessidentified them as fatigue' cracks.

On several of 'the fractures, . black included material was apparent near t as-cast surfaces. Instances of the included material are noted and outlined in ' Figures 2-5 'and 2-6. This material appeared to have been organic binder

.that 1had been . incorporatad from the, core. The included material did not fappear . to . significantly affect the initiation or growth of' the four cracks I

'which were broken open. .

f 2-2 l

id- ,- i>,,, , ~ ,, . - . , . ~,,w,.., ,..,,,,------mw,,---, ,-,,#, + - - - - , ,--, , ,,.-w,- ---

2.3 . Electron Microscopy Cracks from each piston skirt were examined in a scanning electron micro-scope (SEM). ~The character of fractographic features varied with location on the' fatigue crack surfaces. Locations adjacent to the tip of the ridge, near the suspected origins, were more rubbed and damaged than regions t. loser to the -

crack front. Near the crack front on two fractures, i.e., in the fatique region most recently formed, areas containing' fatigue striations were observed. Although these st'riations were not observed on each specimen or in other areas on the crack surfaces, features were observed that were similar .

but without distinct fatigue striations. Worn or abraded features, such as thoso - observed here, are typical of fatigue crack surfaces that result from cycling between tension and compression.

Since the crack that was removd from EDG #101 is typical of all the observed cracks, its fractographic features are discussed. A low-magnification  !

SEM montage of the EDG #101 crack is shcwn in Figure 2-10. As one moves away from the suspected origin ' area - (region A) toward the crack front (region C),

the fracture surface shows evidence of less rubbing and abrasion. High-4 magnification SEM. photographs of regions A, B, and C are given in. Figures 2-11 through 2-14.- In region C, near the most recently formed crack surface, fatigue striations can be seen. Fatigue striation spacings, i.e., the amount

-the crack extended with each fatique cycle, can be seen -in Figures 2-14 and 2-15. The spacing of the striations is approximately 1.2 x 10-5 inch / cycle.

These cracks in the piston skirt ridges were clearly the result of

- fatigue. .First, striations and rubbed fatigue areas were present over the enti re1 crack surface. Second, no evidence of ductile dimples was observed between the graphite nodules. . This strongly suggests that the maximum cyclic j stress was below the ultimate tensile. strength.

In addition, it also suggests that major overloads were r;ot responsible for crack extension.

For- l comparison, an uncracked piece of this piston skirt was fractured under a i bending. load. in the laboratory. In a very narrow region near the tensile surface, ductile dimples mixed with some cleavage were otserved. Such a r;gion is shown- at various magnifications in Figures 2-16 through 2-18. No such features wer:e found on the fatigue crack surfaces.

l 2-3  !

l

L Finally, no cleavage was observed on the pre-existing crack surfaces.

Evidence of cleavage was found only in regions where the fracture was fresh, i.e., caused by overload when the specimen was broken open in the laboratory.

Cleavage features, such as those shown from the exemplar failure in Figtres 2-19. through 2-21, were not observed on the fatigue crack surface. In addi-tion, the graphite nodules were more disturbed in both the cleavage and duc-tile dimple region than the nodules on the fatigue crack surface. This again supports the evidence that these pre-existing flaws were the result of high cycle fatigue and were not caused by ductile or cleavage fracture processes.

2.4 Optical Metallography The ductile cast iron was examined metallographically. Small specimeas, removed from an area near the fatigue-cracked bosses, were mounted, polished,

- and etched to reveal the microstructure. The AF microstructure consisted of about 10-volume-percent graphite par':icles in a matrix of pearlite and fer-ri te. . The graphite nodules were somewhat less regular than ideal; however, this would have had no influence on the fatigue strength of the material. The matrix was about 75% pearlite an'd 25% ferrite; the ferrite appeared around the graphite ' particles, as expected. Figure 2-22 shows a polishe'd and etched photomicrograph of a sample of ouctile cast iron removed from the No. 2 skirt of the piston from the EDG #102 engine. This microstructure is consistent with the heat trectment reoortedly used by TDI.

The microstructure of the AE piston skirt is similar to that observed for the AF piston skirt. The amount of ferrite is approximately one-half that in the AF skirt; this reduction in- the volume fraction of ferrite is consistent with the-increased cooling rate imposed after normalizing the AE skirt.

2s5 Chemical Composition A sample of the No. 2 AF piston skirt was removed and analyzed by Metal-lurgical Testing Incorporated using various chemical analysis techniques. The resultant, chemical analysis is shown in Table 2-1. This chemistry is typical fcr ductile cast iron.

2-4

.. a Table 2-1 CHEMISTRY OF NO. 2 AF PISTON SKIRT Element Weight Percent C 3.46 Si 2.43 Mn 0.53 Cr 0.11 N1 0.76 <

Mo 0.038 Mg 0.072 S 0.005 P- 0.027 2.6 Hardness Brinell hardness was measured on several pieces removed from the cracked piston skirts. The measured hardness values varied from 235 to 255 B'HN; .the average hardness was 241 BHN. These hardness readings fall within the speci-fied Brinell hardness range, 217 BHN to 269 BHN, of ASTM-536-67, grade 100 03. This notation implies minimum tensile properties of 100 ksi ultimate strength, 70 ksi yie'd strength, and 37. elongation. This grade has a pearlite matrix that is consistent with the microstructure observed in the examined AF piston skirts. This is the grade specified for Shoreham's AF piston skirt.

The piston skirt hardness values suggest that the ultimate tensile strength would be between 110 ksi and 120 ksi. This. ultimate strenoth would be con-sistent with ASTM A536 grade 100-70-03.

Brinell hardness masurements were made in the same locations on an AE piston skirt as on the AF . piston skirts. Hardnestes measured on the inner skirt web were between 262 and 285 BHN; the average herdness was 277 BHN.

Hardnesses taken on the boss face ranged between 229 and 241 BHN; the average

- value was 235 BH6l. This is similar to the hardness values, 255 and 261 BHN, measured near the wrist pin by FaA7 and TDI.

The slightly higher hardnesses measured on the AE piston skirt than on the AF piston skirts are consistent with forced air cooling after normalizing.

.These values appear proper for 100-70-03 grade ASTM A536 ductile cast iron.

These measured hardnesses are within the typical range of hardnesses for 100-2-5 L_

70-03 grade nodular iron. It should be noted that ASTM A536 does not specify hardness _ requirements.

2.7 Tension Tests Mechanical property test samples were machined from two of the three fatigue cracked AF piston skirts and from an unused AE piston skirt. The ten-sion test samples were cut from the piston skirts from a location as near as possibletothefatiguecrackedridgesadjacenttothespot-facedbosses. The specimens conformed to the ASTM specification E8-80; they were 1/ t. inch in -

diameter. The specimen deformation was measured with a snap-on extensometer with a one inch gage length. This method of chosing a specimen size and its location in the piston sxirt did not comply rigorously with ASTM A536-65.

This specification recommends that the properties of the cast ductile iron be measured using specimens machined from separately cast and heat treated Y' blocks or keel blocks rather than with 1/ t, inch diameter specimens removed from an actual cast part. The tension test results are reported in Table 2-2.

It is- apparent that none of the tension tests complied totally with the mechanical test requirements for grade 100-70-03 ductile iron castings.

I" larger . tension test - specimens had been machined and tested from separately cast and heat treated test blocks, the results may have been in compliance with ASTM A536-65 for grade 100-70-03 ductile iron.

1; 2.8 Residual Stress Measurements Residual stress measurements were made on the piston skirt removed from the- No. 6 cylinder of EDG #102 and on the unused AE piston skirt. A dissec-

' tion technique' was used to determine the principal residual stresses near the ridges and boss region.

No significant residual stresses were measured in either the AF or AE piston skirt ' iu= this region. The maximum stress measured was 11.4 ksi on the spot-faced region of the AF skirt.

2-6

Table 2-2 TENSION TESTS OF SPECIENS TAKEN FROM PISTONS Specimen Yield Ultimate Percent Elongation Number Strength Strength at Fracture *

(ksi) (ksi)-

AF Fistons 1-101 64.5 94.2 5.2 2-101 63.3 97.5 4.8 3-102 53.6 100.0 6.4 4-102 61.6 98.6 5.4 AE Pistons AE-1 68.0 90.21 2.3**

AE-1-2 70.5 85.36 1.6 AE-2-1 63.5 93.4 2.8 AE-2-2 65.9 91.5 2.4**

• Determined from'extensometer output.

• • Fractured outside of gage length.

O 2-7 w- , w ~- -- r y c ,

.s

,'  ; 3 -

q .: . ,,1

%[

Ji .,.;, ,,

. i . .. wt

..' w

j. * .

i

.): . . .

4 '

n .

.-:q

,.4

s

.g3 .'.,

-[4 s .;

.\ j g -h

' ~

hh)*${bihb$h.II.SjdWe. 1

.1i

..eps. cyg%;ag%Wf n9-%,.y9;&.Yrp ciplw

? '-c' '. .:

%. &,._:av 1./

_ s_=

i:

i s c;.t. W g, n g.gy.,.q.,d  :

. {tigir - 3 h

g.dde y..d.;;;_ .- .. N i

Figure 2-1. Photograph of an AF-type piston from the Shoreham Nuclear Power Station emergency diesel generator piston skirt (left) and

, crown (right).

i 4 7' G.- . eug ~ ' Nr g, s 'lp N, e

[1f 4 b a.

• G .. , p; ye

+ .

t A.y I .

l.

l Q,b .

l 'y4 1

o -

A -

1. 4!

\ #l.3 .., ,

i ftl N l- ,

' 4 .

a,

. s ,

I

~. ..

7

j Figure 2-2. Interior of an AF-type piston skirt. The white coating and red dye are from tne dye penetrant examination. Through holes f ar

l > the crown studs and spot-faced boss are circled.

l l 2-8

'}

Fa AA-PA 14 ._

ff ,

.l^f, .

'[_' ' .jdff

..~ '

y, n s .

<g; l - .' j ; . .

, c. . , y - . .. ,  ; . . .

m j.

~

.* .., . . . . c, .. ' .

a 3r- - '

.- 1, F

.g' ,

1. 9 ,

g

, "; s ,.. e ., ' # - -

. . o ,4 .

.. . , . =

S

' ?, : ', Pfh.NQ g

, a. -

q.gc.;4.. ~,)z. ..,, . ;.e...y.%

..r.....

.w. '.

.....;.q:) . . . ..-. , . .: .. .. o =.-

, l , .

. . a

. ., y _

y l \ ., .. .. .

s - a 4 -

1 . .;. .. 2 ,1  ; .

. r%v**.y ,9/.

.f ' r . .r..' , '

?y. W.,, y .-M?%,....<

.. >.,. ##*P $Y;. yep. xh ,,; _ . .a..R s.,, e. W-+n.iW} d. ~ i

., t$ 'E' .% ' U.5 +%'4%f ~. _

~._y g y 4 % , r N V . N .'. . . .. .. Y ' l-l , . _.].sif .8 *l. : %

• g a

Figure 2-3. A crown-to-ski rt attachoent noss inside a piston skirt re - $

no ve i f rorn ED3 4101. The ho s has 5een machined by spot-tacing from g

2 inches high down to the observed 1 /.. inch, Tne remavel seg nen t af a5 ridge was subsequently broken open and is shown in Figures 8 and 9.

E

- ;m.,~

gWM' n.;..w~

9 k',,~5_ ,

~ , . .

. 4 a

=_7

,E r

A 8

1 -

?*,:94 *%t

. a, .  : e.= M $ ,

q s

a, &

s..(

g '

-m u

- , .ed .4 w. m 6 - &

. 4, -  : m

• • y .;

. 4 y < { .p 4 ' . ,.

,hli ~ '

tr ,l b a g g % p$ a.-.5 ; n*.-

& Lo. 1

.s - rv w .,

y

. .m j y, a <. . ".

h fo$Q'

, g.j;,gys,< .m... -

.s $

3 =

@ 'h , ""* ... , hj$72$ rnq . 7 L ,

c)l .

Wf> - 4D,mj j tyd , - rl;+ c 3 l

i ntf ..

n, <e-6 9

Figure 2-4 A c.rown -t o..s k i r t attachaent Mss f rom ED3 cl02. T ne ,)i ton ]

skirt has heon taste 1 with rol dyr penetrant and soire 1, > s oloper.

red 1inear i n 11 c it i on '; vi s i bl e at 'M e hisa af '* ve tical ridje A

}

j thdt renained aft'r Spat fTcimj. j i

"7 ~) r =

t 3 AA 2- 14 m -

N

• s p ~ - i 69 .

