ML20100N677
| ML20100N677 | |
| Person / Time | |
|---|---|
| Site: | Shoreham File:Long Island Lighting Company icon.png |
| Issue date: | 10/01/1984 |
| From: | FAILURE ANALYSIS ASSOCIATES, INC. |
| To: | |
| References | |
| OL-A-036, OL-A-36, NUDOCS 8412130316 | |
| Download: ML20100N677 (55) | |
Text
h o YWf~
[4
.3 c 3.
< 32 2 o (
g qu
- ~~.
[0 l v NY 5.Y 4 'D
%A 68 L
(o fg 9g[M $',W, failum
)QS k
93
. 3,s,..,. :.c..m.3,:. cossa.s.,
ms ia, sa,-:=e aca: eo e: s ':
s1,,
e5
. e.'o c. o
>:2 5 is. :. 'un n 2 *
- s s/4,0)FaA IB 7396/DOH:R033 lA THE INFLUENCE OF THERMAL DISTORTION ON THE FATIGUE PERFORMANCE OF AF AND AE PISTON SKIRTS Prepared by Failure Analysis Associates C
2225 E. Bayshore Roa:
Palo Alto, California 9:373 Pre ared for TDI Diesel Generator Owne s G oup BU6 LEAR RESULATSEY 90Mul4819N seense u 66 - S.2 2 m$d7(,
lamemeneres L It-C o IIf June }cca.
'* EllTIFIED AppNeast V
REEEIVED --
latervenef
' EJECTED Coal *g Off'r
[ ~N, festratter egyg /d-/- 74 k_),
8412130
? " a o "3 1 6 8 4 1 0 0 1**ojg woe
.. =...,
- ,-=:
. e m.
J STATEENT OF APPLICABILITY This report addresses the structural adequacy of the Transamerica Delayal Inc. types AE and AF piston skirts supplied for use in R-4 series diesel engines.
The modified type AF skirt was originally installed in the engines at Shoreham Nuclear Power Station, Grand Gulf Nuclear Station, and San Onof re Nuclear Generating Station.
These skirts have been replaced at Shore-ham and Grand Gulf by type AE skirts. Type AE skirts have also been installed at Comanche Peak Steam Electric Station.
Evaluation of the type AF skirts at San Onofre will be reported separately.
/
O O
i
\\
/m' (v)
EXECUTIVE SU SARY This report addresses the influence of thermal distortion of the piston crow 1 on the structural integrity of the Transamerica Delaval Inc. (TDI) types AE and At piston skirts.
This report is an extension of earlier work
- which provided the results of experiments and finite element analyses on these skirts under isothermal conditions. This report was prepared on behalf of the TDI Diesel Gene r > t or 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 cyclic stresses, and therefore the fatigue performance, of the piston skirts is influenced by load transfer between the skirt and the crown on the two concentric rings over which the crown and skirt can contact one another.
This load transfe" is influenced by thermal distortion of the crown O
and the initial size of the gap between the crown and skirt. The influence of 3
v thermal distortion and initial gap on cyclic stress levels and the possibility of lift-off of' the ' crown frot the skirt were estimated by combining the results of isothermal finite elenent stress analysis with a crown / skirt inter-action nodel.
The crown / skirt interaction nodel treats the crown and the s<iet as springs whose stiffnesses were estimated by finite elrent techni-ques. Thermal distortion is included as a thermally-induced displar.ement that is calculated by finite elements using a steady-state temperature field in the piston assembly based on experimental measurements.
Predicted and observed gap closure pressures and load transfers between the contact rings agreed well with one another.
Cyclic stresses under isothermal and steady-state opar tion were calcu-lated by use of the crown /siirt model for a variety of initial Tap sizes. The influence of lif t-of f was included in cases where it was predit ted to occur.
- " Investigation of Types AE anc A~ Diston S<irts," Report prepared by Failure h)
Analysis Associates for Transame-ica Delaval Inc. Diesel Generator 0-mers V
Group, Reoort tio. Fa AA-31-2-14, Da! o A'to, Cali f *M a, Maj 1984 0
11 L
I b) t Crack initiation and propagation analyses were then performed by procedures followed.in earlier isothermal analyses.
The conclusions obtained earlier regarding cracking of the AE and AF skirts were unchanged, but now the arrested crack depths predicted for the AF are in better agreevnt with field observations. Overall, the earlier conclu-sions regarding the integrity of the AE and AF skirts are also unchanged.
These conclusians are that cracks may initiate but will not propagate in the AE, and that cracks will initiate and may propagate in the AF.
However, any cracks in the AF are predicted to arrest at depths less than 0.5 inch, which is comparable to field observations of cracking reported earlier.
O O
iii
]
,. \\v)
/
TABLE OF CONTENTS Section Page
- STATEMENT OF ADPLICA3!LITY........................................
1 E X E C U T I V E S'J M MA R Y.................................................
11 1.0 INTRODUCT104......................................................
1-1 Section 1 References..............................................
1-2 2.0 FINITE ELEMENT ANALYSIS...........................................
2-1 2.1 Load Considerations..........................................
2-1 2.2 Stress Analysis..............................................
2-3 2.3 S t i f f n e s s An a l y s i s...........................................
2-4 2.3.1 Skirts................................................
2-4 2.3.2 Crown.................................................
2-5
)
2.4 C rown /Sk i rt I n t e r a :t i on......................................
2-8 Section 2 References..............................................
2-9 3.0 CROWN / SKIRT INTERACTION M0 DEL.....................................
3-1 3.1 Revi ew of Experi ment al Obse rva ti on s..........................
3-1 3.1.1 Diston Assembly.......................................
3-1 3.1.2 Stud Attachnent Loads and St i f f nes ses.................
3-2 3.2 C r0+ 4 /Sk i rt I nt e r a ct i on Mo del................................
3-3 3.2.1 Power Stroke..........................................
34 3.2.2 Exhaust Stroke........................................
37 3.3 Compari son with Experiment al 0bse vati ons...................
3-10 Se:t i on 3 Re f e renc e s.............................................
3-11 4.0 F AT I GU E AND FR AC TU R E AN
- L V SI S.....................................
4-1 4.1 Cyclic Stresses..............................................
4-1 4.2-Fatigue Crack initiation Analysis............................
4-2 4.3 Fatigue Crack Gaosth Analysis................................
4-3 Se c t i o n 4 R e f e r e n : e s..............................................
4-5
' 5. 0 CnNCLUS!0NS.......................................................
5-1 APDEN) t.A:
COM?]NEN' TASr. DESC41DT10N..................................
A-1 iv
p
1.0 INTRODUCTION
The purpose of this report is to address the influence of thermal distortion on the fatigue performance of Transamerica Delaval Inc. types AE and AF, piston skirts. This report is an extension of earlicr work M] which provided the results of experiments and finite element analyses on these skirts under isothermal conditions.
This report draws heavily on Reference 1-1.
The largest stresses in absolute value in the piston skirt result from firing pressure.
The peak firing pressure occurs when the piston is close to its top dead center (TDC) position on the firing stroke.
The pressure force on the crown is transmitted between the crown and the top of the skirt in part on the ring that coincides with the stud attachment bolt circle. This ring is referred to as the inner ring.
Load transfer between the crown and skirt can
'also occur on an outer ring. The crown is assembled to the skirt to provide a w
gap between the crown and skirt at the outer ring of 0.007 to 0.011 inch.
When pressure is applied to the top of the crown, the crown and skirt deform, thereby tending to close the gap.
Load transfer between the crown and skirt is altered when the gap closes, thereby affecting the peak stresses in the skirt.
l Once the engine reaches steady-state operating conditions, large temperature gradients are present in the crown.
These temperature gradients distert the crowa, thereby changing the gap. Sucn changes also influence load transfer and stresses in the skirt.
If thermal distortion of the crown is sufficiently large, it is conceivable that the resulting forces, when combined with the inertid forces at TDC of the exhaust stroke, will exceed the stud preload.
This would allow the crown to lif t off of the skirt on the inner contact ring, which would unload a portion of the contact surface, thereby altering the stresses during this part of the cycle.
The influence of thermal distortion and possible crown lift-off are addressed by co9bining the results of finite element analyses and experimental observatioas reported in Reference 1-1 with additional finite element O
g analyses, experimeatal obse vatioas of stea dy.sts'.e crow-tempe stures and a 1-1
model of the interaction of the crown and the skirt.
Cyclic stresses in the stud boss region of the'AE and AF skirts for various gap sizes and for steady-state temperature distribution and isothermal conditions are obtained. These stresses are then used in fatigue and fracture mechanics analyses of crack initiation and growth to provide predictions of crack behavior in these skirts.
Section 1 References 1-1.
