Sei sulla pagina 1di 4

ELSEVIER Materials Science and Engineering A234G236 (1997) 393-396

Calculation method for the fatigue limit of parts of case hardened


steels
H. Bomas *, P. Mayr, M. Schleicher
Stiftung Institut fir Werkstofftechnik, Badgasteiner StraJe 3, D-28359 Bremen, Germany

Received 31 January 1997

Abstract

Based on the weakest link concept a method is developed, from which the survival probability of every surface and volume
element of a case hardened part, which is loaded near the fatigue limit, can be calculated. Prerequisite of the calculation is the
knowledge of the hardness and residual stress distribution, the surface roughness, and the surface oxidation depth. By
multiplication of the survival probabilities of neighboured elements the survival probability of a limited region or of the whole
part can be calculated, which includes a fatigue limit determination. It is shown, that this method can be applied successfully to
unnotched specimens of case hardened steels. The necessary calculation parameters can be gained from a set of reference parts.
Because of the possibility to formulate a survival probability for every volume and surface element, there are no geometrical
restrictions to the parts which shall be calculated. 0 1997 Elsevier Science S.A.

Keywords: Case hardening; Fatigue limit; Calculation; Weakest link concept

1. Introduction 2. Calculation principles

The here presented calculation method for the fatigue The failure of casehardened parts under loading near
limit of parts of case hardened steel is based on the the fatigue limit is a problem of crack initiation. This is
weakest link concept which was developed by Weibull the condition under which the weakest link concept can
[l] for the strength of brittle materials and later trans- be applied to describe fatigue failure. Two main sites of
ferred to fatigue behaviour by Heckel and co-workers fatigue crack initiation and different reasons have to be
[2-61. The principles of this concept were also applied considered in case hardened parts:
to the calculation of the fatigue limit of homogeneous l crack initiation at the surface (A) due to surface
materials by several other authors [7-lo]. Only a few roughness (R) or due to surface oxidation (SO)
people showed the physical nature of the weakest links l crack initiation in the volume (v>
and the correlation between their size distribution and According to the rules of probability mathematics,
the fracture probability [ll]. The application of the the survival probability P, of a case hardened part can
weakest link concept to inhomogeneous materials like be calculated as the product of the survival probabili-
case hardened steel is new and needs additional devel- ties of the surface P,(A) and the volume P,(V):
opments considering different crack initiation mecha-
nisms and the role of local strength and residual ps = Ps(A).Ps(V (1)
stresses. The weakest link concept is based on the assumption
that material strength is determined by bulk or surface
imperfections which are equally distributed in the
stressedregion. Fracture is caused by the weakest link,
* Corresponding author. Tel.: + 49 421 2185350; fax: + 49 421 which means the worst imperfection with respect to
2185333; e-mail bomas@iwt.uni-bremen.de crack initiation. Thus, the fracture probability near the

0921-5093/97/$17.00 0 1997 Elsevier Science S.A. All rights reserved.


PIISO921-5093(97)00159-7
394 H. Bomas et al. /Materials Science and Engineering A234-236 (1997) 393-396

Table 1
Specimen data (cs = surface carbon content, xc = case depth)

Specimen Material charge Rz (w) xso (w-4 cs (“x3) -yc (mm) Loading Fatigue limit (N/mm-*)

Experiment Calculation

A I 7.8 5 0.70 0.8 RB 823 Reference


B 1.8 0 0.66 0.8 1011 962
V 2.2 0 0.70 0.6 1032 1030
w 1.6 0 0.82 0.6 1045 1009
X 1.4 0 0.55 0.5 1007 Reference