_.,i m a w ~ -

.. _ y 'i!. .; . -

,,p#.

f , ' ;

, . ~-.y f N., "

~ ,. ,.

. ' = _ q. . ,

t, .

Figure 2-5. Macrophotograph of a pre-existing crack cut out

< of the piston skirt from EDG #101, No. 4 cylinder.

I$

f f

i -

l .,

i-1 I,

i i .j-  :

i y *f ri a In,1h)

_ f,, . e'

]

l

!! Figure 2-6. Macrophotograph of a pre-existing crack cut out

'I of the piston skirt from EDG 4102, No. 6 cylinder.

I 2-10

.i FaAA-83-11-13 E . , _ - . _ _ . . _ _ _ _ _ _ _ _ _ _ . - _ . _ _ _ _ _ _ _ _ _ _ _ _ . . _

.t * -

il

.i e t '

g.

.M- ,

w w e# . $K % w,%x.

l; n* O

. *. . m? , 4. .g.v. h,i_ x',l%y

.*". v , .

-;T <) t Q . %,- .

? . ; '.'/5h, YI?sr>;$'g ^'* QQ;~

% we.'k? % ,4 I'~

j 1 l

';f {.

.--iw -J5' % %',9 Q . .. .

p.'

• 2 a.

5-

. . '. [

l

,e Mk).h  %,#. .,:.*.., .

/ p, .. ., , . y a.

s ,.. 'J,.:a p.,

3..

. a ,t. p-

. ,; , f. .r.: . v s.c,e

, eg.? .

• '**'.._ _ _-. <..~...' -

l

,= *-

5 . -q -

Figure.2-7. Macrophotograph of another pre-existing crack cut out of the piston skirt removed from EDG #102 and fractured

,, open.

1

~ .34>sg ._ . -

r .

4l .

g .. . .

~

ig s

9 ,. f.4 s,.

-- 7 ;,

e -

c ~

. s .~, .

.-e.

~

_- .- - . -- . l, n . *,

3 u g

. . ,5

. . s .

y . . , ..

g . a=

. e

4

'R 1 Figure'2-8. Macrophotograph of a pre-existing crack cut out of EC3 #103, W. 7 cylinder piston skirt and fractured open.

This low magnification photograph shows both of the mating fracture surfaces after the ridge segment was opened.

2-11 FaAA-84-2-14. _

f l

l I

r 1' '

. s s -

~

p%. I% f i, .. . .

?kh'h'tF

& 'Y  ? '

l' f'5fW p'* ,

l [ ,,; . : r .' i .i -

sf.5' * ' ._-

bY ,L l' .

fjf -- - .

'~

h Figure 2-9. liacrograph of the fracture surface from the pis-

~

ton skirt removed from EDG #103.

7

! i 3

I a

lI

s

)

l 2-12 i

I$ FaAA-84 14_

{,

7 t s rrn i uo ksi

. s t

- ' ea o

g n o ui rc t tf s ci l* ' . ian

.f f

s prg f a e g m

.~

hnr t

" ~ _# .p i e th y  ;. nag i mi 9  ;,

.}, &,, j[rps* l[ ,a 's ,

de nht uta h

. , g o

ff n O 4 ow k'5[ru '

i '.

, ,?lg;;.&~-

4 j. ) '0 k

c nh a k r ae es o

f'sk%f P }t' ^

y c tr j 0

%.yA"= , -

a

%"  ;,.c a e e**

gs naC i.h {. I,f,"e la ' .

i w

a {j g ,.

i M. t d _

,g / 9,3U.

s .

s on o9 *4 j,WE Ev. .

,y

. it a xo

_ eh ,

k m^S

'9 K - pB

/ g. , .

0 3 er s ,

.kg:g g: r._

pia I , 2 h F: -

'i I eTs r h a i_ [k, * .

ARM, t

.r e

g* 0 Y,%. ,, f1A W"W*g.M[h+

r .

6 o01

1. • • h, I*q

?- 6' 9

~

.',: 3 .,

,Y k

$2 m..

.'. g%

N b

9 h p

a D r

G 2-5 j

~N 7 g E e g

%g 3e - o t m r

.v,' hr o o iug N pf FX g

, yd }~ r - '1 Me r n1 8

x,y(,Y_  %

E d i

• 'p4 .

S nn.

-g iw4 el o1 h;g t yh -

.yA - ics a2

,1 e.

9 o4eo C: ,

kk:t p m .lp t

q. D '*

p7

. ~ +

.k

4

)'

Am oom1 cNa1s -

2 n6

.' . 9' " oe rh s x._GIs%';. . .

0

.ft e r

' 1df u

- eog

• O'Y7.'.

4, 5

5d. -

Pyo\' :

2v emc oeF i

rean urfi g

i F

F*

> {l 5

, g > i + :'E

' .i 4 ,l l; ,i f: .l

i f .

1

-' t.

l

-%g l l 5 v ., .

l , ,

We' V 4 r, . .

j

. :s 1

o

'i ).s, f1

'. LQ

.. . ~ . u-

<. Y ', . lr ra ---

1 Figure 2-11. Area A of Figure 2-10. This SEM photograph shows how the fractographic details of the f racture have 7 been rubbed and abraded. 2KX i

! + y $l l ,

'l.,

... ;p 4 o 3

?

- v k'. 5' I ,i. ,p ""

g J >'

l1. 6

'1 -.

7,f ;.i, '. . , k}/

,z _

(

s  ;\

,1.*,.g <

4 , '. _ t . m .71 w '-

4 i

Figure 2-12. Area B of F i gu r*.: 2-10. This SEM photograph shows an area nearer to the growing crack front than area A in Figure 2-11. It is abraded, but not to the sane e(tent as show, in Figure 2-11. 2KX

2- 14 FaAA-84 14

.J

n. . . - - . . - , - - , , . , - - , - . - - - . . . - , . - .

1

-s ,

h .i'

~

~ .,

. , T

~

k, ,

W

['y i .

" .h, . . , , . .

!g s

.\ [t. > *, 4 ;

f u,% 'g.

a. , .

. *g,3.

l3, i G'g ' -

, u- 4,3,4,. t.

1

._ . . \ .1ti g , - .

g,

s. . l sM , b '!

i> ~

l

, Figure 2-13. A low magaification photograph of area C in Fig-ut e 2-IrJs :This !.194 photo shows the end of the crack front

'and a highlishted. region, 50 tri behind the crdck front

% .which ,is . shown at - higher magnification in subsequent fig-ures i'

~

. A *e 4 8

+

& ,,y wur%.v.cgme.w

-J b i.x  : v ~W.; ,E

..b N'

, pg .

?% :p.

p A.  ;

- s ;

j -

y,;

., $fy? m-,

Y:5 ..

("~> y 9: ,\g

'~

.f'

&3.c .9 ,~

• .h\

't . M' Y , _.

Figure 2-14. SEM photogrdph that . shovts the center region of

< Figure 2-13 at nigher magn i f i c 3 *. i on . Fatigue striations

.; y are visible at renter of the picture. s 2KX

' ' a AA-M M

{ 7-15

-i

\

yps -

..e 7 3 r ~~~

• • • - ~

(

p ) r,

^

-f

.s' 't g.spf& g-a.

o W$,1 h, -

Q

' i:

31 b - (?

'l .

,e -

j Figure 2-15. This SEM photograph is of the center region of Figure 2-14. Fatigue striations, ~1 x 10-5 in./ cycle, are

~

clearly visible. 10KX

t i

m, m es. m-~ n .. ,~

• gg4

'1 sf .m}v[g4,3, y ,{~'. ' - -

y

ff  ;

N '$ , ~

g .,

3.

y f'o ..

!). pg %

u

~- , s, t .

4

(',;. r* *= s h m

.I Et .'

h,kkT*3'.

Y , j k l:,fg g

.p  ?.r r ;. , y, 1

" >; ~~.hd$ N.A'.47ed'p. .4:,;

IS 7'M(, 3, #'h '3 ' 6 ,

~% / gyn; , 7:5 v ;,QqZe;,

%T! h ,

f

, .. ~+ ?w' %. !

hki,.1:.N'. [ h('

' . fi  :

d

.i

• This SEM photograph shows an exemplar overload

, Figure 2-16.

.q fracture area from the EDG d101 piston shown in Figures 2-10 to 2-15. thte the cleavage of the matrix , the shallow

secondary cracks, an1 the <1isturbed graphite nodules.

FaAA-84 14 4 2- K)

.. ~ , . . .

.. __-= ..

l e , - rg pw

,,y ss ***-

f j  !? - y lf'~ >

'i $'

• ,f ,

. h i

I QR * :S . . .-

- r, .

g. .. ;

? '

' "~""

._5%> <-

n .. .

'9T

j r.v.{

, ~ YO

_A C I

4 Figure 2-17. This SEM photograph shows a region of matrix cleavage in the center of Figure 2-16.

1KX l

Y~

/ fy * .

j . N'

!j ef 3p b. -s.. b).S'i' l

o 9-l, i a, -

-- p As;

}

. N.l;q_

o., .. y;4 % . .:;L . -

'a a '.L~

< > N .Q 3' tyj}#) ~ - . .. .N .g 2- ~

s c. &

i 1

i_ t VM ' J , W ._' '

s

~?

,2.= M .%c D' ~ ~+ . %bh)r.l; Q

• 6 7 QdI;7"'l1(1:_  ;

. \ , h ?5$ @ ' % % $ N W ].j Figure 2 A disturbed graphise nodule is shown in this SEM

, photograph of the overload region of the fracture shoan in Figure 2-16. 500X

.. 2-17 l Fa AA-84 14

p e

l

L

[7/t2 ~

^'

s,, g -

jr * ,

?1? .,

.- ,s

~

0 '

, sd ' Tg,;f.7 .ll'/ L'$M,,

' TA,p lc 'RAn? 3; *

- . . .y;;% - ,..

' ' ~C'l ..; .c : _'y 1 '. '

ll

i .~'

f'p l. ;7 . _

y . ;.e g -

3.,' .

( ..

!. Figure 2-19. Another SEM photograph of an overload fracture j -~ j region on a sample removed from EDG #101 and fractured.

-g 200X l

h

9 .n. r b j,f

'd f.  :

,e '*??'

b!? ~

s ,

.f .

V 2 , $ ,' ~.9 '

A L . M.2 k

Figure 2-20. A higher magnification SEM photograph of the

,li center of the area shown in Figure 2-19. flote the mixture 1

, of cleavage and dimpled rupture seen here; this was not on

'!l the pre-existing fatigue crack areas. 503X 2-18 Fa AA-84 14

[j

. *ks j ~

s .. y d

d -: d' ; .

, m-i s

-I I

[ 3- '

'n

d fd l'. Figuie 2-21.

center of Figure 2-20.

Still higher magnification SEM photograpn of the Note the pattern of fractured J pearl i te. 2KX-1

!u .

e l3 +c'y .

rAn p*- ~ v

}  !.@g .g)g:/hk#g .

t

.0;:.. 1 fi' .-

-0 '

. '...<c.

3.. .. .

1 .

' ;1 c

4 ,

1

... ..rh  : ,

a '

'f '1  ;

c . s. . d.yl-

.?.. .

', Is fI, '-}- *p g

, ,2 '"

. ,4 ' t - .. 'f.t'y

's -

?, f ] _I Y, . 9;' q.. ,

Y 4

,l .'? 5..f.'l ;j _g 6., . .'!. ..

3 f Figure 2-22. Inis is a metallographic photograph of a polish -

4

'" .ed and etched. specimen taken from EDG d102. The dark areas are graphite nodules surrounded by white ferrite. About 80% of the matrix is pearlite. 185X

. 2- l ^/ Fa AA-84 14.

.g$

a

4 1

3.0 EXPERIMENTAL STRESS ANALYSIS The - preceding section described cracking observed in the modified AF

]

piston skirts. Section 4 will provide the results of fiaite element stress analysis. of the AF skirt, along with corresponding results for the more recently designed AE' piston skirt. The AE design has been supplied by TDI as a suitable replacement for the AF skirt. This 'section presents the results of experimental measure,nents of strain under static load in the AE skirt. These measurements-provide the means of assessing the suitability of the AE skirt as well as'a benchmark for the finite element calculations.

3.1 Comparative Description of Piston Skirts The AF piston skirt and crown were described in Section 2. A comparative description of the AE and AF skirts is provided in this section. Rasically, the two skirt designs are similar, and only the major differences need be discussed here.