" Investigation of Types AF and AE Piston Skirts," Report prepared by Failure Analysis Asso:iates for Transamerica Delaval Inc. Diesel Generator Owners Group, Report No. Fa AA-R4-2-14, Palo Alto, California, May 1934 O
O 1-2
4 2.0 FINITE ELEIENT ANALYSIS The results of finite ' element analyses of stresses in AE and AF piston skirts are presented in this section along with analyses of thermal distortion of the, crown.
The skirt analyses were performed using the same global and local
- finite element models that were previously constructed for the iso-thermal analysis reported in Reference 2-1.
Spring constants of the skirts and crown were also evaluated by finite elements.
These stiffnesses are used in the crown / skirt interaction model presented in Section 3 to provide esti-mates of cyclic stresses for use in the fatigue and fracture analysis of Section 4, 2.1 Load Considerations Three loads were assumed to be acting on the piston: gas pressure, reciprocating inertia, and friction.
In addition, since the piston is a two-piece design, initial internal load is associated with the bolt preload.
As part of the analysis of the crankshaft of these same engines [2-2], a table of gas pressure, accelerations, and friction was developed that provided values for every ten degrees of rotation of the c ankshaft.
This covered rotation from O' to 72Y thereby encompassing the entire four-cycle combustion process.
The combined loading was foun. to be highest at too dead center dJring the power stroke.
At this point, only pressure and inertia are acting on the pisten since the velocity (and therefore friction) is essentially zero.
A peak firing pressure of 1670 psig [2-?] was assumed to be acting when the piston is at top dead center.
A value of 379,000 pounds was obtained for the pressure load froe the pear. firing pressure and cylinder bore.
At top dead center of the power stroke, the pressure load is somewhat of f set by the inertia load, which is exerted by the crown on top of the skirt.
The piston acceleration at top 3
2 center was found from the crankshaft analysis to be 25.1 = 10 in/sec.
The crown weighs 144 pounds.
Therefore, the inertia force is 144 = 26.1
=
3 10 /386.4 = 9727 pounds. Subtracting this from the pressure force provides the maximum net force on the to: of the sH et, 369,333 pounds.
Tnis corresponds O'
to an effe-tive pressJre of 1627 psig.
21
l
,m The other entreme of. the stress cycle in the skirt occurs at top dead center of the exhaust stroke, at which time a tensile load equal to the inertia force of the crown is applied at the top of the skirt.
If lift-off between the crown and skirt does not occur, then the peak stress due to i ne rt.i a can be obtained from the peak stress due to firing pressure by multiplying the peak pressure stress by the ratio of the inertia force to the peak pressure force (and changing the sign to account for the different direction of the loads).
If crown / skirt lif t-off does occur, then the load path of the inertia force at top dead center of the exhaust stroke is different than the path at top dead center of the power stroke.
Under such conditions, the peak stresses during exhaust must be evaluated by a separate analysis, rather than by ratioing the power stroke results.
The following four sets of boundary conditions on the top of the skirt were considered for each skirt design:
- 1. Crown mounted on the top of the skirt with a frictionless interface at the inner contact ring, pressure of 1627 psig is applied to the top of the crown, and the crown and skirt are allowed to deform withoat interfering with each other at the outer loading ring.
- 2. Uniform vertical displacement on the inner crow,/sk i rt contact ring of a nagnitude to react the loading coa esponding to 1627 psig effective pressure on the crown.
- 3. Uniform vertical displacement on the outer crown / skirt contact ring of a magnitude to react the load corresponding to 1627 psig effective pressure on the crown.
1
- 4. A stud load applied on the stud washer landing area and reacting on the oute* loading ring which is constrained to have a uniform vertical displacement.
Of these four cases, Case 1 provides the nost realistic estimate of naxinun stresses in the isothe-mal skirt in the absence of gas closJre and also provides checks on the crown / skirt interaction model.
This loading condition was the only one co9sidered in Re'erence 2-1.
Cases 2 and 3 provide estina'.es ' of the sii*t soring const 3nts th!! are repired for the caown/srict pQ irteraction model dist,ssed in se: tion 3.
co,piri so-or case I wit 8 cases 2 2-2
iv.
and 3 provides information on the suitability of simplified boundary con-I~
ditions at the top of the skirt.
Case 4 provides stress levels at top dead center of the exhaust stroke appropriate for crown / skirt lift-off.
A rigid wrist. pin was assumed in Cases 1-3; no wrist pin was utilized for Case 4.
All finite element runs on the skirt models were performed for uniform te9perature.
2.2 Stress Analysis Stresses and displacements for the four loads and boundary conditions discussed in Section 2.1 were calculated for the AE and AF skirts using the ANSYS finite element computer program.
Details of the models employed are provided in Reference 2-1.
Spring constants were evaluated from the global models, and stresses were obtained by combining the global and local models as described in Reference 2-1.
The stress results are summarized in Table 2-1.
The third algebraic mininam principal stress (oggg) is provided for firing pressure results, and the maximam principal stress (og) is provided for the lif t-of f condition.
These stresses are. pertinent to the fatigue analysis discussed in Section 4 The equivalert stress, e,, is also included in Table 2-1.
The results of Table 2-1 show that the AF piston skiri. p N eslly has considerably larger stresses than the AE. This is especially true for the two conditions of primary interest -- pressurized crown and lif t-off as calculated from the local models.
The appreciable increase in g, when lift-off occurs is apparent from the results in Table 2-1.
The relatively small dif ference l
between the stresses for the AE and AT skirts for uniform displacement on the inner contact ring indicates that this idealized boundary condition does not point out the large stress difference in these two skirts that results from the more realistic boundary conditions provided by placing the crown on the skirt.
Fairly large stress differences between the two skirt designs were observed to be present in strain gage measurements and are indicated by differences in cracking behavior under service conditions -- as reported in Reference 2-1.
The resalts of Tabie 2-1 also show that the peak stresses in
(
the stud boss re; ion are ma:t more influenced by loading on the inner contact ring, being virtually zero for loa ding only on the oute contact ring.
?-3
,a Y,$
2.3 Stiffness Analysis The results of finite element analyses of crown and skirt stiffnesses are presented in this section.
These results are used in the crown / skirt interaction model described in Section 3.
2.3.1 Skirts The relative displacements 'of the inner and outer contact rings between the crown and skirt control gap closure and the load split between the inner and outer contact rings.
These relative displacements can be expressed in terms of spring constants.
The values of the spring constants for the AE and AF skirts were obtained from the finite element calculations performed to obtain the stresses discussed in Se: tion 2.2.
The following spring constants are required:
OG stiffness of skirt at inner ring due to loading at inner ring, kg = F /d ;
stiffness of skirt at outer ring due to loading at inner ring, k kN; stiffness of skirt at outer ring due to loading at outer ring, kg g = F (/ 6 g og=k/3;.
=
g stiffness of skirt at inner ring due to loading
- at outer ring, k g g 4
As discussed in Sect i ca.
3, the re:ip o:ity relation of linear elasticity requires tnat kjg = kog.
These stiffnesses were evaluated as part of the finite element stress analysis.
The constants k and kg are directly obtainable from the uniform g
displacement finite element runs.
T9e evaluation of kog is subje:t to some uncertainty in the present case, because the non-loaded contact ring does not remain planar when the loaded conta:t ring is subjected to uniform vertical displacement.
Hence, a uniform displacement of the non-loaded ring is not clearly defined. However, kog can be more clearly defined when displacements in the complete crown / skirt model are considered.
This is discussed more fully te Section 2.3 Fortunately, the pre:ise value of kog is not required, because this spring coqstant does not have a st"ong influence on the en:
4 resul ts.
Teole 2-2 suva-izes tM cal-d ate: spaia; constants fo* the two
~
s k i rt desigas.
2c
n The vertical displacement on the top of the skirt along the stud bolt circle was evaluated.as part of the finite element analysis of stresses due to stud ~1oads under lift-off conditions.
These displacements are relevant to crown. lift-off, and are presented in Figure 2-1 for both the AE and AF skirts.
A stud load of 6,600 pounds was used.
As discussed in Section 3.1.2, this value is based on experimental measurements.
A strong angular variation of skirt displacement is seen in Figure 2-1, which is especially marked for the AF skirt. The displacement is smallest over the wrist pin, as expected.
2.3.2 Crown The crown is fabricated from cast steel and mounts on the top of the skirt.
A 0.007-0.011 inch gap is present at the outer load ring at assembly.
The crown enters into the analysis in two ways; for isothermal conditions, the pressure load applied to the top of the crown is transmitted through the crown O
to the top of the skirt in a manner which corresponds to neither a uniform displacement nor uniform loading coundary condition on the skirt, and thermal distortion of the crown due to non-aniform temperatures during engine oper-ation influences the proportion of the pressure load that is transmitted across each of the loading rings.
Two seprate finite element models of the crown were constructed. One was for placing on top of the skirt and is described in Reference 2-1.
This model was nee $ed only to represent the circumferential variation of the crown /
skirt interf acial pressure, and was relatively coarse.