II 1.0 0 0.8 0.59 PP 501 469

Hl 0.65 0.9 PB 473 490


H2 III 2.0 0 0.96 1.5 R=O 593 Reference
H3 1.06 1.5 617 553
VE IV 1.0 0 0.9 1.0 PP 586 536

fatigue limit is not only a function of the load level but 3 A4


also of the size of the stressed region. Therefore, every (6)
a=2’l-M
volume element A V and every surface element AA has a
Now, the equivalent stress amplitude for unnotched
survival probability which depends on the size of the
specimens under uniaxial load with the local stress
element, the local equivalent stress amplitude gDv, and
amplitude ga and stable principle residual stresses grl, oE2,
the local zero mean stress uniaxial fatigue limit gw. It can
and or3 can be written as:
be described as follows:
Surface: g,,V = ga + hf(a, + grl + flr2 + a,,) (7)
P,(AA) = 2 - (A~i.%)(qm’qva ImA (2) M can be deduced from the local Vickers hardness HV
and a parameter H,, which has to be fitted from reference
Volume:
specimens:
Ps(AV)=2- (A V:VO)((~DV,‘~V
VI” ”
(3) M = (HV - E&)/867 (8)
This is principally a Weibull description, but has the
The local fatigue limits are, following Murakami [14],
advantage that it works with fatigue limits gWA and a,,
supposed to be proportional to the Vickers hardness plus
according to a fracture probability of 50%. A, and V, are
120 HV:
the area and volume, which the surface fatigue limit gWA,
Surface:
and the volume fatigue limit cow”, refer to. For easier
calculation, values of A,, = 1 mm* and V, = 1 mm3 are oWA = FA (HV + 120) Y, Ys, (9)
chosen in the next chapter. cDV is the equivalent stress
Volume:
amplitude according to the Dang Van criterion, a high
cycle fatigue criterion which can be applied to any owv= F,(HV + 120) (10)
multiaxial stress-time history [12]. In the present context,
The coefficients FA and F, can be determined with
the fatigue limit can be described by two stress parame-
reference specimens.
ters: z,, the maximum shear stress amplitude, and the
Y, is the surface roughness coefficient according to the
maximum hydrostatic stress pmax during one cycle, which
German standard DIN 3990 which depends on the mean
depends on the principal stresses:
surface roughness R, (DIN 4768):
pmax = l/3 max(0, + o2 + ~7~) (4)
Y, = 1.490 - 0.471 (R, + l)‘.’ (11)
The Dang Van criterion describes the regions of failure
and survival in the za -pmax plane which are separated Following Melander and Preston [ 151, the effect of
from each other by the following straight line: surface oxidation on the fatigue limit can be character-

z, + ap,,, = b (5) Table 2


The parameters a and b can be determined by the Calculation parameters

measurement of the fatigue limits under two loading


F* mA -u, (W F” mv HO
conditions. If this is done under push-pull (R = - 1) and
repeated pull (R = 0), the slope a can be expressed with 3.675 21 3.1 1.375 25 293
the mean stress sensitivity after Schi.itz [ 131:
H. Bomas et al. /Materials Science and Engineering A234-236 (1997) 393-396 395