The major differences between the AF and the AE designs are concentrated

, in the region of the bosses through which studs extend to attach the crown to the skirt. The stud attachment bosses were considerably enlarged in the AE oesigni Figure 3-1 provides direct comparison of the two designs in this

, region. The thickness _ of the n.aterial around the stud hole for the AE design was substantially increased (from 1.44 to 2.62 inches), and the extent of the thickened area around 'the stud hole was increased (bottom portion of Figure 3-1). In addition to this modification, the following changes were made.

- Thickening of the walls of the cavity tnat extends from the top of the wrist pin boss to the top of the skirt. This thickening is shown in the bottom portion of Figure 3-1.

- Thickening -and filling in of msterial around the wrist pin hol e.

- Thickening and . tapering of the circumferential rib that runs between the wrist pin bosses.

l-3-1

i- .

The complete piston consists of a skirt (either AE or AF) and a crown, as shown in Figure 2-1. When the crown is mounted on the skirt, contact occurs only over a ring located just .inside the stud bolt circle. The outer edge of the crown is manufactured with a clearance of about 0.008 inch to ccmpensate for thermal expansion of the crown. Hence, there is no contact between the crown and sk!rt at the outer rim at romn temperature when no pressure is

applied. Figure 2 shows the crown-skirt mounting. The pressure at which the gap closes and the proportion of the load carried by the two load rings is not well known, and is therefore of interest in the experimental work.. The proportion of loeding on each of the rings is useful in selection of boundary 1 conditions.for the finite element calculations to be reported in Section 4.0.

3.2 Test Set-Up The loading which produces the largest stress in the piston skirt is the firing pressure. The pressure' loading is much larger than that imposed by inertia loading or friction, and thermal stresses will also play a minor role The major

-- especially in their contribution to the cyclic stress history.

portion of _ cyclic stress levels in the piston skirt can therefore be simulated by pressurizing the top of a piston crown and reacting the resulting force through the connecting rod.

Stress coat and strain gage tests were performed on an AE piston skirt which was subjected to hydraulic pressures as high as 2000 psig. This is well I above peak -firing pressures - reported for the R4 or RV4 series engines. An L

. actual cylinder liner was used in the test with two opposing pistons placed crown-to-crown in the liner.- The region between the crowns was pressurized with a hydraulic pump, and the pressure load on the instrumented skirt was reacted against a support plate with a short piece of connecting rod. The connecting rod was in a vertical position, thereby simulating the top center position of! the piston. An 0-ring was placed in a nachined groove in the piston crown in order to seal the pressurized oil, and the hydraulic pressure was read by a pressure gauge. The leakage past the 0-ring was very small so Lthat constant pressure on the crown was ea;ily maintained. The instrumen-tation employed in the strain gage test is described in Section 3.4.1.

3-2

3.3 Stress Coat Test In order to identify the arecs of highert stress with confidence and accuracy, a brittle lacquer, or stress coat, test was performed to identify precisely the highly stressed regions of the skirt. These results were used to provide guidance in the placement of strain gages for subsequent quanti-tativa evaluation of skirt strains.

A stress coat test was performed on an AE piston skirt with attached

crown. The gap between the crown and skirt was measured with a feeler gage to .

be uniform and between -0.007 and 0.008 inch. The region of the four crown stud mounting bosses was sprayed with Tens-Lac. Calibration blocks were also prepared in order to provide estimates of the threshold strain for cracking of the brittle iacquer.

-The strains in the piston skirt are primarily compressive, whereas cracking of brittle lacquer occurs when subjected to tensile strain. In order to.use the brittle lacquer 1or evaluation of compressive strains, the pressure was applied -to a given level, held for approximately 15 minutes, and then rapidly released. ,The hold time allowed the brittle lacquer to creep so that it would _be placed,in tension when the pressure was rapidly released. Tests on .the calibration blocks (nat were tested ander the same conditions (compression, hold, rapid release) showed a crackisig threshold of 1500 to 2000 pin /in.

A pre'ssure of'_500 psig was fi rst applied, held, and then rapidly

~ released. A careful inspection revealed no cracking of the lacquer. The same procedure war followed with a pressur e of 1000 psig, and again no cracking was observed. The pressure was then increased to 2000 psig, and localized crack-

- ing was~ observed. in the stud boss region. Figure 3-3 presents a photograph of the. cracked region, which was observed on three of the stud bosses. A roughly circular region of about 1/2 inch diameter was observed on the three stud bosses that. exhibited cracking of the _ brittle lacquer. This served to closely define the- regl'o n with largest strains, which was then concentrated upon in

-the placement of strain gages for obtaining quantitative strain levels.

3-3

3. 4 ' Strain Gage Test' The piston skirt used in the stress coat test was instrumented with foil resistance strain Jages. Rectangular rosettes were used in all cases with one arm of the rosette oriented in the dirxtion along the axis of the piston (E-direction). . A total of 15 rosettes were mounted, with Figure 3-4 summariz-ing the locations. Two rosettas were pla;ed on each stud mounting boss in the region where the - brittle lacquer was observed to crack (for a total of eight

~

rosettes). .All gages in these highly stressed regions had a gage length of

~

0.030 inch. As indicated .in Figure 3-4, two rosettes (A and J) were mounted -

-in the stud boss region away from the peak stress region. The " column gages"

- provide an estimate of nominal ' stresses in the wrist pin cavity wall. The K and M ' rosettes were placed on the longitudinal ribs between the wrist pin

bosses.- Rosettes P, R, and N were placed in the wrist pin cavity and the skirt ~ wall 90* from the wrist pin in order to provide information on. the pressure . required to: close the- gap between the outer load ring and the crown, as well as to estimate how the load was divided between the two loadino rings.

Some of- the gages located . in regions of lower strain gradients had -gage

' lengths'of 0.060 inch.

Tests were conducted- by applying the pressure in steps. At each step, the pressure was held constant while the strain gage readings were automat-ically recorded by the instrumentation system. Data was mcorded for both increasing pressure steps and decreasing pressure steps. Several cycles of

-loading were recorded.

Two separate- test ' series were conducted; one with a conventional crown and tone with a crown that -was- machined to widen the gap so that it would not close under an applied pressure of 2000 psig. Thus, the loading was known to be'only on the inner ring in the test with the modified crown.

, 3.4.'11 Instrumentation The -instrumentation. required for the piston . test included strain gages, data acquisit. ion equipment, and an accurate way of expeditiously recording the data'. from ' up' to r 60 - channel s. The instrumentation chosen for this test and a brief explanation of its capabilities are discussed in this section.

3-4

x.

The . strain gages chosen were of two sizes. The smaller gages (0.030 inch

gage length) were used in the areas of suspected high strain on the bosses of the piston skirt. The 0.030 inch gage length was chosen to minimize the area ever which the ' strain was averaged. The gages were arranged in 45* stacked rosettes- and were encapsdated for protection. In other areas of the piston

'where lower strain gradients were expected, a larger 0.060 inch gage length strain igage was - used. These were also arranged in 45* stacked rectangular rosettes and encapsulated.

The, equipment used to measure and record the strain readings was a Micro-1easurements Vishay System 4000 which included an HP 9825 Computer for the

" executive ~ unit," a .Vishay 4220 Controller, and three Vishay 4270 Strain Gage Scanners. The System 4000 is a computer controlled daca acquisition system wnich can ~ acquire. the - raw data as -well as use the computer to provide appro-

'priate ' data' reduction.

-The system software is' loaded in the executive unit and essential data (such c . gage factor and which channels have gages are input by tne user and recorded .on magnetic : tape in the executive unit. The executive unit slso contains a ! built-in- printer to document the input of parameters and the

-recorded data.

The 4220 : Controller, ' including its interface with the encutive unit,

.provides the Llogic :and timing conversion between the executive unit and the

~

scanners. The- 4220 Controller digitizes the analog signal from the scanners

.and < routes -it to the executive unit for recording. The controller also pro-vides the logic for the scanners.

The! 4270 -Strain Gage ~. Scanners contain 20' channels per scanner. The scanner ican scan at- a rate of 30 channels per second based on a 60 Hz-' power

.line. The scanner also provides internal shunt calibration to check each channel.- At each . pressure that strain gage data was - gathered, the strain +

reading from. each f or the - 45 gages was recorded upon command on magnetic cassette 1 tape. The magnetic tape data was 1 also printed on paper tape for permanent storage and display of the test results.

.p 3-5

3.4.2 Test Results The strain gages were zeroed prior to installing the crown on the skirt.

~

The TDI-specified crown installation procedures were followed, and strain gage readings then taken. . Only very 3 mall strains were recorded at this point.

' Pressure was then applied in steps, as mentioneo above, and strain gage read-ings recorded.

,The maximum stresses in the piston skirt under peak firing pressure are of primary interest. This pressure is approximately 1670 psig. Some of the circumferential gages in the rosettes in the stud boss region did not work; however, enough were operable to provide accurate and representative results.

Table 3-1 summarizes the strain readings for the stud boss rosettes that. were complete. The results are for 1600 psig on the first cycle of loading ~ for a conventional crown. The principal strains were calculated from the rosett.e readings using the conventional equations for a rectangular rosette [3-1].

The following elastic constants were used to calculate the stresses from the strains.

E

• 23.6 x 106 psi v = 0.3 These values are reported in Reference 3-2, -and the valut of E is in accord-ance with the value suggested by TDI.

The results presented in Table 3-1 show that the principal stresses and strains are very- closely aligned with the z and e (axial and circumferential) directions in the skirt, because cz and c eare nearly equal to cyg and ci, respectively. Additionally, it is seen that the stresses are nearly uniaxial and the value of oggg is well below the yield strength of the material.

The stress ' levels for the ' rosettes which had inoperable circumferential gages can be . estimated by assumingecz/c = 0.17, which is the average value .

for the rosettes included in Table 3-1. This allows the results presented in Table ,3-2 to be obtained. A comparison of the results from the two tables show that-;the stress values 'are comparable. The results do not show a large 3-6

. Table 3-1 STRAIN READNGS AND CALCULMED STRESSES FOR INE COMPLETE STUD BOSS ROSETTES AT 1600 PSIG WITH A CONVENTIONAL CROWN C D E F H c

z -1740 -1668 -1722 -1605 -1583 ce 270 366 304 270 246 c45 -732 -787 -815 -602. -632 cg 270 375 310 272 247 cgg -1740 -1677 -1728 -1607 -1584 ogg, ksi -6.5 -3.3 -5.4 -5.5 -5.9 oggg, ksi -43.0 -40.6 -42.4 -39.6 -39.24 I.

L I

l 3-7 1

4

variation' from rosette to rosette nor from stud boss to stud boss. Therefore, the Lyalues. measured :should be - close to the largest values present in the skirt, and 'are the' values to be compared with the peak stresses calculated by finite elements'-(once - adjustments are made for differences in pressure and .

- load splits between inner and outer rings).

The absolute value of aggy (~o z ) for Rosette C is plotted in Figure 3-5 for the ;three' pressurization cycles applied and for increasing pressure be-tween steps (up) and decreasing pressure (down). The upper part of i;he figure C  ; presents results for a' conventional crown, and the lower part of the figure is for the modified crown .(in which gap closure does not occur). These figures show the degree of reproducibility and hysteresis in the test results.

1 The . solid -line in the upper part 'of Figure 3-5 is through the data for Run 1~up. A bilinear -variation of stress with pressure is observed, with a slope change;at just below 1000 psig. This slope change is believed to be due to.. closure of.'the' crown-skirt gap at -the outer loading ring. Evidence of this is shown .in the: lower portion of Figure 3-5, which presents corresponding

- results obtained by use of the_ modified crown. In this case a linear oggi - p

- variati.on -is observed, with a slope equal to that for a conventional crown at-pre'ssures below- 1000 psig. - The _12% difference at the _ peak firing pressure between - the modified ' and . conventional crown results (in conjunction with finite ' element'~ calculations ~ that reveal" that the peak stress -in the -stud boss regionais controlled primarily. by loading on. the . inner ring, see Section 4.2) indicate' that 88%:of' the pressure load is carried on the inner ring at the

- peak. firing pressure. .This result is for room temperature; a greater percent-

- age of load;on the outer ring is expected at: operating temperature because of thennal . distortica of - the crown. Such . distortion would reduce the peak stresses.