The other crown model was for evaluation of thermal distortion of the crown and stiffnesses of the crown due to pressure on the top and loading on the outer contact ring.
This model mast accurately predict the crown dis-placement at the outer ring relative to the inner ring, and, thus, was constructed with a ruch finer mesh.
Figure 2-2 shows the axisymmetric model that was composed of 327 elements and 425 nodes.
An axisymmetric bilinea*
displacement elewnt was utilized for this refined crown model which was run 9Ca; the MAC anaTysts program.
O 25
a 4
%.J The vertical displacement of the outer contact ring relative to the inner contact ring when the crown is subjected to service conditions is of interest in evaluating the response of the crown and skirt to pressure and temperature loadings.
The finite element model shown in Figure 2-2 was used for the calculations of crown deformation for various service loads.
The inner ring of the crown model was assumed to be supported and the displacement of the outer ring was calculated for three conditions: (i) uniform pressure on top of the crown; (ii) axially oriented load around the circumference of the outer ring; and (iii) crown subjected to steady-state temperatures corres-ponding to engine operation.
The stiffness of the crown when subjected to pressure and outer ring loading was found by div' ding the total load by the corresponding deflection of the outer contact ring. The fol'owing results were obtained:
Dressure stiffness k [p) = p A/6 = 47.4 Mps/mH g
Outer ring loading stiffness kqp) = F/6 = 16.4 kips / mil p = pressure, A = bore area, F = outer contact ring load, and 6 = corresponding displacement of outer contact ring relative to inner ring.
The vertical deflection cf the outer ring relative to the inner ring at steady-state operating conditions is also of interest.
This parameter is denoted as 6,,and its calculation reaaires information on the steady-state T
temperature distribution in the crown and skirt.
In order to gain a better understanding of the crown temperatures under steady-state operating condi-tions, the results of peak temperature measure 9ents as a fun, tion of position in the crown were supplied by TDI [2-3i.
The measurements were made with
" tem.-'ugs," which were inserted into holes drilled in the crown and provided a pass' ve measare o' the maximr temperature to w*ich tne plug was subjected.
As **: -*ed by TOI. the measurements were made in an P. s engine it 451 ROM 2-6
~
with, a BMEP of 213 psig, and closely follow the running conditions at Shore.
ham.
The peak c.cown temperature measurements reported by TDI are shown in Figure 2-3.
Note that the temperatures on the tuttom of the crown are nearly consta,nt and equal to about 200*F.
This suggests that the piston skirt is nearly isothermal under steady-state operating conditions, which implies that thermal stresses in the skirt are small.
Independent calculations of operating temperature in the piston assembly were performed by Failure Analysis Associates.
The transient radia-tive and convective heat transfer analysis utilized reasonable values for coolant temperatures, and convective heat transfer coefficients, and combus-tion gas temperatures dete e.1.9d from the indicator diagram data from Reference 2-1.
Key features of the calculated temperature field, including peak temperature and temperature gradient through the central portion of the crowq, were in agreement with the TDI measurements of temperature.
The temperatures at various positions 09 the surface of the crown shown in Figure 2-3 were used to estimate the steady-state operating temperatures everywhere on the crown boundary.
These boundary temperatures then served as boundary conditions for an axisymmetric steady-state heat conduction problem.
The finite element model shown in Figure 2-2 was used for numerical generation of the steady-state temperature distribution througho't tne crown.
This temperature field was then used for finite element calculation of thermally-1 induced displacements in the crown.
This provided the following result for i
the (downward) movement of the outer ring relative to the inner ring under steady-state engine operating conditions:
6T = 0.0106 in:h This result is used in Section 3 to o5tain estimates of the influence of thermal distortion on stresses in the skirt 39d possible lif t-off of the crown from the skirt.
O 2-7
r v) i 2.4 Crown / Skirt Interaction Information on 'the -interaction of the crown and skirt under isothermal conditions was provided in Reference 2-1 by experimental and finite element analyses that included the crown mounted on the skirt.
The finite element results are briefly reviewed in this section.
Figures 2-4 and 2-5, which are drawn directly from Reference 2-1, sunriarize the finite element results relevant to crown / skirt interaction. These figures show that there is consid-erable angular variation of vertical displacement in both the crown and the skirt in both the inner and outer rings.
Hence, the contact rings do not remain planar, and uniform displacement of the contact rings does not provide an accurate representation of the boundary conditions on the top of the skirt.
However, these figures show that the gap between the outer rings of the skirt and crown is virtually independent of angular position.
This is evident by the flatness of the dashed line in these figures, which is the dif ference between the vertical displacement on the outer ring of the crown and skirt Q(3 (6co - 63o).
The uniformity of the gap around the circumference of the pres-surized crown / skirt means that gap closure will occur uniformly around the circumference - despite the fact that the contact rings do not remain planar.
This observation is consistent with the deduction from the experimental results reported ir. Reference 2d that the g3p Closes nearly simultaneously at all angular p.isitions cr. the oute ring.
Tne stiffness of the crown when attached to the skirt and sub,Jected to
[k [p)] and the value of k c
jo can be estimated from the' results pressure
. presented in Figures 2 4 and 2-5.
The value of the crown stif fness is given by the expression c(p)
- 6 6
(2*I) g g
The value of k jo is given by the expression
.[l 6
-e 5'
So -1 3
(2-2) k to F
q o
l 2*
I
v where F = total applied force, o
= vertical displacement of outer ring of crown, co 6gg = vertical displacement of inner ring of crown, 6
= vertical displacement of outer ring of skirt, so 8,54 = vertical displacement of inner ring of skirt, and kg = skirt stiffness at inner contact ring.
The value of kg reported in Section 2.3.2 was used for the calculation of kjo.
Table 2-3 sumarizes the results of.these stiffnesses.
The value of k (p) e agrees well with the value of 47.4 kips / mil determined by finite element analysis of a stand-alone crown, as reported in Section 2.3.2.
The slight difference in the values is due to the different degrees of model refinement.
The values of k shown in Table 2-3 show a wide degree of variability in jo spite of a relatively constant value of the relative displacement.
Fortun-ately, an accurate value of kjo is not required, because this relatively large q
stiffness plays a secondary role in the crown / skirt interaction -- the inter-A action being dominated by the smaller stiffnesses k, k {p), and k (p).
g e
g nominal value of 500 kips / mil was selected and is the valued included in Table 2-2.
Section 2 References 2-1 " Investigation of Types A: and AI Piston Skirts." Report prepared by Failure Analysis Associates for Transamerica Delaval Inc. Diesel Generator Owners Group, Report No. Fa AA-Ba-2-14, Palo Alto, Calif ornia, May 1984 2-2 " Emergency Diesel Generator Crankshaf t Failure Investigation," Failure Analysis Associates Rep 3*t No. Fa AA-83-10-2 Palo Alto, California, October 19R3.
e 2-3 Private communication from Al Fleischer, Transamerica nelaval Inc. to David Harris, Failure Analysis Associates, 6 January 1994
' O 2-9
1 i
e V
Table 2 1 IEXIMUM PRINCIPAL AND EQUIVALENT STRESSES IN STUD 8055 REGION AS EVALUATED BY FINITE ELEMENTS FOR VARIOUS 800NDARY CONDITI0ftS FOR THE AE AND AF PISTON SKIRTS (all stresses in ksi)
AE.
AF Loading Condition Global Local Global Local Pressurized crown, oggg
.42.7 68.1 41.4 92.2 p
Inner ring, unif. displ., o;;;
28.5 43.4 29.4 47.7 Outer ring, unif. displ., og g y
-5.71
-3.29
-2.22 4.20 4.45 7.32 5.24 17.8 g
Lift.off, a; = omax No lift.off, Sax (l) 1.77 2.40 Pressurized crom 38.2 61.3 42.8 79.9 w
h Inner ring, unif. disp 1 24.8 39.1 29.4 43.1 8
Outer ring, unif. displ.
9.9 4.7 3.6 3.6 u.
b Li f t-of f 3.8B 6.78 5.55 19.8 (1) Equals.(9727/369,300) = (o;;; for p essurized crown).
V i
l 2 10
~-
Table 2-2 kb
'i" SLM4ARY OF SKIRT STIFFNESSES FROM FINITE ELEENT ANALYSES (all values in kips / mil = millions of pounds / inch) t.
AE AF 4
k 81.2 70.5 j
k 95.3 94.2 o
kjo=kj
-5M
-600 o
0 I-1 4
1 i
O 2-!!
]5 e
Table 2-3
)
STIFFNESSES EVALUATED FROM FULL' CROWN /5K!RT FINITE ELEMENT IODELS t
AE AF 634 - 65o, nils ave 3.8 4.7 max 4.1 5.0 min 3.6 4.6 kgo, kips / mil ave 493 685 l
max 824 1547 i
min 389 558 O
6co - Oci, mils ave 6.4 7.0 max 6.8 7.6 j
I min 5.6 6.4 l
I 3 n ava 53 g (7, ' i n '*ii k
I max 65 SA min 5:
A?