crack initiation in the volume. The specimensof charge


II [17] are also cylindric with a net diameter of 12 mm.
The material is 16MnCr5. The specimensof charge III
[18] are bars with rectangular cross section and were
loaded by plane bending with a stressratio R = 0. The
bending height is 17 mm. Specimen H2 was taken as
additional reference in order to determine the parame-
ter Ho describing the mean stresssensitivity. The speci-
mens of charge IV [19] are cylindric with a net diameter
of 12 mm. The steel is 16MnCrS5.
The hardness and residual stress distribution of all
specimens is known, so that the survival probability
could be calculated. The calculation parameters de-
scribed in chapter 2 were optimized until the calculated
and measured fatigue limits of the reference specimens
were identical. Table 2 shows the optimized parameters,
which were used to calculate the fatigue limits shown in
400 600 800 1000 1200 Table 1 and Fig. 1. It can be seen, that the involved
Measured Fatigue Limit [N/mm21 effects on the fatigue limit of the examined unnotched
specimens, concerning their manufacturing, geometry,
Fig. 1. Measured and calculated fatigue limits of case hardened, and loading, are well described by the presented calcu-
unnotched specimens under rotating bending, push-pull and plane lation method. Further work will show the applicability
bending.
of this method to notched specimens.
ized by an equivalent short or long crack, depending on
its depth xso and a fit parameter x0. Considering this, a
factor Yso after Topper and El Haddad [16] is used to References
describe the effect of surface oxidation:
[l] W. Weibull, Ingenieur-Archiv 28 (1959) 360-362.
y,,= x, [2] J. Biihm, K. Heckel, Zeitschrift fur Werkstofftechnik 13 (1982)
(12) 120- 128.
J---x0 + Go
[3] K. Heckel, J. Kiihler, Zeitschrift fiir Werkstofftechnik 6 (1975)
The Weibull exponents mA and m, are determined by 52-54.
reference specimens. [4] C Kra, Beschreibung des Lebensdauerverhaltens gekerbter
Proben unter Betriebsbelastung auf der Basis des statistischen
By multiplication of the survival probabilities of
GroBeneinflusses, Thesis, Universitat der Bundeswehr, Miinchen,
neighboured elements, the survival probability of lim- 1988.
ited regions or of the whole part can be calculated. The [5] F. Scholz, Untersuchungen zum statistischen GrBBeneinfluB bei
fatigue limit is the nominal stress amplitude with the mehrachsiger Schwingbeanspruchung, Fortschritt-Bericht, VDI,
total survival probability 0.5. Reihe 18, Nr. 50, Dusseldorf, VDI-Verlag, 1988.
[6] G. Schweiger, Statistischer GrdBeneinfluCi bei unregelmLl3iger
Schwingbeanspruchung, Thesis, Hochschule der Bundeswehr,
Munchen, 1983.
3. Application to unnotched, case hardened specimens [7] R. Kuguel, ASTM Proc. 61 (1961) 732-744.
[8] J. Liu, H. Zenner, Materialwissenschaft und Werkstofftechnik 26
(1995) 14-51.
In order to study the described method with easy
[9] C.M. Sonsino, Konstruktion 45 (1993) 25533.
calculations procedures, it was applied to unnotched, [lo] H. Bomas, T. Linkewitz, P. Mayr, F. Jablonski, R. Kienzler, K.
case hardened specimens of two steel grades under Kutschan, M. Bather-Hiichst, F. Miihleder, M. Seitter, D.
rotating bending (RB), plane bending (PB), and push- Wicke, in: G. Liitjering, H. Nowack (eds.), Fatigue ‘96, vol. I,
pull loading (PP) including different mean stressesand Pergamon, Oxford, 1996, p. 141-146.
[ll] H. Bomas, T. Linkewitz, P. Mayr, F. Jablonski, R. Kienzler, K.
different case hardening processes. Table 1 shows the
Kutschan, M. Bacher-Hiichst, F. Miihleder, M. Seitter, D.
most important data of the examined specimensfrom Wicke, in: B.L. Karihaloo, Y.-W. Mai, M.I. Ripley, R.O. Ritchie
the steels 16MnCr5 and 16MnCrS5 (steels for case (eds.), Advances in Fracture Research, Pergamon, Oxford, 1997,
hardening after German standard DIN 17 210), which pp. 1321-1328.
have partially been published before. [12] K. Dang Van, G. Cailletaud, J.F. Flavenot, A. Le Douaron,
H.P. Lieurade, in: M.W. Brown and K.J. Miller (eds.), Biaxial
The specimens of charge I are cylindric with a net
and Multiaxial Fatigue, EGF 3, Mechanical Engineering Publi-
diameter of 6 mm. The material is 16MnCrS5. The cations, London, 1989, pp. 459-478.
specimensA and X were chosen as reference becauseA [13] W. Schiitz, Zeitschrift fur Flugwissenschaften 15 (1967) 407-429.
exhibits crack initiation at the surface and X shows [14] Y. Murakami, JSME Series I 32 (1989) 167-180.
396 H. Bomas et al. /Materials Science and Engineering A234-236 (1997) 393X396

[15] A. Melander, S. Preston, Mater. Sci. Forum 1022104 (1992) (1993) 111-124.
199-210. [18] H. Bomas, P. Mayr, Antriebstechnik 27 (1988) 40-45.
[16] T.H. Topper, M.H. El Haddad, Can. Metall. Q 18 (1979) 207. [19] C. Jansen, H. Bomas, P. Mayr, in 28. Tagung des Arbeitskreises
[17] G. Lowisch, H. Bomas, P. Mayr, in: H.P. Rossmanith, K.J. Bruchvorgange, Deutscher Verband fur Materialforschung und
Miller (eds.), Mixed Mode Fatigue and Fracture, ESIS Publ. 14 -priifung, 1996, pp. 247-256.

Potrebbero piacerti anche