- Further evidence of a gap closure pressure of about 1000 psig (at room

. temperature) is provided by considering the strain bta from the rosettes in

! - the wrist pin, cavity'_ (P at inner ring, 0 at cuter ring) and at the top of the

- skirt 00* from the -wrist pin -(N). Figure. 3-6 presents the axial strain (cz )

measured in Lthe inner part of the wrist pin cavity (P) with a conventional crown.- Once again, 'a bilinear -relation with pressure is observed, with the 34 bz ea y a- q wway --2-H----w-a-- g-,r-- TvW'-$ v+W" 'T F' T-Tt*~fT ^-FW=Mw'77'-'97-T*-+-78" ?-4' "^~T*--r*= *==$f WF*+

. o .

Table 3-2 STRAIN READINGS AND CALCULATED STRESSES FOR THE STUD BOSS ROSETTES WITH MISSING c FOR A CONVENTIONAL CROWN AT 1600 P G B G I cz -1930 -1768 -1471 cg = 0,17 c z 328 301 250

. c45 -1145 -798 -1014 ci 379 303 340 cII -1981 -1770 -1561 ogg, ksi -5.6 -%9 -3.3 ogyy, ksi -48.4 -43.6 -37.8 l

L l

.s l

e 3-9

slope change occurring at 1000 psig. In this instance, the slope change is not large.

Figure 3-7 shows results for axial strain with conventional and modified crowns for the rosette at the outer edge of the wrist pin cavity (R). 0nce again, a bilinear variation with pressure is observed for the complete crown, with the slope at -low pressure being equal to that for a modified crown. The slope ' change again occurs at 1000 psig, and is particularly marked in this instance. Apparently, once the outer gap is closed, relatively large loads-a,re transmitted through the outer ring over the wrist pin.

Figure 3-8 presents results for the rosette 90 from the wrist pin (N).

'The bilinear relation is again observed, with a slope change at 1000 psig --

.which - is the - same behavior observed on all other rosettes. A comparison of

.the rosettes under the outer ring (N and R) shows that the slope change is in opposite directions for the 90 position versus the wrist pin. The marked increase in the slope' for the wrist pin results (Figure 3-7), in conjunction with the less marked decrease at' 90 , indicates that, once the gap closes,

- most of the load on the outer ring is reacted over the wrist pin, with a smaller portion of the load being transmitted in the portion of the outer ring away from the wrist pin. Such behavior is ' explainable if each of the contact rings on - the crown is considered to remain planar when the skirt / crown is pressurized., The -vertical displacement (u z) in the contact rings would be uniform, and " dishing" distortion of the crown is responsible for gap . closure and varying loa'd split- between the inner and outer loading rings. A uniform

. vertical displacement over the outer ring would result in most of the outer ring load being reacted over the relatively very rigid wrist pin boss, with much less load overLregions away from the wrist pin boss. This would explain the. difference in direction of the slope change for the wrist pin (R) and 90 rosette (N).

Thus, a uniform displacement over the loading rings appears to be a reasonable boundary condition for the stress analysis of the piston skirt.

This observation, along with informa' tion on the load split between rings and

' absolute magnitude of stresses, provides guidance in appropriate bounda ry conditions and load conditions for the finite element analysis.

The experi-3-10 L.1__

-mental .results also provide inputs to the assessment of the suitability of the AE skirt for the individual applications.

References 3-1 R.C. Dove and P.H. Adams, Experimental Stress Analysis and Motion Measurement, Charles E. Merrill Books, Inc. , Columbus, Ohio,1964.

3-2 Iron Castings Handbook, edited by C.F. Walton and T.J. Opar, Iron Cast-ing Society, Inc. ,1981.

l I

r

')

3-11

T-'

~

F -~

/ ,A 1,

/

AE AF

/

A e AE AF AF I

Cavity AE i l

I l I

i Figure-3-1. Comparison of AE and AF piston skirts in the region

. ' of the stud attachn ant bosses.

. l l

4 FaAA-84-2-14 4

-- ,, - ,m, ,---.

s V

e E

7 m o

o C .E N  ;

c o

! l k % -

N

- s !Uj

% u w'sj Oy *:

8*

r.a

& $"3 ss eeu o

~

. . . c

.e 25 w wc Fa A A-84 14

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

'NW;;;-7'"cey

~

! ~+-

j7f. r .. . , ,

1

.f

<.i. t'.vV[% g4;./ d{'Q,.g;/ s'i.+.:* ' r f.1,8l %. Q' [ ./.7. 1,' 7 ' .' i ...w.,. N 97.5-g-< ty. , .' 7Mf)ts i i .3 : QP*l.h i n t.1 5 L ^f. W'f.'.. ' .g "94.

-n-

.: ~ . .n.:yvk .: .. ;q  %

i . ,': .. .

.- J. . V

,. ;,.y ..r.. w. ,,,, . , - }_' :4Q.

p ,i .. .5.yj J  ;.R:;,. - -.. . 't, -

j

. v ~n:[s .

• . "y y. ..j,.'.. , h...c.., .

, L. .. ,.- i .

). [. s..

.. s c

< .;v .

v w - [. s .- :e.: e Q. -

7

.%.:.)- 'Yy .Y, ,

g ".: D..;. Q f ? .'?::q'l.lj(:2' . , ' +

,lllh.' l:},].)l_' .L-;d:', ,?

. > . .j. ' 7 . ' . h, . . ..;su$- t . 4 _s

_ J. ,..: r. ; ' ?v - s y, q.;

l z.t:.

',. m f. &. . . ,'s. 1 .

...,1 . . . . . . , . ,, c. , . .

i

' . *. , y. . 1,7.. ..a . ,. ,.: u.;.-F :

f', , .:.f.. ",; . . ; , . 2, . ,- s . y*

g

.t.. .

N

• -( .: k.': df..:m.. .} ._.f.,. .,l: ... u

>...a.,i> a.

. [-

e

.A

.[.f.,..;;&y-f .

. w .

l ,.

u - -

Yh.

l ?plb)Q f 'YY ' '.".d? I . k 0 W _ Y,.. .;l; .

3 '

" . &.%.  ?' $.Y.hb' s' ' ' c.i ' .N +

Q ; ' " ' "'." .

• v;

a. G V y. : :

. , . m:.' f.,. . :: .; ,-f S.

I .

<: ?g,. ,v . spo..~. :,. . - f'[ .-

3. .%)n

.x . . ' ,, a ..g ffen y

. m . ... * .

g 3,. ... -;* y .,4 -

,s v.

.c

" ee-

.45...

-....1

'..y.*r - .'" - 9/

._w,4. w'

< , . .r 5

.ej, .

+ , s \$.Pd.#n$4

..,::.  ; gig 9. . >r, .w4l -;m, ' , . . . .

r.w. ' . ' f.: , . c' , s-

.o?s ,.,

Te,j:.c'.h+ T@r '4 RN..c b..u _9 ..

,..>w.

q.. -

.  ; wJ,:i f f, .y t.  ;.y:1'. a .-

.v' . :"..'. ec.

a

r. e -

&m .y ... ..+. rn.p ...,.~ ., x . c...a

.w..,....,....y_.  %. ,...np .

4 -

. . :4. ,;.

.m .5 3. :,%. v.

~ .

. , . . *v.

[.y$ Ml~$c %.'4..1p3.y gf. " O c. I

• l $4 e..Y:3

.l - .

v..

• ' ~ '

j s & <,;;a ";,c*., ggg 1. , c.f.

• r c.. ';j&;.t;y,j;y,; , ..s} _..

nf.wl:{gr .,f, i &.;; W~ 4: . ; rgk. . Q t .. -

.+ ~-

yy:

n. w - - ,

.n . a . ...c..

s g^ .r g ; .  ; . - a,* 1. .z . f' q .p.s . ..

._f. y*,;;,;  :: .; Q w -::y.

y, n .- .g 7 ..,

. m,y. , ; , .~w.

m ,m;n'y .,~ i: p@;. s. ,.',';,<  ; ,%s,3 _,

fg.$. w~ ~hgky.;a:. g. g, .q.

ww

,; ; .v"k p %.p! ,- y,s

. , 2v;o . .y av ' ry.; - > -

- - ' y, 6, :..p p

-- l l

l Figure 3-3. Photograph of cracked lacquer.

i 1

i I

i l

I I

I l

I FaAA-84-2-14 l 1

-_-_-....s.,-..

y. -

r Pair H.I Pair F.G

\ -

% K

\

61 (:l,:

..::0,k,hk::,.. , ..;: h.h .N'

I ,h;h.:5

{5 ..Y:bh. ..

. {l: . -) l]:h!l

_'Of iG i.Q 'N

!!n aC R yp( O *+* L g4gy N. .:: -

% sm M

l w?!Nii 4

- j A \ A E

J j T g Pair D,E Pair B,C 4 P

.k R Figure 3-4. Location of strain gage rosettes on instrumented AE skirt.

S 60 AE PISTON WITH COtTiENTIONAL CROWN - ROSETTE C w

m . + RUN 1 UP i / .

$ 50 -

+ RUN 1 OOWN

,,f8 a RUN 2 UP .-

_j . 40 - a RUN 2 OOWN '

{ . O RUN 3 UP n ,

O RUN 3 OO',VN U

z 30 g

y -

E 20 -

H =

W -

a u) =

E

@ 10 a .

. O g O 500 1000 1500 2000 PRESSURE Epsi3 0

4E PISTON WITH MODIFIED CROWN - ROSETTE C

$ 60 g -

+ RUN 1 UP

$ 50 - + RUN I DOWN

% . a RUN 2 UP 40 - C RUN 2 OOWN

< O RUN 3 UP

a. -

w u

z 30 E

' 20 -

s '

@ 10 J

.O ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' '

g: 'O 500 1000 1500' 2000 PRESSURE [ psi]

l

~

Figure 3-5. om as a function of pressure for stud boss rosette C with conventional and modified crown.

FaA A-64-2-14

A DN ' dim COWENTIONAL CRWN a n

600 /

f C -

+ RUN 1-UP f

/ .

E 500 - + RUN 1 DOWN f.'

$ . a RUN 2 UP . /"

/ '

RUN 2 OOWN ,/

[

u_

400 O RUN 3 UP ,

E O._RUN 3 DOWN +

300 -

.E 200 -

w-g- -

. y 100 -

U -

J 0

O 500~ 1000 1500 2000 PRESSURE.Lpsi]

Figure 3-6. Axial str.ain as a function of pressure for strain gage rosette located at-inner ~ portion of wrist pin cavity.

J' e

FaAA-r?4-2-14

200

'AE PISTON WITH CONVENTIONAL CROWN 2 + RUN 1-UP O D '

+ RUN-1 DOWN y' 150 ,

. a~RUN 2 UP o_

a RUN 2 DOWN b .' O RUN 3 UP

3 100 -

0 RUN 3 00WN t z .- .

_: .- - /

50 '

E -

z

. i i 0 . . . . . . . . . . i 0 500 1000 1500 2000 PRESSURE Epsi]

200

.AE PISTON WITH MODIFIED CROWN 7 . + RUN 1 UP

o. -

+ RUN l'DOWN h '150 o RUN 2 UP.

m

C a RUN 2 00WN h-  ; .O'RUN 3 UP'

!. 3 100 -

x .

_J

[

.<. . /

I

, U -

- . i I 0

r.

'O 500 1000 1500 2000 PRESSURE :[ psi]

Figure 3-7. LA xial. strain as -a function of pressure for strain gage rosette located at outer portion of wrist pin cavity.

Fa A A-84-2-14 j, , .~ -- , ., _ - - , , , - .

6 700 -

r,

.AE~ PISTON WITH CONVENTIONAL CROWN

/

/

C 600 -

+ RUN 1 UP /

0 -

+ RUN 1 DOWN

,/

j- 300 -

c RUN 2 UP

/ _

.a a-RUN 2 OOWN ,-- g

.b 400 -

O RUN 3-UP / @

O RUN 3 OOWN '

E u +

300 -

z . +

200 n d -

+

E +

-$ 100 -

U O

O 500 1000 1500 2000 PRESSURE Cpsi] .

.AE PISTON WITH MODIFIED CROWN "C 600 -

t RUN 1 UP 0- + RUN 1 DCWN j' 500 -

a RUN 2 UP o C RUM 2 DOWN 6

'b 400 -

O RUN 3 UP E

s 300 -

z -

a d- 200 -

E

$ 100-U:

. . . . . i . . . . i . . . . i O 'O 500. 1000 1500 2000 PRESSURE [ psi]

l Figure 3-8. Axial strain as a function of pressure for strain gage rosette located near outer. load ring at 90 from wrist pin cavity.