F' = PA = 369 kips 4
i a
O l
l l
2-1?
l-
. ~,
"s 4
12 i
i e i
a i
i i
e I
l AF 1.0 1
I s
e i
I l
O.8 W
AE i
I 1
I a
i l
l O.6 i
i i
I i
3 1
O.4 I
i Stud +
re-boss I
I hole l
8 0.2 i
t i
I 4
I i
0 O
20 40 60 80 ANGLE FROM WRIST PIN,6(degrees)
Figure 2-1.
Vertical displacement of skirt at center row of nodes in inner contact ring for 6,600 lb. load applied at washer landing area.
O FaAA-84-5-18
. l V
\\\\\\
s N
r fr f~;
b
_SN
- f>
II I
L
?
^
~
> :d b, i l}JL - 4
\\i I I / / //
~
\\'
lill//
~~-
\\
lill//
- \\
lill/
O L 'x.
I'll'
';\\
l
\\\\\\\\\\\\
Figure 2-2.
Finite element mesh of axisymetric crown for evaluation of crowe themal distortion and spring constants, O
+
, - ~,,., _ _.,.. _,.
L i
\\
483 681 F*
j 583 631 407 378*
662 437 i
h 388 l
\\
\\ 246 N
332 Nx s
\\
328 N
kN 202 l
205 Figure 2-3.
Results cf te plug neasuremeats et peak terperatu e as a fua.: tion of position or crewr (supplie:: by TD!).
6 O
Fe A A-84-6-18
f3 I
+ G 5
3 s-s o+ 3
$C g
e t-O e-*
u e
o e
$d
- C e
m o+ 3 m
L c
a wa
~
~
5C '
.E 1 O
eG
.g e
e 2
es a
e
.e oe o
o e
s z
4 0
08 c
e a
(
m i
a
- +
ev E, w o
o o
u o
- c. <
U m
o
+ m o
o u
oc
,e
-=
eL
-o in ww i
+
a.
3k aL F
u
+
G o' m e.
+
E 5
b5 l
C+,
C
~
~
+
i 2
- c i
i h
O Y
'E
+
.l ea j
tu j*
- l d
Ee 55 i
+
- =
i
(
t z
v i
O
{+
C
.l v.5 o o g
c I
C +
1 wa l
c e e
t+
su C
p o 1
y y i
C, c
p c
N m
- e I
g i#
c v
- i, e5, e
e-
+
ee Of E
G P
+G O
O t2 2 6 c-G
+
[
O' O-4F o o
e o
e o
n l
(sl!w) rft '1N3W30Y"IdSIC 'lV Ol16 3 A tJ 1
Fe A A-84-5-18
O e,
,e
,e t
I 5
o
+
0 0 +
d >
O i
e s-em u
L-3 IE a
cw
+ 0 0 +
d >
cY 3
ge A
5 E
Q UC ec a
,G+
(,5-
+e c
c o
o vv 3
.o o
e
+
t
-u
.u m.
w ct <
o D
e u
+
e o
.e c
c g
e
- C T9 d
I
~
- z
-L 5
o'
_U E
02 a
c
~
u 3
n s.
ac
\\
L Q o
u
+
6-M se -
m c
3 g
o o
e C+
n y
ae c
E h E.
c O+
s 1
s o
o o
0+
G e'
o O'=
O
~c M
U" t)
C M
C E~
k G
w e c-W k
eu 5
O o i
+
a 5 c e
zw G F z
b c
==
T C O O 4-Q
+
w w" d >
c sv e
"m gE a
V y d b C6 8 8 V
q g=
E O
+
w,
=
+ 0 4,
i O
i G F c-
+ e 2.E
= k c'
L I
I I
O C
I: e o
e o
e o
w e
2
.(snu) n.1N3W30Y7dS10 7VD11W3 A l
l l
i Fa A A-84 18
(ol V
3.0 CROWN /SK!RT INTERACT!fM N) DEL A crown / skirt interaction model that includes the influence of thermal distortion of the crown on the load split between the contact rings and the possibtlity of lift-off between the crown and skirt is described in this section.
Such ef fects are important because of their influence on cyclic stresses in the stud boss region under isothermal and steady-state operating conditions.
3.1 Review of Experimental Observations A review of the experimental observations under isothermal conditions reported in earlier work [3-1] on the AE and AF skirts is provided along with some new closely related results.
3.1.1 Piston Asse :bly The re.ults of strain gage measu ements at namercas locations on AE and a
AF pistons with hydrostatic pressure applied to the crown indicated that the gap closed nearly simultaneously around the circumference of the piston.
The pressure at whicn the gas closed (unde" isothermal conditions) is denoted as p*.
The observation of si,ultaneous ga; cloure aroand the piston circumfer.
ente is consistent with the finite ele-ent results presented in Figures 2 4 and 2-5 in which the gap (dif ference in displa:enent) is nearly independent of angular location in spite of the appreciable angular va istion of the individ-ual displacements.
Once the gap is closed, some load is transmitted to the outer contact ring, thereby red;cing the stress in the stud boss region below what it would be in the absen:e of gap closure.* The ratio of peak stud boss stresses with and without gas closure is of interest, because such stress reductions have an ir portant influen:e on the value of cyclic stresses in this critical region.
Additionally, experimental obse'vations of stress reduction and gap closure provide results that a*e suitable for co,parison with predic.
- The finite ele,e
- res;lts of Tao'e 2 1 s m that loa:ia; of t% oater ring Droduces only reinialal stresses in tM stad boss regtoa, 11
b l
'(
tions made by use of the crown / skirt model. Table 3-1 sunenarizes the experi-mental results of current interest.
i 3.1.2, Stud Attachnent Loads and Stiffnesses j
- The amount of preload on the studs that attach the crown to the skirt must be known to determine the possibility of crown lif t-off, as discussed in Section 3.
The stiffness of the belleville washer stack is also required.
The values of these parameters were measured experimentally, and the results will be reported here.
Two foil strain gages were mounted on the central portion of a crown attachment stud, and the strain produced by attaching a cromm to a modified AF skirt following procedures reconnended by TDI, and as contained-in Reference 3-2, was measured.
Multiplying the measured strain by Young's modulus for steel and the area of the central portion of the stud provided a stud load, Fgo, of 66 % pounds.
A Icad-displacement measurement was performed on a stack of Belleville washers assembled according to instructions in Reference 3-2 for modified AF skirts (i.e., two stccks of 13 washers each placed concave end-to-concave end, with -! centasi Plignment bushing).
The measurement was performed in a closed loop servohydraulic test machine, and the stiffness of the washer stack was foJnd to be k, = 0.15 kips / mil.
The spring constant of the stud, which is idealized as having a length equal to the length of its reduced section (3.765 inch) is equal to 2.52 kips /
mil.
Since the stud and washer stack are in parallel, their combined stiff-ness will be
,-1
)
(g
+ p)
= 0.142 kips / mil (AF) k
=
g stud w
The stiffness of the was5er stack for an AE pis*on is twice that of a modi'ied AF, because a single stack of 13 washers is used for the AE. Therefore, k, =
0.3'1 kins / nil, whic% results in a co*bined stad/ washer stiffness of a
3-2
/
(
k
= 0.268 kips / mil (AE)
These results are used in the crown / skirt interaction model.
3.2 '
- Crown / Skirt Interaction Model The finite element analyses reported in Reference 3-1 and sumarized above provided information on stress levels-in the piston assembly, and allow-ed estimates of gap closure pressures at ambient temperature., _ Additionally, the finite element results sumarized in Figures 2-4 and 2-5 suggest that gap closure is uniform around the circumference of the skirt, which is consistent with experimental observations.
However, information on the influence of thermal distortion on_ the load split between the loading rings and the possi-bility of lif t.of f of the crown f ro9 the skirt during the exhaust stroke was I
not provided directly by the finite element results.
All of these additional factors could be incorporated into finite element models, but such a model would be' a significant extension beyo9d those employed to date -- the priinary extension being the use of gas elements in both loading rings.
This would result in nonlinear effects which would require an iterative solution to an already large problem. The inclusion of thermal distortion effects would also produce additionel complexities.
a Rather than undertake additional expansions of a t h ree-cimen s i ona'.
c rown/sk i rt model, a simple engineering model that accounts for all of the important variables was constructed.
Tne model was based on the observation i
that the gap appears to close uniformly around the circumference.
This is consistent with both the experimental obseavations a9d finite element calcu-lations.
The nodel was based on the assumption that the loading rings on the crown and the skirt remain parallel to one another.
For the purpose of eval-uating stiffnesses, the loading rings were assumed to remain planar.
The stiffnesses and crow 7 thermal distortion reported in Section 2.3 provide ne:essary inputs to - the cro** fsk irt interaction model.
Tw: basi c conditior.5 -
were considered; (i) top dead center of the :c9pressic, stroke (or beginning o' powee stroke), where the rusi r compressive stressas in the piston skirt 31
.Q s
(
)
MJ
=l occur, and (ii) top dead center of the exhaust stroke, where lift-off of the
-crown may occur.
Lift-off would alter the load path for the reaction of the stud preloadi, ahd, as shown in Table 2-1, would result in increased maximum tensile stresses in the skirt.
3.2.1 Power Str# e The maximum firing pressure of 1670 psig is considereo to occur at top dead center of the compression stroke (which coincides with the beginning of the power stroke).
Figure 3-1 shows the forcts acting on the crown and skirt at this point (inertia forces are included as a reduction in the pressure).
F is the total load on the inner ring, and F is the total load on the' outer g
o ring.
Both of these loads are distributed around the circumference of their corresponding ring.
F is obviously zero prior to closure of the gap, go.
o The stud preloads do not enter into the problem unless lif t-off occurs, which is not possible under firing load conditions.
The displacement of the inner and outer rings of the skirt are given by the following expressions:
i o
01*k
- k 11 to (3-1) i o
i 6
+k
=
oo oi i
where k
= stiffness of outer ring due to load at outer ring on kgo = stiffness of oJter ring dJe to load at inner ring kgg = stiffness of inner ring due to load at inner ring koj = stiffness of inner ring due to load at outer ring The reciprocal theorem of linear elasticity (page 20, Reference 3-3) requires that kjo = kog.
The following relationships are employed to simplify the notation:
I i
ii I3~ )
k E k 31
)
1
. o
(
. Q' ~
Equation 3 1 then becomes F
F 0
- .k, 0
(3-3)
I F
t-3
. 01"Y*k i
of Force equilibrium of the crown results in the following relationship F, + Fg = pA (3-4) where A is the cross-sectional area of the cylinder.
Let 6 be the downward displacement of the outer ring of the crown e
relative to the inner ring of the crown.
- Pressure, p,
produces a downward displacement, F,
produces an upward displacement and thermal distortion results in a do.<nward displacement.
The following relationship holds 6
= k (p) + 6T
- k (F)
(3-5) g c
c where k (p) = stiffness of crown due to uniform pressure g
k (p) = WW o' '.cown de to ouw Hng losmg g
6T = charme disia' tion of crown, w'11ch is considered 4
to be known (see Section 2.L 2).
The following expressitin follows from the geometaical relationship between the displacements 6g - 6, + 6g = g, - g (3-6)
The current value of the gap is denoted by g, the initial value is go, and g =
0 when the gap is closed.
The following parameters' are considered as u%no-ns: F,F,6, 6, g, o
g g
4 and 6 However, if g d 0 then Fo = 0, and vi:e ve ss.
Tnerefore, there are O.
g
.five unknow15. he five epations necessa*y for a so?
- r are gi een by Sua-
.i
)
3-5.
)
8 I
l V
1
' tions 3 3 (two equations), 3-4, -- 3-5, and 3-6.
Under ambient conditions with no pressure, the gap is equal to its initial value, go.
- The pressure at which the gap first closes at roon temperature (hence 6T = 0) is of particular interest.
This pressure is denoted as p* and is given by the following expression (6T = g = F, = 0)
'o
'o (3-7) p* =
=
3 3
3]
pA [
+ -]
3 3
A[
+-
c(p) i io c(p) i Tne notation kj = [(1/k ) - (1/kjo)]-1 has been introduced for simplicity.
g The forces on the load rings once the gap has closed are of interest, because the load " split" influences the stresses.
Taking g = 0 (because the gap has closed) and using the five simultaneous equations leads to the follow-ing expression for the force on the outer ring
(
6
-g
+ pA [
+1-1 c (p) i io o'
1
. f + /
,2
,(F) c o
i oi (3-8)
- EA [k
]
6T*90 c(p) i 1
1 1
- F P
k (F) c i
o The notation k' = [(1/k ) - (1/kgo)]~I has been introduced for simplicity and o
to f acilitate co,pariso, with experimental results.
to F - is useful in comparisons with experimental obser-The ratio of F j
n vations.
Using Equatio9 3 a to obtain an expression for F in terms of p and j
F and.iividing this into Equation 3-9 foe F prwides the follo ting result 9
g (for D > p*):
36
pA[p
+
+6
~9 p
T 0
1=
(3-9)
F 1
1 1
pA[k (F) i k,1 - 6
- 9 k
T 0
c c(p) o Gap closure pressure and load splits calculated by use of Equations-3-7 and 3-9 in conjunction with finite element stif fness values are compared with (room temperature) measurements in Section 3.3.
3.2.2 Exhaust Stroke The analysis to determine w'1 ether the crown and skirt separate on the inner ring durir.g the exhaust stroke follows procedures similar to those p
employed above. Whether or not such "lif t-off" occurs has an important influ-ente on the peak tensile stresses du-ing the exhaust stroke.
The crown is attached to the skirt by foJr studs, each of' which has a static preload of FBo, as discussed in Section 3.1.2.
Figure 3-2 shows the forces acting on the crown and skirt, where now a. separation of 6 on the inner load ring is con-g sidered.
The separation distance, 6, is taken to be inde:endent of angular L
position, which is consistent with the assumptim thn Jic loading rings remain parallel to one another.
This assumption underestimates the defor-mation of the skirt under actual stud loading conditions,.because, as shown in Figure 2-1, deformation of the top of the crown is far from uniform under stud loading conditions.
Additional possi5ilities are lif t-off over part of the inner contact ring at a given angular location, or over a part of the circum-ference of the loading ring. Rotation of the inner crown ring relative to the inner skirt ring, resulting in stud bending, could also occur. The assumption of a uniform 6 will lead to conservative predictions regarding lift-off, g
i.e., lif t-of f may be predicted under situations where it does not actually occur.
Additional comments along these lines are provided in Section 4.1, where displaceneits based on the model are compared with the finite element results fe,r stJd Ioading.
37
O Once a lif t-off distance of 6 is present, the force in a single stud g
is given by:
(3-10)
FB=FBo + k,, 6L where k, is the stiffness of the stud / washer Combination *.
g For the purposes of estimating skirt deformation, assume the bolt loads to be distributed around the inner load circle.
By use of Equation 3-3, the skirt distortion is given by the following expression:
F F
)=4
+[
(3-11)
)+F(
6, = 4F (
g i
of o
of i
o Force eqailibrium of the crown provides the following relationship 4FB = F, + Fg (3-12) d g is the crown inertia force (see Figure 3-2).
The distortion of the where F crown due to the force F,is given by (see Equation 3-5)
F (3-13) 6 (p) = g (F) c c
Displacement compatability provides the followin; relationship 6 + 6, = 6T~90 F)
(3-14)
-6 Recall that 6 is positive upwsed, and 6 will be decreased by a larger gap, a L
g larger skirt distortion, and a larger upward outer crown ring displacement due to F,.
A larger thermal distortion (6 ) will increase lif t-off.
T l
- The stud is assumed to be rigidly attached to the crown, and inner-ring c rowe, distortion and "th*007 thickness" skirt displa e9ent are ass med to be Q
negligible.
3-R
o 1
. '%)
o(p)].
Equations-There are now five unknowns [F ' 'L, o, F,'and B
g o
c 3-10 through 3-14 provide the five equations necessary for a solution.
This system of equations provides the following results:
, 6((1 + w) = 6T~90 k
F,
~
g sw (3-15) g g
+ F; [g
+7g3]
c(F) o oi The constant w is defined by the following:
5" [ c(F) + d + d 2 ] = 4ksw [k
+ d + d]
(3-16)
I w = 4k i
o og c{p) k,!
kf k
Once 6 is known, the stud load is given directly by Equation 3-10, and the t
force F, is given by the following expression obtained by combining Equations 3-10 and 3-12.
F, = 4(FBo
- sw L (3-17)
~
Positive values of 8 and F confirm that lift-off has occurred.
If g
n lif t-of f has not occurred, then the results of Section 3.2.1 must be used if it is desired to estimate Fj and Fo (wit
- an appropriate valae of p to account for the inertia force).
For a given set of conditions, the minimam allowable gap to preclude lif t-off (denoted as gg) is obtainable from Equation 3-15 by setting ( = 0.
This provides the following expression F,+F[g
+[g I]
I g* = 6T-g y
sw c(F) o 01 (3-18) l
+ ll T
k,FBo + F [A
=6 I
k-g c(F) g Lif t-off will occur if 9, < 93.
Using tne relationsn's fo* 93 given by Equa-tion 3-19. Equation 3-15 can be rewritten as follows:
3-9
l
.- (O, U
9"* - 9 o
(3-19) g=
Comparison of gap closure pressures and load splits as observed experi-nentally and as predicted by the crown / skirt model are presented in the next section.