~

FaA A-84-2-14 k

4.0 - FINITE ELEMENT STRESS ANALYSIS The results of the finite element stress analysis for both the AE and AF piston skirts are presented in this section along with descriptioqs of models

. employed in the analysis.

4,1 Load Considerations Three loads were assumed to be acting on' the piston; gas pressure, recip-rocating inertia, and friction. In addition, since the piston is a two-piece design, . initial internal load was associated with the bolt preload. As part of the analysis of the crackshaft of these same engines [4-1], a table of gas pressure, inertia, . and friction was developed, indicating their values for every ten degrees of rotation of the crankshaft. This covered rotation from O' to 720* and encompassed the entire four-cycle combustion process. The combined inading was found to be highest at top dead center during the combus-tion stroke. At this point, only pressure and inertia are acting on the-piston since the velocity (and therefore friction) is essentially zero.

The gas pressure load acts on the surface of the piston crown. As dis-

. cussed in Section 3.4 and shown in Figure 3-2, the crown contetts the piston skirt at the inner ring just inside the bolt circle. The outer edge of the crown is manufactured with a clearance of about 0.008 inch between the crown and the . skirt. This gap can close due to pressure and/or thermal distortion of the crown. The strain gage results presented in Section 3.4.2 suggest that the gap closes at about 1000 psig when the piston is at room temperature, and

'that about- 88% of the peak load is carried on the inner ring. Under steady operating conditions, the gap closure pressure could be lower and the portion i

of load carried by the outer ring higher.

These considerations established the basis for the boundary conditions on the upper _ portion of the skirt. Two extreme conditions' for the circumferen-tial variation of' the crown-skirt interaction were considered: (1) an inter-

facial pressure on the contact rings .that does not vary with angular position, j this is referred to as uniform load and corresponds to a very compliant crown or skirt; and (2) a vertical displacement on the skirt at the loading rings 4-1 L

f e (that does not vary ' with . angular position, this is referred to as uniform displacement and corresponds to a rigid crown. The experimental results pre-sented -in Section 3.4.2 indicate that the uniform displacement condition is closer to the actual situation.

s Boundary conditions for loading on the inner and outer rings were also considered. Loading on the inner ring only (single ring loading) and a 50-50 load split (double ring loading) between the inner and outer ring were con-sidered for the case of uniform loading. The stresses for varying degrees of load distribution were found through a linear interpolation of the values found in the two load cases. Figure 4-1 shows a graphic plot of the nodes on the. top of the skii t with the arrows indicating the load distribution along the inner ring.

The magnitude of the pressure load, determined during the crankshaft analysis, was 381,300_ pounds distributed over' the surface of the crown. This load is transmitted to the skirt through the inner and outer rings as pre-

- viously discussed. The pressure load is somewhat offset by the inertia load, which is determined by the acceleration of the piston multiplied by the mass.

In ~ the analysis of the. piston skirt, the mass acting on the skirt is essen-

- tially that of the crown, about 130 pounds. Multiplying this mass by the acceleration found at top dead center (26.1 x 103 in/sec2) we obtain a force of-9460 pounds. The combined loading .of 371,840 pounds was, therefore, used for.the pressure load cases.

At the end of the exhaust stroke during engine operat. ion, the force acting on the piston ski rt is reduced to inertia loading only, due to the absence of gas pressure. The value of the inertia load is the same as at top center of the combustion stroke,-9460 pounds. Since the inertia load acts on the same nodes as the pressure load (but is opposite in di rection ), the effects of the inertia load only can be evaluated without additional computer runs. By ' reversing the signs on the stresses and multiplying by the ratio of the inertia load to the combined pressure load, the' stresses due to inertia l loading are obtained. The ratio is approximately 0.025, or that is, the inertia load is ?.5% of the combined pressure loading.

n 4-2

_- (

7 The stresses due to bolt preload were evaluated on one of the early models and found to have virtually no effect on the stresses in the area of the radius' between the boss and the vertical wall. The bolt preload was,

.therefore, omitted from further consideration.

4.2' Piston Skirt Analysis Stresses and displacements under gas pressure and inertia loads vere calculated for both the AF and AE piston skirts using the ANSYS finite element computer program. Two models were developed for each piston design: (1) a -

full .(global) model of the skirt, and (2) a refined local model of the crown-bolt boss in the region of .the highe.st stress (as identified from the global

- model 'results). In . the following subsections, the results of the finite element analyses -- including maximum stresses, their locations and directions

~

-- are presented with particular attention to the observed crack locations in the AF piston bosses.

Dimensi7ns for the - AF piston models were obtained from measurements of actual castings. . Dimensions for the AE piston models were obtained from available engineering drawings and by measuring actual castings. The elastic constants'given in Section.3.4.2 were used in the analyses.

Each of the " global" finite element skirt models was ger.erated with ANSYS

eight-node solid elements incorporating trilinear displacement functions.

Each node may have three components of displacement.

.Results from the global model' analyses indicated highly localized stress gradients at the intersection of the vertical wall and the stud attachment boss ~in both-designs. For this rcason, substructured models of this region in l 'both designs were developed with a much greater degree of mesh refinement than was possible in the global models. The substructured models were loaded using a technique known as " displacement recovery". In displacement recovery, the boundaries of the -local model, as defined by boundary nodes, describe surfaces in the global model for which nodal displacements can be recovered. By impos-ing the displacements on the local model, the local model is made to respond

.as if- it were an integral part of the global model. This technique provides 4-3 L

p -

results in regions .of high stress gradients and small geometric details with-out incurring prohibitive computer memory allocations associated with high degrees of refinement in large global-models.

Comparisons of the ' _AE and AF piston skirt peak stresses are given in Tables 4-1 and 4-2 which show the maximum third principal stress and the Von Mises' effective stress. Details of these values are discussed in subsequent subsections.

. The gradient of these stresses below the surfaca as calculated by the local models is shown in Figure 4-2. This is a graphical representation of the . normalized stress (o/%x) versus distance below the surface. These gradients are necessary for the fracture mechanics calculations of crack growth, as will be discussed further in Section 5.

The stress _ gradients are determined by starting at the surface where the highest stress _ nodes are- located. The stress at the nodes in the next layer beneath the surface are scanned to obtain the node with the highest value.

Each subsequent layer is also scanned fir. ally forming a string of nodes along which the highest stress level at any given depth can .be found. On both the AE and AF pi'ston skirt local. models, this string was a smooth line following a logical trail back to the surface.

4.2 1 AE Piston Skirt Analysis Global Model . The global- model of the AE piston skirt consisted of 1895 solid elements and 3883 nodes. Figure 4-3 shows the AE global model and the region selected for further analysis using a refined local model. As was also done for the AF piston, a quarter-section was modeled and symmetric displace-ment boundary : conditions were imposed on the x = 0 and y = 0 planes of sym-metry.

' Local 'Model . The local model of the AE skirt stud attachment boss detail consisted - of 984 solid elements and 1358 nodes. Figure 4-4 shows the local model with shaded planes representing those on which displacements from the global analysis were imposed.

4-4

, - ([:' , I <?

Table 4-1 -

PISTON SKIRT PEAK STRESS COMPARISON (Single, Ring Uniform 01splacment) 6 AE'~ AF

~ ....................................................

Stress Type Giobal_. . Local Global Local

, (ksi): (ksi) (ksi) (ksi) n.

Third Principal Stress 49.96 81.18 88.4 118.7 '

(0111)

Von Mises' Effective Stress 37.34 72.83 80.9 98.2 (CE)

O a

w i

e h

,..t.'

n 8

e

.j ,

L e

4-5

'N  ; .

(- ,

r ~ , , , ,a n ,- --. ,,r- - ..,-..n< . . , -- ., .n. -n

Table 4_2 PF. TON SKIRT PEAK STRESS COMPARISON FOR LOCAL MODELS WITH UNIFORM LOAD OR UNIFORM

_ DISPLACEMENT BOUNDARY CONDITIONS Values of oIII(max) ksi AE AF Single Double Single Double Ring -

Ring Ring Ring Uniform Displacement 81.2 _- 118.7 --

Uniform Load 164.3 85.5 304.0 154.0 Values of cE(max) ksi AE AF Single Double Single Double Ring Ring Ring Ring Uniform D'isplacement 72.8 -- 98.2 --

Uniform Load 156.4 82.5 296.0 136.0 4-6

g.

• t

to -

Stress Results.- The global AE piston _ skirt model showed peak values of the-third principal stress in the area of the intersection of the vertical wall and the stud attachment boss. The peak - value of the third principal stress in this area was 49.9 ksi. Figure 4-5 shows a color schematic of the stress gradient at . the surface on a portion of the global model. This value occurred under .a uniform displacement on the inner ring of the top of the skirt that provided the desired load (single ring uniform displacement). The global 'model was also run using a uniform load along with inner ring (single ring uniform load)..

To . locate the high stress and to determine the stress gradient undar the surface, 6' local model was produced (shown in Figure 4-4) which more accur-a'tely models the geometry. Displacements from the global model results with

. single ring. uniform displacement were used as boundary conditions for the o

. local model . The principal stress -peak is in the radius between the vertical wall and the stud attachment boss.- The stress gradient at the surface is shown in ' Figure 4-6. The peak principal stress at the surface in that region was.found to be.81.2 ksi.

IThis local; model was refined to include 50". more elements and run under

..the osame boundary conditions. The'_ peak stresses in this refined model were only slightly higher than previously found, indicating that covergence had been achieved with the local model. -

-4.2.2 AF Piston Skirt Analysis Global Model . ' The global model of the AF - piston skirt consisted of 1141 solid elements and :3472 nodes. Figure '4-7 shows the AF global model and the region .. selected for substructuring.- The global model represents a quarter-symmetric section of the piston skirt. Symmetric displacement boundary con-

' ditions were -imposed on the two planes of symmetry: x = 0 and y = 0. The

z-axis coincides with the centerline of the piston skirt.

Local Model . The local model of the AF skirt . stud attachment boss detail cons'isted of 720 solid elements and 1105 nodes. Figure 4-8 shows the AF local

'model; the . shaded planes 'are those- on which displacements from the AF global model analysis- vere imposed.

4-7

.1

, .s o; -

Stress Results. The global AF piston skirt madel indicated a maximum th' erd principal stress in the area of the bosses and had a peak value of 88.4 ksi for uniform displacement loading. This peak stress was at the inter-section of the vertical wall and the stud attachment boss. Figure 4-9 shows a color schematic of the surface stress gradients. This value occurred under a uniform displacement of the inner ring only (single ring uniform displace-ment). Computer runs were also made using single ring uniform load. ,

In order to understand better the location of the high stresses and the stress gradie;it below the surf ace, a local model was constructed (shown in ,

Figure 4-8) of the area of high stress found in the global model. This local

. model more accurately duplicated the geometry. Displaceme'nts from the global model results' with single ring uniform displacement were used as boundary conditinns for the local model. The principal stress peak in the local model was found at the radius between the vertical wall and the stud attachment boss.

and can be seen -in Figure 4-10 which is a color schematic of the stress grad-ients on the surface. The peak principal stress was found to be 118.7 ksi in that region at the surface. The local model was refined by adding 507. more elements and rerun under the same boundary conditions. The peak stresses only increased slightly indicating that convergence had been achieved using the

- local model. -

References 4-1 " Emergency Diesel Generator Chankshaft Failure Investigation Shoreham

' Nuclear Power Station," Failure Analysis Associates' Report No. FaAA-83-10-2, Palo Alto, California, October 1983.

4-8

i i

l f

i 1

.~ s l

.i

~. . .. ..

F 7

j I

Q -t , ,

- { ';;: . . .

i b, j

t. .

nlj 1 t ;, . '

t 3

i

,. i .

I i

Figure 4-1. Color graphic of nodes on top of skirt indicating load distribution along inner ring only.  ;

i 1

I O

I l [

i i

FaAA-84-2-14 h

d

)

t n

_ - - - _ - - - _ e m

e c

a 1

g ~p s

% i d

~ 6 g 0 i n

- r l - e a

c ' l g

l E

o

~

i

(

s n

A n

~ o r/ ' )

n i s

i r

4 ( a p

) E m s l C o a c

' c A o '

4. F t n

l 0 R e F U i d

A S a r

g W

O s L s

, E e

' r B t s

H s T

N P s

o E b D

s' t

l 2 o

' b

' 0 t r

i

- k s

n

% o t

_  % s

' i P

g N 2 4

e r

u

. - - - _ - _ - - g 0 i 0 8 6 4 2 0 F 1 0 0 0 0

. ^xaEbb mmmWFm ON]jgOz m${i7A

O t

N l

\/

_- /

N

/

N/ -

/ ,

/ /

! /

l

\

/

/

/ '

y

/

g3glon OI

/ ,\ gy toca\ 00b*\ ~

\

i

• e e -

' ' M ('*

p Nk/'[e

/

Giub; f O ,

/

/

/ /

)

~

fN' A/ 3 s '55 s

/ ll\'  %

f

, ' \

~

. pigure 4-3* g piston 9TobaT model-FaAA~M'*~ \

g E ~;. . ' g ,

1

,N *: .

N' !. t

u- .

.kk,_ N w

pe

' C C s' C N ajl xw N a' :s .

1-s

-Q -

, 4 .

- . xg . ,.. ,

f s i .

f -4 -

4

. l -j ' t ,

i > <

- ; .- u l

l'

, +

, _ ~,

i e

t l

i 1

4

l. l*

,:{

,  : j ,

j. ,i*

' ! .f .i .

l i.

.f, .

• g

.R '

,I l l 6 it 1

'i l l- f.

f.,s y 1 i

n, I

l t-a N , , ,

i

\ ,

4 i

'I .

i o l

.N ,

l.

. I f

.  ; I fi  !

f.I'. mlI I

[j { l

,i ^ ,

'f

h x. l l'!.. i I [l

.1 m

, ,1 ,

s g.

s [l- f jr '

l lY

_ _ .g h; [ /' ,

~

' ~ QFM l Ilil/

~NJ  %

ly/W'R P ~

i.

gi

• S-.

N$,._~k

'l 'Q

_N x: . .. .F X Nk N

I' ,

,N  % _,

m .~

. r =.

+

' w-'*

N5% '-:

q  ;- -

N- L s. . ; . 3 l , _

q

, x.  %

N- .%).

4 3

. Figure 4-4. LAE piston local model, Fa A A-84-2-14 f

d I

1 .

< . I 1

s P 4-i

.'4 .

s i Wrisi pin boss J . .

2 y k 6

j . . 7

-59384. TO -57153 4 ,

- , .-67153 Ter-53259, .

s 'N  ;

~ ,

-53459. TO -49365.

,-49345. TO -46471. .

Washer Ianding -

j spotf aced

~-

-45471. To -43%f7.  ;

-41577. TS -37643. .

. .,.,,x 8 < '

t

. ~ <; ,Y; - Q  %

.  ; -37643. 70..-33799. i

~~

• -33739 TO -2989W. N l 1 -21896A TO -26491. 1 ..

~

i Bolt hofe ' bore '".

a'- *4.. -

- 9..:.

' c:.. $? i .-26481. TO -22187. *.

l

- ' ' : t ..

,,_ l kf[j;f&i~~y[,*'N -22197. TO 'e18213.

l j ', .

~

K g - -

gj ' , . ? ;.: ' ;

• • i .j h2~,: -418913'. TO -14319. '

. . ,t- . . ,

4 ?%; g . ,

. .. :; i . .~.1 - -14319. TO -18425. ,

s.

} .

-6531.-IO

... . -4534.

~

i g#,

m AE PISTON 1/29/94 =

g ,$ . . .

g SUSCASE 1 ,  ;

' '7 9

Y g i cm -

.o 4 >

i Figure 4-5. AE piston global medel with stress contour of average element stresses.

~

,i , y <- .

< ~

n u . i v ,

. 0000 7; -10,000 .

4 ,

. Wrist pin 'bosa -

-!s*U,00 'C' -90444 "

..f .

e -

-45000 ':i.-90004, -

9 -

, *? , _ a e .c ., ~

-g;000 b

$$ -74000

-5:000 . T: '

- 4, .

4,. .-

-74000 To -eyegg e +_

_g '

• 4 . -eea00

': -5E700,3 4

3 -

=,

, -55000. TO -50000 H.

! s. . - .

l . '

-50008. T j -42000 g

1 Washer fandinig -

/ h l ,. ~ == ="s~ a F

,sg- 9 -<=cee I ro in**e-

~ -34000. TO -26400. K~ -

,4 < '

l Rli! 1. r L.. #. . . , ;- s <

. -26099. TO c19490. LF g- =

, ,. . . 4 ., 7 y 7 1a. .

' 3-

.? GA:, 2,y -18990. TO '

.- 189t( .

c - 'O  : $ ,g

~

" -teeee . To - -2ess.

g y -j _ - -. - f ,. ., .

,.7 :g;. ,

%y;j?

2000.

j .

, ,,- -209e. TO AE LOCAL ( S I N G L E '.P I N G v. LACEMENf, .

l'. Q . Bolt, holt, bOF%

Figure 4-6. AE piston local model with stress contour of average element stresses.

/'

(

\ '

\

t N

\ \

\ \'

\ .

'\

\. \\

\

\

(

\s'\'

s s , .

Re91 0" 0 \

u-tocal modet ' -

s , s

'\*

} .-

s

\ \ T)

,, ,M ,s, ,

e 4 gp piston global model-Fa Ah-84-2-14 g-

?

3 T \a\'\ \

-l,j. '

1

%\\\& .

.:,,f ,

_ f

\WN\.\; , j i -

\ --

1 /

/,

s' /

\

~~  !

,0,j,j lll

' ? ,'

l L!

Ip

/--  !

^ \9;jIfl:,Ilf:-(f;

.?

! w iQ ,

Eh j '

'A I'

~

W~ s % _s

,1 j l' f _,

7, i -3 lj s .

7~.[s,M s l\ j i j

-j j yM l .' ' ', '

, , ., t __

m.

i

.i d f. ',r .!

ll- 'k/! \/_ \\ '\, \ ,

(; .

-

p u/ f l

/ "l 4 .

1 -l f

\ },

. m . l7 l

/

\ \

8

( k \

1

\

[ ,

'\ _s .

t

.~;

i

\

/ ~

,/'

}':

,a

\

. /

't -

jg Figure 4-8. . AF piston local model.

FaAA-84-2-14

.y

-+.

l

.* , .y . .

, ,' t

• y s l

^ #

- -94989. TO +91615.  !

Wrist gQ.n,bose spojMced -  :

, -91615. TO 45965, i

' -85445. TO -78475. i

. . -78475. TO -71995.

, . 4 .

~

i

- V' ' '

-71905. to -65334.  ;

t Washerydin .

.. c i spotface4

~%; .

..-66335. TO -58765. -

, . . . .< f

• m ,s -58765. T O' - 5 2,1 9 5 .

f.

^

. . .  ;- 4 ..

. -52196. TO -45625.

'j f.['l; =

Bolt hole bore; d

1 @ - '

• -46625. TO -39855. l

- i$3 , .

, ,. , 4

?

lj h

? .g.

-39855. 70 ~3248d. .- -

l

. n t .

4 jd D2485. TO -25915

\

.i q, ; -

, ';- i-

,. -25915. TO -19345. -

y

. . [. ,. . R.. ..

-19345.TO -12775.

g

,47 (4 _ E .-lg g

-12775. TO 1: ; ,

-6295.

E

.. N' '

C g i

-6286..TO

. , . . A2929.  ;

$ L$LCO ar . PISTON ($1NGLE RIKG5.01SPLA .

ADING) ,

.. j i ^

suecasi E

~

1 -

~

.4~' . .

Figure 4-9. AF piston global model with stress contour of average element stresses.

l '.

?

m i .

f  ; M84 .b . f.

w ..

. . y; . .; . r. +

14

~ ~

7%3,.sk ' 'f I

g;. ;,,:

%  ?: .:7% WristJin_ boss. ' '

W::1Me g. " *

=.TQ Y' . s.:' a J-:), .

l . :,a ttacer' 4 '

. h .po . :.; ; .

.g '

Q . **eWn. .

..s. u .. .. .

. . A.  ; g:n W .93 see' - t 2,s 4 m o .. a 17 . .c- ..;.

y&;3 ;T.V. ; -: n.:

.s..Q . :;._y .. AyGg r.

. ...,. yp.ny;;, .g ;, _ y qw e s.s g yg y.p n g .

4

. m .> n c 3 k, .. ::

.. . U. 3. .~ .,. '. ::y :

~,

. R -ntssee.

-;].3.m,I.Nh.'Mi N.N f c

$ gg$[D;.:i:N((W ;;.gn.}7

=  ;

. 7

o

:.

o M;;p$.

.. . s -

3

~

k c . .~ , .. ..c.o?.r...

?. .r,7o fieN ,sse. .

4 Z; ..:..<y. .v.;,e

. sy 4 ... .. wa. .

.y m.+;.:. .-~

. ... s 3 3.t.

. ~ ;-

a , c . .

.g

.. .. ..,x

. - 4,_4 5A T ' -

. ' . { $$'..yC, ; :. 4,pkg'yj h:i sse >

I.hT[.. .q. . . ,;f ..Qj-[ih..

.g, l.

gQ n..y v.; y .; . . ; ' 4 - 'y

...;.y.y.

j. ' Np.f + 0(y' KL .l. . i M'. y ' '

t: . . . -

' ' s ..6;,2.Q.W._.eggggf *]y,s.ee M. /N c ..  :

. ..i  : . ?

• r.. .

' h??'yasher landing- - c 1 wl.?d ,see.

~

spottaced.J.

~

? - I 7

.- y< 2w.3 fg. .. ,.y[.

g .- g.-

,Q..

.. ., Af n~.

a .Q.

g jpf;'.jg:.

3. c.

z t g.y. .

y; h

,;;;h

~

]l'..'.3

.4y.

.S 7.*.j:Q,4 ,f k.M k:?+;$ p x . ,c . .

+,p.

v. e ps 9 69sse.

g.y;Qy: .m .

. a.

sy 4 . 2w n. .v. . ra.

n

+

' c '-

n,. . . - *.

. . . w'-

esas W

n. . ov. 3 y . . .:;x.

,1;;. - .. ..

y J.~ u.F. A x.

^

s ;:. . .- ..

..+ n . .mm .

'.- -995+se : L-u-pM1.3

.V

.m

• ^7 W @::

c.97ec&r%.:  : U_B.olt"hoto twose w m . _ . _ _

A..3 '

}s ..x-([9.z.ggtp g; g

-' . } 5 t; ._ ' " :' TO % 9580. 2

[s.. ,

M.~

l ly

~ ); ..  ?~l V

. (l$.G:l';.

f ?.l. f...l.: Q).fl}f / %,. w: ..9.'brp, . .:'ig:

. ,. lts -. . i. f Wi. %y. . g . . .

. .. g . '

• 4,

• g,.

4

}TOS{-39544 t K

f ,. . .

.,.s..

T j. g;x ' g o.q . .yg4. -

• ^

w' y

5

~,.

q f , y,

m. . > <

,, ..1, ,

.t , 5 5 ,:  %'n.,,: . :. . ., '. ... q.g&c. .:7 ' , ue '*

as

..l .= T -295e9.

..~ s ..

. .: p.

g. . as y up O w %,qh..n p:e e u c .J,j w;m? li g. '.

% ' i . n.y - ' '

^ .

L '

k .e L

"~

y~,.. 1. w ... . 4 ;g.  : '? -29500: TO -t9509.

4-

.fs. n ,,. .,:. r . s_

1. wa y%... ' ' .

~

s

.ng .. c. : ; r s M ..

fit 4 - 3 '.  ?

g[:g.%.. . . M M's

.:4.. .f. p n%M yf.P P- ~N *" ~ '

- - -19589/ T0 . -9589.

J.W y.f yg : .

, , m..  ;.mf: 334SQ i : ..a,. '. -

.d[e.- ..

J ;. v 1%.; s:7;

.K. ye

,. p . .-

7 .. ,

_4 m, -4549

' { .~~.\ TO ' -4500.

? A fsyff , h~.n . '. .n

..; o. g-g.y.;:;ga..

.:.:~g g .y - 3j{ . ,d gg.-  :

k

[KEYORT1'=10 k:. e.n;

> :u m .

d Tdy Mgw.qs g . t. A $ INGLE]RIf k. -~. .. 40!N9) :A. 9 [

g. , t hh l, f .7  ;. .

7., .

Figure 4-10. AF pisten locai model with stress contour of average element stresces.

)

r .

U 5.9 FATIGUE AND FRACTURE ANALYSIS The metallurgical evaluation of the AF pit, ton skirt provided in Section 2 concluded that .the observed cracks resulted from fatigue. Additionally, no macroscopic defects -appeared to exist prior to the parts being placed in service. It therefore appears that the cracks -initiated and grew due to +he cyclic stresses imposed during running of the engine. The experimental and

- finite element _ stress analyses results presented in the previous sections are combined with the fatigue and fracture properties of the material to analyze .

the possibility of crack initiation in the piston skirts. The growth behavior ,.

of initiated cracks is also analyzed, and comparisons made of the performance '

of AF and AE skirts.