The possibility of lif t-off and the effects on cyclic stresses are discussed in Se-tion 4.1 where cyclic stress magnitudes are estimated for piston skirts at ambient and steady-state operating temperature conditions.
3.3 Comparison with Experimental Observations The gap closure pressure and load split between rings can be evaluated from the crown / skirt interaction model using the spring constants evaluated by finite elernents.
The gap closure pressure, p*, can be evaluated by means of Q
Equation 3-7.
The ratio of peak stress with closure to the corresponding value in the absence of closure can be evaluated knowing the ratio of forces on the outer and inner rings, as given by Equation 3-9, assuming that the peak stress is governed by the load on the inner ring.
The results of Table 2-1 show that this assumption is a good approximation, because the stresses for a given load that is applied by a uniform displace 9ent on the inner r'ng a e much higher than corresponding values for loading on the outea rir 9 T ie ratio of stresses with and without closure is given by the following ex? es-sion:
closure i
1 (no closuee), F, + F (3-20) 1
}.
1 The value of F Fj is obtainable from Equation 3-9.
o Table 3 2 su narizes the calculated gap closure pressure and stress ratio; the nominal, minim and maximum values correspond to the gap values shown ir Table 3-1.
A ceparison of Tables 3-1 and 3-2 reveals good agree 9ent
\\
betwee$ the resuits predicted by the crowe/sdet intera: tion nodel and the 3-10
Ov experimental observations.for the case of the AE piston.
The load split pre-diction for the AF skirt is also good, and the gap closure pressures agree withirt about 20"..
The comparison between the experiments and calculations can also be made by estimating skirt stiffnesses from experimentally observed load splits and gap closure pressures and comparing the results with the finite element stiffnesses. Table 3-3 summarizes the results of such comparisons, which show better agreement for the AE than for the AF skirts.
Experimental values of the outer skirt stiffness (k') are generally lower than calculated.
- However, the finite element values fall within the range of results calculated by use of experimental observations.
Overall, the agreement between the various sets of data is quite good, which indicates the validity of the crown / skirt interaction model.
Fortun.
Ox ately, as shown in the next section, the predicted cyclic stresses and crack growth are not highly' dependent on the accuracy of the predicted skirt stiff-nesses.
The major value of the crown / skirt interaction model is its ability to predict cyclic stresses in a piston skirt at operating temperatures, which is discussed in the following section.
Section 3 References 3-1.
" Investigation of Types AF and AE Piston Skirts," Report. prepared by Failure Analysis Associates for Transamerica Delaval Inc. Diesel Generator (hrners Group Report No. Fa AA-84 2-14, Palo Alto, Calif ornia,
May 1984 3-2.
Instructions for AF Piston Skirt Modification contained in letter from T01 to LILCO referenced as part of Stone and Webster Engineering Corporation Engineering and Design Coordination Report No. F-38313, December 3.1991.
3-3.
S.P. Timos5enko and J.N. Goodier, Theory of Elasticity, Second EJition, McGraw Hill Book Comoany, Inc., New Yora, 1951.
O 3 11
Table 3-1 SumARY F EXPERIMENTAL (BSERVATIONS RELATED TO CROW 4/ SKIRT INTERACTION (From Reference 3-1)
AE AF l
Gap, go, nominal 7.5 8,0 mils min 7.0 7.5 max 8.0 9.5 Gap cicsure pressure, p*,
nominal 1000 800 psig min 820 700 max 1050 1000
' closure /8(no closure) at nominal 0.88 0.80 1670 psig min 0.75 0.92 max O
3-!?
u-.
- -,. -.. -. ~. _ _... _ _ _. _. - - - - _ - -.., _ _ _. ~
9 O
V Table 3-2 GAP CLOSURE PRESSURE AND STRESS REDUCTIONS DUE TO GAP CLOSURE AS PREDICTED FROM C40dN/SK!RT MIDEL AE AF Gap closure pressure, p*,
nominal 1040 1050 psig (Equation 3-7) min 970 980 max 1110 1240
' closure /8(no closure) nominal 0.85 0.85 at 1670 psig (Equations min 0.83 0.83 max l
0.R7 0.89 O
3-9 and 3-20) l 6
r i
31':
- m.
3
('
()
N
~
I
. 1, Table 3 3 CIBIPatl%04 E SElei Siltrarssts As tympairs renst tapfelsEnTAL an%revailas Afst Olnum/58IRI imitRACling sting 1 WilN CDnel%PilanteG finlIt iLtsu RI v4LWI5 At AF Crumments k[. tips / mil nominal R 3. 7 44.4 Evaluated by use of Equation 3.F.
min 4%.7
?%.9 using tgrog = 47.4 ktps/ ell and z
mas I?!
M t. F various Ahterved P* and g, from g
Table 3 1.
n;.tfps/=Il nominal F.95 11.9 Evaluated by vse of Equation 3 9 and 3.?n ustaq k, from ahove, tcff)
- t..
,ge 74.0 R.0 g
16.4 tips /mtl amet various obterved W
mae 7 59 )
=
Ioad spl4ts and q, Irom table 31.
n E
t 41.7 70.5 From Table ?.?.
g g
t, 95.1 94.?
w k
Sml Sm1 go aw t{
96.9 R?.I From k. k,. aM k,
g g
w g
t*
119 116 s
e
~
' 1 1
1 l
l P
I m
r dl MI dI h
F F i Fi Fo o
Fo Fi Fl Fo o
e o
o b
~
~~~~~, _ _.,.,s._,,~
o
-~
6, 6,
- k..-
i i
Figure 3-1.
Sche.atic re:resentation of loads and displacerents en inner and oster contact rings of Crown and skirt.
I i
w-e, - -
-,,w
.y----
p-
-+e-y
-,-+--y y-
--- -- + - - - -
i 4
i 1
d I Inertia force, F
r g
NN N.
,. 7 s.
\\
p' y
r
/m, s
I
/
]
l.\\.
s,'
t.
C
\\
y y-a A
' F, 6
YF g
B (4 pl:s.)
F B i
E A
o 6
l l
T v
l 1
/
i i~
i Fig.** 3-2.
F:':ts a:tir.;
caca. ar: 5+tet et 100 dt!
Ot*te" :# es*t.it st*069, 1
s
4
.f 4.0 FATIGUE AND FRACTURE ANALYSIS Previous work reported in Reference 41 included analyses of fatigue crack initiation, propagation, and arrest in AE and AF piston skirts under isothermal conditions. The cyclic stresses employed accounted for gap closure and load split in the At piston by linear interpolation between results for the two gap sizes employed in the testing of the skirt.
In this section, a similar analysis is performed, including the influence of thermal distortion of the crown.
The crown / skirt interaction model is used to predict the load split in pistons with isothermal and steady-state operating temperature dis-tributions, and these results are then used to adjust the stresses in the crown / skirt model to account for the load split and possible crown lif t.of f.
Once the cyclic stresses have been calculated for various conditions, the possibility of crack initiation, growth and arrest is analyzed by fatigae and fracture nechanics procedures that are described in Reference 4-1.
4.1 Cyclic Stresses The finite element stresses for the complete crown / skirt finite element model and for lift.of f, as summarized in Table 2-1, form the basis for esti-mating the cyclic stresses in pisto95 wit') isothermal and steady-state tenperature distribations and various initial gas sizes.
These base line stresses are adjusted by use of the crown / skirt interaction model to at:3vnt for thermal distortio9, gas closure, and other operating vaaiables. Referen:e 4-1 shows t*iat the finite element results are generally conservative relative to experimental isothermal stresses.
Hence, the follo=ing results are also conservative.
In order to encompass the load splits observed experimentally, cyclic stresses are evaluated using the range of skirt stif fnesses estimated fro 9 experimental observations.
The skirt stiffnesses evaluated by finite eleme9t calculations are also co9sidered.
Hence, the foar sets of kj and kf shoni in Table 3-3 for ea:N skirt design are considered.
Table 4-1 p*ovides a sum may of the ecsalts, e,3, for 89 i sothe r-'dI Di st oa, is alws/s ea;31 to g for n:
!ift.o f.
e fo* a c'stoe wit 9 ste3dy-state te7 e ature disteiostice r
gg 41
-q 1
i depends upon whether or not lif t-of f takes place.
A value of ( = 0 implies no lift-off.
, The results of Table 4-1 shoa that the cyclic stress amplitude is lower in the case of steady-state temperature distribJtion than in the isothermal case eve 9 when lift-off is predicted to occur.
He9:e, isothermal operating conditions are generally more severe than steady-state conditions.
A smaller gap always results in a more favorable gin, but may produce a less favorable because of the possibility of lif t-of f with a smaller gap.
The results amas of Table 4-1 snow that lift-off is almost always predicted for an initial gap of 0.007 inch, but never for a 0.011 inch gap.
A comparison of the predicted lif t-of f dista9Ces with predicted skirt displaceme9ts under stud loading is shown in Figures a-1 and 4-2, which pre-sents the results from Fig;re 2-1 alo99 with correspondt9; predicted values of for the two ty:es of pisto9 skirts. The values of ( are calculated 69 and 6L using the skirt stiffnesses fro
These figures show that the skirt d.eforms more unde washer la9 ding loads tha9 is pr?dicted by use of the spring constants for uniform innea ring displa:ement.
This is not uneupe:ted; the stiffness o' the ski-t under c,i'orm displa:ement loading is dominated by the relatively very stif f r:3 0 o..
tte wrist pin.
Figares 3-1 1
and 4-2 shoa that the skirt can def o-m suf Ticie 'ily to pre:lude lif t-off whe9 a 0.009 in:h initial gap is present.
However, the ca':alated lif t-of f for a 0 007 inch gap is mu:h large' tha9 any skirt displa:e'e9t due to stud loadin;.
Therefore, it appears that Ilft-o'f is linely to o::;* =1t9 a 0.107 i n:h ga p,
but not for a 0.039 inch or large gap.
Also, the la ;e washe-landing dis-nlacements for the AF means that this skirt can more closely folloa the crow 9 at top dead center of the exhaust stroke, and li't-of f is, therefore, less likely than in the AI.
4.2 Fatigue Crack initiation Analysis Tee cy:lic stresses in pisto s'iets w a isst e-s' and steady-state i
l temperat;re dist ibJtio95 p'ese9Ie$ 19 ?SO'?
t*'* I"O E0*D'"II *iSN $hI f0I4 5A crack initiatio9 c* iter:a Fro-Re'erence 4-1 to assess t9e possibility o' cre:ks initiati9g in AE am! At pisto* skirts unde
- a va*iety o' conditic s.
42
Figures 4-3 and 4-4 show the results for 0.007 and 0.011 inch initial gaps for isothermal' and steady-state temperature distribution conditions.
Two results are shown for each set of conditions. corresponding to the minimum and maximum values from Table 41.
Allowable stress envelopes for maximum and minima, yield, strength values from Table 2-2 of Reference 4-1 are indicated, which is the same procedJee as employed in the earlier report.
Figares 4 3 and 4 4 show that conditions for crack initiation are more severe under isothermal conditio9s and with a large gap. Even though lift-off is predicted for steady-state operation with a 0.007 inch gap, the resu* ting cyclic stresses are less likely to initiate cracking than with a 0.011 inch gap which does not esperience lif t-off.
Figure 4-3 shows that cracks might initiate in AE skirts under certain conditions.
In contrast, Figare 4-4 shows that cracks are predicted to initiate in the AF skirt under isothermal condi-tions for any gap size co9sidered, a9d initiatio9 might occur under cyclic stresses corresponoing to steady-state operation.
These observations are consiste9t with those reported in Reference 4-1 for isothermal conditions.
Although fatigae cracks may initiate in the piston skiets, these cracks will not necessarily propagate, becaase they would grow into a region of decreasing stresses.
4.3 Fatigue Crac'.,.r4 Analysis Tne results of the previous se: tion reveal that cra:ks may initiate ir the stud boss region of AE and AF piston skirts.
This se: tion analyzes the cracks to determine if they will groa, and if so, whether they will subse-que9tly arrest.
Arrest is likely to oc:ur be:aase o' the ve*y steep stress gradient in the stud boss region (see Figare 4-1 of Reference 4-1).
Fatigue crack growth analyses were performed for the cyclic stress co9ditions included in Table 4-1 by procedures des:rtbed in Re'erence 41.
The normalized stress gradient in all cases was take9 to be the same as that shown in Figure 41 o' Reference 4 1.
Tne only modification to eaalier pro:ed;res was to considea a co9bi o tion of stresses corres:09di93 to isothe*W 493 steady-state co* di-tion Cal:ulatio95 were first perforsed f cycll; st' esses under isothe-.a*
3 D:ssible staess eedistaid;;4 0-dae t: yiel ii e ; u n :3mside*et.
r> + +
4-3
Fatigue crack growth due to the redistributed stresses was analyzed. This was followed by imposition of stresses corresponding to steady-state operation.
In some instances, cracks could grow deeper under steady-state conditions than isothermal, because chas increases if lift-off occurs.
Following proceda*es in Reference 4-1, two sets of calculations were performed, one for nominal tensile properties and one for worst-case tensile properties.
Cracks were predicted not to propagate in the AE skirt under any condi-tion considered. In contrast to this,~ cracks were predicted to grow in the AF skirt in certain cases, but always to arrest.
Table 4-2 summarizes the crack growth and arrest results.
In contrast to the results preseited in Reference 4-1, cracks were predicted to grow in AF pistons with gaps within the T3*-
specified range of 0.007 - 0.011 inch on assembly.
The cyclic stresses as a functio 9 of gap were evaluated in Reference 4-1 by a procedure that differed from the crown / skirt interaction model employed here.
This resulted in dif ferent values of the e:timated cyclic stresses.
Table 4-? shows that, if cracks are predicted to propagate at all, they will arrest at depths ranging from 0.10 to 0.49 inch.
The predicted depth of arrested cracks is larger than in Reference 4-1 becaase of the different mea s of estimating cyclic stresses and the Consideration of lift-off betweeS t%e skirt and the crown.
Observed crack depths reported in Re'erence 4-1 eae in the range of 0.10 to 0.30 inch.
The values in Table A-2 witm a r. :
- t"
.t lif t-off bracket this observed range.
As discussed eaalier, the cro afseirt interaction To el tends to ove predict lift-off, whicn wo,ld lead to ovea.
predicting 9nat. This, in turn, woald ove* predict the arreste: crap de:tt.
Overall, cracks are always predicte: eitner not to geo-or to arrest at depths comparable to obseaved values.
Hence, the earlier conclusions [4-!]
i regarding the integrity of the AE a9d AF skirts under isothermal condt.io's are also applicaole to steady-state ocarstion.
sec
'1 1
l' Section 4 References 41
" Investigation of Types AF and AE piston Skirts," Report prepared by Failure Analysis Associates for Transamerica Delaval Inc. Diesel
- Generator Owners Group, Report No FaAA.84 214, Palo Alto, California,
- May 1994 Z
f i
j 4
l 6
}
l i
i l
46 j
.--.-,.__,.,...._,-,-..n.
..,-...n,-.,
p-
.a l
Table 4 1 CKtIC %ftf 5W5 le af M M Pl%fes Saltf5 MR 191iMeasut ANR %ftAnf.5TAlf (2 5IIInt$
AT at
>...................................... \\...........................................
Buteg T,f.
noe6 mal eIa seea f.I.
neelnaI eln mee 9,
I
(............
64.1
...........)
(.............97.7............)
g, Ieff.eff I
(....-.......
F.17...........)
(.-..........
IF.A............)
no 1itt.off I
(............
I.FF...........)
(............
7.43............)
4j
?
96.9 R t.1 45.7
!?l 27.I 44.4 75.4 A l.1 b*
7 Ita 74.n F.9%
741 Ilh 11.9 R.0
=
- a, e
8 i
F 1.999 1.401 0
7.len 7.n77 1.17e n
7.179 g
9
- n. I'.7 n
n n.74%
n.Il4 o
n n.7%s il n
n n
n n
n n
n 4
7
. % F.%
.49.7
.En. )
.%R.4
.74.0 77.4 7% 4
.74.4 g,dhermal he 9
.47.1
. A.*.9
.s.7.n
.61.4
.47.7
.F4.1
. 7 ?.4
.AI 7 II
.u 7
.%.?
.h t.n
.64.1
.44.1
.al.?
19.9
.ne.1 e,,,,
5 F
11.n
.47.7
.%n.9
-11.%
.41. F
.49.4
.63.4 31.5 steady-state 9
- 17.4
.4%.9
.%7.1
- 54. 4
.49.4
.% ).1
.6%.9
.44.9 II 47.7
.41.1 54.4 41.6
.%%.9 5 R.0
.6 R. I 51.4 e
6 F
F.'e l F. 74 1,71 F.16 14.4 14.1 7.46 IR.4 sN,e.ty-state 9
f.17 1.11 1.11 1.40 17.4 7.4n 7.4n 11.9 1I I.1F l.11 I.!!
I,11 7.4n 7.4n 1.40 7.40 aTI vieues In bl. g, anOg in mih unig eg :
- 1. Fra= Tahle 7.l.
- 7. f rne f ehle 1.1
- 1. E rwe f ersat ion 1 1%.
- 4. f reue f. pat ings 1.M and 1.9 wit h A,. n.