5.1 Material Properties

. The fatigue and fracture properties of the piston skirt material are presented in this section.

5.1.1 Fatigue The pistons experience 1.35 n 106 stress cycles every 110 hours0.00127 days <br />0.0306 hours <br />1.818783e-4 weeks <br />4.1855e-5 months <br /> of engine operation. Crack initiation under high cycle fatigue conditions is therefore of concern, in which case the endurance limit of the material is the property of interest. Reference 5-1, and references cited therein, indicate that a lower' bound for the endurance limit of cast iron with the properties of the 100-70-03 material used in the skirt is 30 ksi.

The endurance ' limit of 30 ;ksi is applicable to fully reversed uniaxial

. loading, in which case the mean stress is zero cad the stress system is par-ticularly simple. In order to perform the analysis on the piston skirt, two complications must be accounted for: non-zero mean stress, and multiaxial stress system.

The case of non-zero- mean stress can be. treated in the standard manner by

.use ; of a Goodman diagram. Figure 5-1 shows the definition for cyclic loads employed here. Figure _5-2, which closely follows Figure 15b of Reference 5-1, 5-1 E _

shows how the allowable cyclic stress for infinite life varies with the mean stress. This figure, which extends into the compressive range, is similar to that employed in Reference 5-2 for their fatigue analysis of nodular cast iron piston skirts. The allowable stresses indicated in this figure are such that yielding. in tension ce compression is precluded, and at zero mean strcss the value of oa is equal to the endurance limit, o.

n The value of the yield strength of 60 ksi was used'in the generation of Figure 5-2. This corresponds to a typ1 cal value measured at FaAA and as reported in Section 2.7.

The requirement that %ax _I. 'ys ' %in .2 -8ys, combined with the definition of og and %ean, along with the requirement that ca " "n when %ean = 0 defines the q lines in Figure 5-2.

The condition outlined in Figure 5-2 are conveniently summarized in Figure 5-3, which plots directly the allowable cyclic stress for infinite life for a _given mean stress. The representation in Figura 5-3 closely follows that inc'iuded in References 5-2 and 5-3.

The results presented in Figures . 5-2 and 5-3 are for uniaxial stress.

Figure 5-3 can be generalized to multiarial stress systems by procedures presented in References 5-3 and 5-4. However, the experimental results pre-sented in Section 3.4.2 show that the stresses are nearly uniaxial in the highly stressed region of the stud mounting boss.

~

Therefore, Figure 5-3 can be used directly in conjunction with the principal stress that has the largest l- cyclic amplitude (oggi in tnis' case).

5.1.2 Fracture Mechanics Properties l Analyses based on the previous section can be used to predict whether an initially uncracked component will have an infinite life. If a finite life is predicted, then a crack will initiate. Once it has initiated it may or may not grow, and its possible subsequent growth can be treated by fracture mech-anics principles. The fracture mechanics properties of the material are i

required in this analysis and are summarized in this section.

Fracture Toughness. The fracture toughness of the material governs the point at which a' crack grows catastrophically and is the critical value of the t

5-2 l-

3

~ applied stress intensity factor [5-5]. Reference 5-1 and references cited therein provide information on the fracture toughness of nodular cast iron.

The fracture toughness it, influenced by temperature within the range of inter-est (room temperature to - 300*F), and increases with increasing temperature.

Also, the toughness tends to decrease with increasing strength levels. Table 5-1 summarizes some relevant fracture toughness data from the literature.

This . table shows K Ic of 40 ksi-in /2 l

to be a reasonable but somewhat conser-vative.value of the fracture toughness. This was used as the nominal value of KIc in the fracture mechanics analysis. This value is especially conservative for use at the operating temperature of the piston skirt. -

Fatigue Crack Growth Characteristics. For a given material and environ-ment, the rate at which a fatigue crack grows is dependent mainly on the cyclic value of the stress intensity factor (AX = K max -Xmin) [5-5]. Other factors, such as the mean value of K (measured as R = Kmin/Kmax) also have an influence. Information on the crack growth- rate, da/dN, as a function of AK for nodular cast iron is included in References 5-1, 5-7 through 5-10. Rep-resentative results of da/dN vers % AK are included in Figure 5-4. A reason-able and somewhat' conservative representation for the crack growth charac-teristics is given by the upper end of the shaded band in Figure 5-4. This result is conservative for all cases ,shown except R = 0.7, which corresponds to . cyclic loading with large mean tensile stressos. The cyclic stresses in the piston skirt in the region of interest are primarily compressive, and therefore do not have such a high positive R-ratio. The following equation provides the relationship between da/dN and AK that was used, h = 2.47 x 10-12 [3g)5.15 (5-1)

The constants in this equation are ?pplicable when da/dN is in inches / cycle and AK is in ksi-inl /2, l

Equation 5-1 overestimates crack growth rates when AK becomes small, because- there is a threshold below which cracks will not propagate. Reference t 5-9 provides some information on fatigue thresholds for various R-values. The value of AK th for R = 0.1 is indicated to be 12 ksi-inl /2 This value appears 5-3 Im-

_ _ = . . _ . .

Table 5-1 SUIMARY OF FRACTURE TOUGHNESS DATA FROM THE LITERATURE FOR NODULAR CAST IRON WITH STRENGTH LEVELS SIMILAR TO 100-70-03 T, 'F Elong. % Ref.

KgI-in/2 ks l oblt, ksi ki.

41 70 133 80 3.6 5-6 36 70 85.3 76 1.6 5-6 48 220 116 90 --

5-7 40 70 116 90 -- 5-7

~ 85 70 90 58 --

5-7

> 80 25 75 62 >3 5-1 47 75 --

104 --

5-1 o Nominal value for calculations: 40 ksi-inl /2, l

G 9

5-4 IL

to be somewhat optimistic, because Reference 5-11 suggests a value of 6.4 k'si-inl /2 for a variety of steels for values of R close to zero. This value is stated in Reference 5-11 as being somewhat conservative., and is used as the applicable value for the nodular iron considered here.

An approach to non-propagating fatigue cracks is provided by References 5-12 and 5-13, in which case the influence of R-value on crack propagation is treated.- In this case, the Forman relation for crack growth is combined with the treatment of R-value from Reference 5-13 to provide a treatment of the

- influence of R on AK th. Cracks are taken to not propagate for a given AK and R if AK < Akth(R) . Reference 5-11 provides the details, and the procedures are included in the BIGIF code. This code is used for the fatigue crack growth calculations presented in Section 5.3, and is a widely used code for fracture mechanics analyses.

5.2 Fatigue Crack Initiation Analysis This section of the report combines the stress analysis results of Sections 3 and 4 with the crack initiation criteria from Section 5.1.1 to

- assess the possibility of cracks initiating in the stud coss regions of the AE and AF piston skirts. Attention will be focused on the location with the highest principal stress, and the stresses will be considered to cycle between the values -for pressure plus inertia, and inertia loading only. The inertia only stresses are obtained by reversing the sign of the pressure stress and multiplying by the ratio of the inertia to inertf a plus pressure loads. This ratio is 0.025, as discussed in Section 4.1.

Table 5-2 presents the selected stresses from the experimental and finite l element results. . Stresses for the highly stressed region of the skirt are provided along with the corresponding values of the cyclic and mean largest

- principal stress. The results of Table 5-2 provide a direct comparison between t:1e experimental and finite element values for the peak stresses in an AE skirt. It is seen that the experimentally measured peak stress is 667. of the finite element value for a uniform displacement along the crown-skirt loading ' ring. The disagreement between the experimental and finite element values can probably be attributed primarily to the over simplified boundary 5 ,

L.

e ,. :. -

Table 5-2 SELECTI:1) MAXIMUM AND HINIHilli l't.ASl~lc STRESSES FOR lilGill.Y STitESSED REG 10flS OF AE AND AF (1111% loading on inner ring) ,

all values in Lsl AE AF unifonn uniform unifenn uniform experimental 'displacemcnt load displacement load min o -47.2(I) -71.4 -144.6 -104.5 -267.5 T (press.b}nertia) m max oggg 1.18 1.78 3.62 2.61 6.69 (inertia only) oiII,a 24.2 36.6 /4.I 53.6 137.1 oggg,, -23.0 - 3 4 . 11 - /fl. 5 -50.9 -130.4 (1) 0.975.x maximum value from Tables 3-1 and 3-2.

t

conditions, such as _ the assumption of a rigid wrist pin or completely rigid or compliant crown. Other contributing factors are finite element modeling

-errors, especially_ the match between the global and local models, and possibly missing the exact location of peak stress in mounting of the strain gages.

None of these factors is believed to produce major errors, but when taken in combination could account for the 337, disagreement.

The results of Table 5-2 are plotted on the allowable stress envelope for infinite life in Figure 5-5; The results of Figure 5-5 p* edict that cracks will initiate in all cases considered except for the experimental measurements on the . AE piston. The stress values in the AE skirt are lower than the corresponding values for the AF skirt, thereby indicating the superiority of the AE desigr..

The finite element stresses in Table 5-2 are well above the compressive yield strength ~of the material. Therefore, localized plastic deformatic7 will occur on at least the first few cycles o# loading for the cases considered.

This plastic deformation in compression will cause redistribution of the stresses in the highly stressed region, which shifts the mean stresses towards tension s Calculations of stress redistribution were performed using the contained plasticity analysis included in BIGIF [5-12, 5-14]. This approach utilizes the Neuber approximation [5-15] for contained plastic deformation in

. conjunction with a Ramberg-Osgood representation of the material 's stress-

-stra1n behavior _ in order to provide an estimate of the influence of plastic deformation on the . stresses within a contained plastic zone. The redistri-2 buted stresses are used in the fatigue. crack propagation analysis to be. pre-sented in Section 5.3.

• . Although the results in Table 5-2 predict that fatigue cracks -can init .

iate under stress levels corresponding to the finite element stresses, these cracks will not necessarily propagate. This is because the initiated crack will be growing ~into a region of decreasing stresses, because of the steep stress gradient. in the stud boss region.

b^

5-7 E.U

.c -

In sunna ry, the results of finite element calculations predict that cracks will initiate in both the AE and AF skirts. The experimental results predict that cracks will not initiate in the AE skirt.

5.3 Fatigue Crack Growth Ar.nlysis The concern raised by the discovery of cracks in the AF piston skirts is that the cracks may grow during operation and result in a failure of the pistons. This section analyzes whether the initiated cracks could grow in the pistons. The presence of a crack in the piston skirt does not necessarily lead to unsatisfactory performance of the piston, because the initiated cracks may not grow. Even if they do grow, they may arrest as they grow out of the localized region of high stress. The behavior of any cracks that do initiate can be analyzed by use of fracture mechanics principles. The fracture mech-anics properties of the material sumrr.arized in Section 5.1.2 were assumed in

'this analysis, which was performed with the BIGIF fracture mechanics code

[5-U].

t The elastically calculated stresses from the finite element results are -

sufficient to cause yielding in both piston types. Residual tensile stresses are predicted in the localized region where the stresses exceed the yield strength. The contained-plasticity capability in BIGIF was used to obtain the elastic-plastic redistributed stress fields. Figure 5-6 presents the princi-pal stress with the largest absolute vciue as a function of distance into the piston skirt for the AF piston subjected to pressure and inertia. Also shown are the redistributed stresses which exist after yielding. This figure reveals the very steep stress gradients in the skirt in the localized highly stressed region near the stud boss, and shows the substantial redistribution of stresses resulting from the clastic deformation. The elastic-plastic redistributed stresses show a substantial tensile component that can cause cracks to grow. These tensile components result from an upward shift in the mean stress due to yielding caused by the high compressive stresses under peak cylinder pressure. .

The stress intensity factor range, AK, due to the plastically redistri-buted cyclic stresses was calculated using BIGIF. The crack was idealized as.

5-8

a plane . strain edge crack' in a strip of finite width. An initial crack depth of 0.01 inch was assumed. Calculations were performed for 62 ksi yield strength with 98 ksi ultimate strength, for both the AF and AE skirts. BIGIF-calculated stress intensity factor ranges, AK, are compared to the threshold value for crack growth, AKth, to determine crack growth and arrest points.

The BIGIF calculations account for the variation in AKth, with stress-ratio, R. Crack growth is only possible when AK(R) > AKth(R).

Although it was shown in Section 5.2 that the AF skirts experience stresses sufficient to initiate fatigu:: cracks, the finite element stresses s

for the AF piston under 88% inner-ring uniform-displacement loading are predicted by BIGIF to be insufficient to cause the crack propagation observed in the AF piston skirts. As previously discussed, the uniform-displacement

. loading . condition represents a perfectly rigid crown and, therefore, an extreme bound on skirt-crown interactions. The uniform-load boundary condition, however, represents a perfectly comrliant crown and, therefore, an upper bound stress condition in the skirt stud-boss region. The uniform-displacement boundary condition is believed to more closely represent actual loading conditions than uniform load, however, the resulting stresses are too low to produce the observed cracking. For this reason, a sensitivity study was performed in which the fracture mechanics calculations were made using various stress values ' between the extreme bounds (uni form-displacement and uni form-l oad ) . - Table 5-3 summarizes the results of this study. Using maximum stress values equivalent to 10% of the way from the uni form-displacement

-condition to the uniform-load condition, the AF piston shows crack growth and arrest between - crack depths of 0.016 and 0.056. Since observed crack depths in AF pistons range from 0.10 to 0.30 inch, the stress values were increased another increment. Using maximum stress values equivalent to 15% of the way from the uniform-displacement condition to .the uniform-load condition, the AF shows crack growth beginning at a crack depth of 0.015 inch and arrest at a crack. depth of 0.130 inch. In all cases, Kmax remained well below the critical value K IC (= 40 ksi-inl /2), indicating no catastrophic failures.

5-9

Tablo S-3

SUMMARY

OF FRACTURE f.tECllAtJICS CRACK GilOWTH AND ARREST RESULTS Effective Boundary Condition Ratio inillallon* Astust" Peak Elastic K an a x (uniform displacement / uniform load) Crack Dupth Crack Dupth surf ace Stress, ogg, (In) (iii) (k si) (k sl6)

• Y AF 0.016 0066 -120.0 6.24 y 90%/10%

5 .

AE (sio orownli) (no usoweli) -78.6 3.58 AF 0.015 O.130 -127.8 6.65 86%/15%

AE (no growtli) (sio urowili) -82.4 4.00

w.

• e-= .ey me +-w

• Initiallon: AK > AKTil

• • Arrest: AK < AKTil

b References 5-1 fron Castings Handbook, edited by C.F. Walton and T.J. Opar, Iron Casting Society, Inc., 1981.

5-2_ R. Reipert, H. Moebus, and K. Sche 11mann, " Computer Design of a Steel-Nodular Cast Iron Piston Capable of Withstanding High Loads for Appli-cation in Medium Speed Diesel Engines", Paper No. 83-0GEP-8, American Society of Mechanical Engineers, New York,1983.

! 5-3 H.O._ Fuens,' "A Set of Fatigue Failure Criteria", Journal of Basic Engineering, pages 333-343,- June 1965. -

5-4 H.0. Fuchs and R.S. Stephens, Metal Fatigue in Engineering, John-Wiley "}

and Sons, Inc., New York, 19807-~

~

5-5 D. Broek, Elementary Engineering Fracture Mechanics, Sijthoff and Noordhoff, Alphen aan den Rijn, Netherlands,1978.

5 R.K. - Nanstad, F.J. Worzal z, and C.R. Lcper, Jr. , " Static and Dynamic Fracture Toughness of Ductil' Cast Iron", AFS Transactions, Proceedings of 79th Annual Meeting, Vol. 83, pages 245TE36,1975. ,

5-7_ B. 0stensson, " Fracture Toughness and Fatigue Crack Growth in Nodular Cast Iron", Scandinavian -Journal of Metallurgy 2, Vol. 2, No. 4, pages 194-196, 1973.

5-8 D.G.. Smith, ,and K.P. Jen, " Fracture Properties of Nodular Iron Cast-ings, Grade 80-55-06", Tennessee Technological University Department of Civil Engineering Report TTU-CE-82-1, Cockville, Tennessee, October 1982.

5-9 M. Castagna, P. Ferrero, R. Medana, and E. Natalu, " Fatigue Properties of 'In-Mold' Ductile Iron", AFS International Cast Metals Journal, Vol . 4, No. 4, pages 63-72, December 1979.

5-10 M.S. Starkey and P.E. Irving, "A Comparison of the Fatigue Properties of Machined and As-Cast Surfaces of SG Iron", International Journal of Fatiguej page-129-136, July 1982.

5-11 S.T.' Rolfe and J.M. Barson, Fracture and Fatigue Control in Structures, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1977.

s5 P.M. ' Besuner, et al., "BIGIF - Fracture Mechanics Code", EPRI Report NP-1830-CCM, Electric Power Research Institute, Palo Alto, California, 1981.

A. Yuen, .S.W.f Hopkins, G.R. Leverant, and C. A. Rau, " Calculations

~

' 5 Between Fracture Surface Appearance and Fracture Mechanics Parameters for Stage II Fatigue Crack Propagation -in Ti-6A'-4V," Pktallurgical-Transactions, Vol. 5, pages 1833-1842, August 1974.

5-11

[* ,

5 P.M.' Besuner, and S.A. Rau, " Stress and Subcritical Crack Growth Analy-sis _ Under Contained Plastic Conditin.s," EPRI Report NP-81-8-LD, Elec-tric Power Research Institute, Palo Alto, California,1981.

5-15 H. Neuber, " Theory of Stress Concentration for Shear Strained Pris-matical Bodies with Arbitrary Nonlinear Stress-Strain Law," Journal of

' Applied Mechanics,.pages 544-550, December 1961.8 E

v m

e J .- e t- 5 -

w- n, ,,r.-, - . -g. ,

amean = 1/2 (amax

• umin) aa = 1/2 (amax amin) a max. A

/ I Ga I

---

-- -- ---- ---..---- ~a mean'A~

. O min Figure 5-1. Definition of terms for cyclic applied stresses.

• d

B ayges - A

/

/

/

/

g/* /

/

/

/

/

\

' On " g ,/

/ B

/

D /

/ *

/

/ c l /

I /

s I / t

-60 / q'i 60 i

t / {

l

/

/

/

7

/ *

/

/

\

/

/

/

/ C

/

/

t

' /

/

/

/

/

/

/

/

E D t

I Figure 5-2. Mean and cyclic .; tresses for infinite fatigue life.

I Fs A A-84-2-14

s

7. . ,

.; e e ., . -_.

9 F.

, r -

i

+ +-

f 4. f g ,,WY 77 ,

_'- 5 e. Y 4

.e .c :

a g o

n e Ua(ksi). -

100 -

.a; 4

4 e c*--UYs '

..+ ,/iN -

/ N-

, . r

• • M/ N N

Cl. 6n \ s

, = ,c. .

N 3,

, , - E' ,

, , A'% %m 6

-100 4 -50 0 50 a 1001 m(ksi)

.>+ ,

. E3 *

• 6ys Uult.

..y. t>

rre

, e f

,y.-

+

Figure 5-3. Mean and cyclic stresses'for infinite fatigue life.

6-

, . , e

~

- +

- y. ~

~

N

.w 1

.s .

0 4

g 4.s >

.t ,

P

- Fs A A-84-2-14

^' '

• m _. , _ ,

]

10-3 i i 1 l 10-4 -- -

A -

3 10-5 _

p, _

o 'd

%' k}

y/. o',

z y 10-6 b l

10-7 - --

B and, Fig.19, p.344, re f. 5-1 I I

.10-8 I~ 'I 8- 10 20 30 40 60 dK (ksi-in /2) 1 Figure.5-4. Representation of da/dN - AK data for no'dular condition.

FaAA-84-2-14

O O

m

- m CJ L

a O M CD G1

.c n

3 o

e- .

O m C

1m L L

Cn

~ O g C o

"O

_ c)

~

W u t e o -

g=m =

m 2 d w L

_. -e Cf) a (O

^

l I UJ 1 l l 1 1 I I o

O gr O O. O O O l- 'm O CO (0  % N Cf)

~

w a.

g a (ism) 2001IldWY SSBB1S Ol10 A0 d <  %$

W _ C W

2 vb co e N t ,

I w

g 5,,

, I a

e 5%

% o O

, O g "

g mc

• O 9 l u-

- ma

c. "

uo cr} mC

- 3 a m%

m u O O ca m

• bm O I mm

' J e- .

. .e!

W '3 iM O *'

@ o "g e 1 6 E 3 m E CD o _ -

L

. .J O

~

O e

1

- /

. FaAA-84-2-14

B y

• n

'40 -

INERTIA

, Elastic-plastic 20 - Elastic 0 """*****--I I- --I I i y 0.10 0.23 0.30 0.40 0.50 0.60 5 -20 -

DISTANCE INTO SKIRT (in) w _ _ _ _ ---

w ,,, e - - ,

E -40 -

F- gen"

< -80 -

a: .. ..* .

@_gg

,f mmG MESSURE + INERTIA E Elastic-plastic I Ela s tic

-100~

-120 ,5 S -140 -

i-n

?

4 Figure 5-6. Elastic and steady-state clastic-plastic stresses as a function of distance into the skirt material for the AF piston.

i .,

' 6.0 AE PISTON SK1RT INSPECTIONS FaAA. has conducted oddy current inspections of seven engine-operated AE

~ skirts. This high-resolution procedure, developed to differentiate between superficial dye-penetrant indications and fatigue cracks, is described fully in FaAA Procedure-NDE 11.5.

Included in these inspections were four skirts from the SNPS DG102 engine, which had completed over 300 hours0.00347 days <br />0.0833 hours <br />4.960317e-4 weeks <br />1.1415e-4 months <br /> total operation at 75% load and over, ' including 100 hours0.00116 days <br />0.0278 hours <br />1.653439e-4 weeks <br />3.805e-5 months <br /> at 100% load. One skirt was inspected from a P.V-16-4 engine at the Kodiak Electric Association. This engine had experienced over 6,000 hours0 days <br />0 hours <br />0 weeks <br />0 months <br /> 'of service with the AE skirts, at a peak firing pressure reported by the utility to be approximately 1,200 psi. No skirts were ' inspected from the TOI R-5 Jevelopment engine after operation at 2,000 psi or above' for over 600 hours0.00694 days <br />0.167 hours <br />9.920635e-4 weeks <br />2.283e-4 months <br />, according to TDI.

Thes'e inspection reports are available from FaAA. None of the skirts disclosed any indications above the acceptance size defined in FaAA Procedures NDE-11.5 t

e 6-1

J1 .

7.0 CONCLUSION

S

. Four cracks in three modified AF-type piston skirts removed from the SNPS diesel engines were found to be. the result of fatigue loading. These cracks were concluded to be representative of linear indications observed in the crown-to-skirt stud attachment bosses of all 23 AF skirts.

Comparative finite element stress analyses were conducted on the modified AF and the replacement AE skirts. In addition, strain gage measurements were carried out -on a piston with its crown subjected to hydrostatic pressure loading. ^~

The results of these tests and calculations showed the following:

Stresses under simplified loading conditions and at uni form skirt temperature are -significantly lower in the stud attachment boss area in the AE than in the modified AF skirts.

- The experimentally-measured peak stresses were about 30%, lower than the finite element calculations performed using boundary conditions indicated to be most reasonable from the experi-mental observation.

The discrepancy between the experimental and finite element results can be attributed to the simplified assumptions as to boundary conditions, especially the distribution of load

' between the crown and skirt, and possibly to some difference between = strain gage lccations and the location of the peak stress.

- Cracks are predicted to neither initiate nor propagate when experimental stress numbers are used in the fatigue analyses (applicable only to AE skirt).

- Finite element stresses predicted cracks to initiate in the AE and AF skirts.

- When the assumptions of crown-ski rt load distribution were modified such that cracks would be predicted to initiate and grow to a small depth in the AF skirt (as actually observed in the modified AF skirts at Shoreham), the identical loading assumptions led to the prediction that cracks would not grow in the AE skirt, even if they wero to initiate.

7-1

c_ .

From both- the a experimental stress measurements and the inspect;ons of seven engine-operated AE skirts, which disclosed no cracks, it appears unlikely that cracks will initiate in the AE skirts.

. The experimental measurements and finite element calculations of .;;ress did not include thermal . stress in the skirt or the effect' of - thermal -distortion on the load distribution between the crown and the skirt.

G 9

4

.r 4.

i e

7 .

p r--' w. . , - - , - , ,, w- , ~av- -e