%. Iress Iseat tces ).7ft and 19 eit 4 8 - 11.6 alIs.
g A59.ats g, with no Ottt.aff if 8 - n. adlesst ed f rna flatte *Icaeat g with lif t.nf f If A > n hy 3
g w.eaq f +e et enn 1 3n to deter =ene s t.et t=*u lead an<f rat inta9 hy Eq /fmo <ount for t=<reased stwf I. 4.s er the watee me.f f r f easte element tale eolat snm.
L 6
a O.
i j.>.
3.F 3 e
ee ee e I
,a
~R3 f
e e e e ad
&ia i
6 t
e
=
l lg e
s a
i 4 1 88a i
~
C.
C.
u2*
d i
4#
.la**t
-2 3,
4 R..
k..
- g=,4 e
c g~1 d
c' e
w 3
m 3 ::'
R
,i 3 7. E 6
g E
c' c' e' e' e c'
g-e 6
er s
v 5
U 5
l e r
-R-g g
.. c e c.
e e c
~
c' c' e' d d c' I
e.
.e A
I t
2 l
T g
1
~,=
~,=
i
-. =
k
~, =
G" 4
{
"I la 5
wg
~
=
1 1
a i
-I t
Es
.I.-
s I
E e
3 5
E *.
a ado
- A di1 Cae sdO-A dil EletBS Cd
r to 2.32 mils g
1.2 6L for 7 mil gap Edge of boss 1.0 I
e 6
S t u d h o le -e --.-
i i
i I
t e
O.8 a
i I
t I
I e
1 7
g i
e i
e
'i 0.6 0.44 mils i
w t
6C I
... _l...
i e
l L
I 6L for 9 mil gap O.4
' _ _ _ _1_ _ _ _ _ _ _ l_. -.. _
.!..E _4 r ik, g
0.2 p i
i 4
i i
i 0
O 20 40 60 80 ANGLE FROM WRIST PIN,6 (degrees)
Fi;/e 41. Lift-off distaa.:es cor:arei wit *, inaea rir.; skirt ciscla:ene ts cse to washer loads for the AE cistoa shirt.
Fa A 4 -44 5. it
r-~
4 to 2.35 milt n
Edge of koss n
6L for 7 mil gap 8
1 Stud hole ~
i 1.2 l
s I
i e
4W$
e r8 [I i
f f
0.72 mils !
1.0 i
l 6
i I
f 0.8
?
Y<
i
/
i w
i
(
l t
0.6 e
j
--.....l.----y.y.. -....
6L for 9 mil gap
/'
l 0.4 I.
p
-. 4Fg /k g 1
0.2 t
0 0
20 40 60 80 ANGLE FROM WRIST PIN. 6 (degrees)
Fig.ee 4 2.
Li'*-O'f di$ tat:e5 C3*;3*f a't* inne" ring $ birt c15Clacements d.,f 10 w35*te ICads for the AE Cintor. Skirt.
(
l.
I FaA A*84 S* 18
r
- t
- 6 0 m' So Temperature m
W a
7 isothermal E>. O a
7 S t e a d y-s t a te c
11 isothermal
- 40 9 e e
11 Steady-state o
/.
_. 4 O
Man 0')*.+*'
Y e
20
~ M,n o,
y a
e i
i i
-80
-60
-40
-20 0
20 40 60 80 ME AN STRESS, om (k si) r ;, c :.3,
$; cess s 3,es f:- At ; ;;: skir; for v3,ie. 5 conct:iens p13;teg O' 73?' C' e**.?a!!'t l'.r(15 a-**4t.:e at a fw' tica O f Ec:a. Ltress.
l Fa A A.84 3 18
r :-
- 60 N go Temperature
$e e,
7 isothermas yj 4
11 Steady-state C_
g c
7 lsothermal
- 40 38
~ q e
11 Steady-state a.
p 4
Y s %.,.****'.**
o Maa o
- 20
\\ Min O ys s
i i
i n
i i
i s
-80
-60
-40
-20 0
20 40 60 80 ME AN STRESS, om (k si)
Fi g,*t 0 0 1 rest sta 15 f:# a~ c'1t:r shirt #:* va*10 ! ::*: ti: 4 :,';*,*t:
- ;*a:' C 4 ' * := a : *. 6 $1*fli a ;'1*.de 41 a fv *tt:r of mea
- Stet 15.
a F a A A.8 4.$.16
r 5.0 CONCLU$10NS A crown / skirt interaction model was developed that provides estimates of thh influence of thermal distortion of the crown on cyclic stress levels in the stud boss regions of AE and AF piston skirts.
This model extends results beyond those previoJsly reported, which co9sidered only isothermal conditions.
The crown / skirt interaction model utilizes crown and skirt stiffnesses evalu-ated by finite element calculations.
Comparisons with experimental observa-tions showed ge9erally good agaeement with the model.
Lift-off of the crown from the skirt was predicted for gap sizes less than aboJt 0.009 inch at steady-state operating temperature. This lif t-of f alters the cyclic stresses.
Calculations of cyclic stresses unde
- isothermal und steady-state oper-ation were made using a range of stiffnesses encompassing experimental obser-vations.
Lif t-off was co9sidered in cases where it was predicted to occur.
Crack initiation and propagation a931yses were performed by procedures followed in earlier isothermal analyses.
The conclusions obtained earlier regarding cracking of the A! and Af shiets were unchanged, but now the arrested crack depths predicted for the AF are in better agree 9ent with field observations. Ove*all, the earlier co9clusions regarding the integrity of the At and AF skirts a*e also unchaaged.
These conclusions are that cracks may initiate DJt will not propagate in the cE, 89d that Cracks will initiate and may Dropagate in the AC.
However, any cra:h s in tne AF are paedicted to arrest at de;;hs less than 0.6
- inch, which is compa atle to field observatio95.
s.1
~1 e-APPENDIX A Component Task Description A-1
r--
4 t-DR 03 341A 1 P!sTONS Classification A PART NO. 03 341 A Completion 03/05/84 PRIM 4RY FUNCTION:
The pistons react to the cylinder firing pressure and pro.
vide a reciprocating mechanism for converting co9bined inertia and combustion pressure forces into mechanical torque through the wrist pin, connecting rod, and crankshaft.
FUNCTIONAL ATTRIBUTES:
1.
The piston crown must have suf ficient strength to resist the high temperature and pressure firing loads.
2.
The load transfer between the piston crown and skirt structure must not prcduce alternating stresses sufficient to cause failure of the skirt.
3.
The wall structure of the skirt must be resistant to pressure. induced deformation which could result in skirt fatigue in proximity to the stiffening ribs.
4 Preload in the crown studs msst be sufficient to preclude failures of studs / nuts / washers.
5.
The pistri sk.irt must provide a suitable sliding surface against the cylinder liner.
6.
The pistri ring groove east be sufficiently wear-resistant to provide su "Icient ring life.
58ECIFIED STANDARDS: None EVALUATION:
1.
Determine the historical evolution o' the AC. AF.90di f i ed. AM, AN, a91 AE piston designs, including casting, heat treatment, dimensional, and material changes.
2.
Determine maximum firing pressures and temperatures for 0$R-49, 050V.
16 4. DSRV-12-4, and DSRV.20 4 designs.
3.
Develop finite element models for AF.nodified and AE pistoa designs with pressure loading (static conditions).
4 Conduct the rm3/mechani c al analysis to determine tneamally. induced load transfer due to crown disto *tio9 5.
per'.m ret tilaag'ce? esa-inatisa e' fricture AF cision skirts.
A.?
S e
t 6.
Perform eddy current examination of AE piston skirts from TDI DSR-42 and R-5 engines, and Alaska stationary diesel generator.
7.
Conduct fr Icture mechanics analysis of possible crack propagation in AF.
modified a,id AE designs with differing stress conditions.
- 8.,' Conduct esperimental static isothermal stress distribution test on AE skirt.
9.
Evaluate the effect of piston skirt loading on wear.
10 Perform LP and eddy current inspection of SNPS AE pistons following 100 hours0.00116 days <br />0.0278 hours <br />1.653439e-4 weeks <br />3.805e-5 months <br /> at 100% load.
- 11. Assess the similarity of the AF-modified AH, and AN piston designs.
- 12. Complete report on AF-modified, Art, AN, and AE pistons.
- 13. Review information provided on TER's 0-159, Q-194, 0-203, 0-310. Q-326, 0-335, 0-339, 0-393, 0-412, 0 413, 0 419, and 0-422.
REVIEW TDI ANAlfSES:
1.
Examine TDI strain gage testing (static) on skirt stud boss region.
INFORMATION RE011 RED:
1.
TDI drawings for AN and AE designs including studs, Belleville washers, 7
preload, and material specifications.
2.
Historical infomation on casting changes, heat treatment changes.
3.
Maximum cylinder firing pressure and tenperature for DSR-48 DSRV-16 4, DSRV-12 4, and DSRV-20-4.
\\
A-3