ISSN 1068-798X, Russian Engineering Research, 2008, Vol. 28, No. 3, pp. 217–220. © Allerton Press, Inc., 2008.

Original Russian Text © D.V. Vikhrenko, 2008, published in Vestnik Mashinostroeniya, 2008, No. 3, pp. 27–30.

Estimating the Life of Complex Parts

with Stress Concentrators
D. V. Vikhrenko
Minsk Automobile Plant (MAZ), Minsk
DOI: 10.3103/S1068798X08030040

In most cases, determinate calculation of the life of Thus, Eq. (1) takes the form
machine parts is based on the Basquin power law
σR ⎞ m
N det = N 0 ⎛ --------------
- , (2)
⎛ σ ⎞
m ⎝ Kσ Rrat⎠
N = N0⎜ R ⎟ , (1)
⎝ σ RN ⎠ where σR rat is the rated stress in the part determined by
tests or strength calculations, disregarding the stress
where N is the life of the part; N0 is the number of concentration (Fig. 1).
cycles to the break in the fatigue curve; σR is the fatigue Thus, in calculating the life of the designed part, the
limit of the material with asymmetry coefficient R; σRN rated stress and various correction coefficients must be
is the stress acting; m is the slope of the fatigue curve. determined. However, the parts considered do not
always correspond to the existing classification, and
Besides Eq. (1), other formulas have been pro- strength calculations do not always give reliable infor-
posed—for example, the three-parameter equation [1] mation on the stress distribution.
As an example, consider an automobile component
σ RN – σ R⎞
N = N 0 exp ⎛ – --------------------
- , such as the crankcase for the drive shaft. Under the
⎝ ν ⎠ action of a static vertical load, the cross section in the
region of the spring mount is hazardous according to an
where ν is the slope of the fatigue curve in semiloga- analytical calculation by the beam theory of material
rithmic coordinates. strength (Fig. 2, cross section 1–1). In that case, the
In both cases, the dependence of the life on the stress stress is determined using the formula σrat = Mfl/W
acting is presented relative to some basic number of where Mfl is the flexural torque and W is the torque of
cycles, with the corresponding fatigue limit, whose the cross-sectional strength.
value is determined, as a rule, by testing smooth cylin- By the finite-element method, we find that the max-
drical samples under flexure with rotation. imum stress arises over the rounding radius of the cross
section (transition from the sleeve to the head), on the
However, in calculating the life of real parts, various inside of the crankcase (Fig. 2, cross section 2–2),
factors that greatly affect the fatigue strength must be although the stress should decline in this zone accord-
taken into account. Thus, in determining the fatigue ing to beam theory, on account of the increase in W at
limit of a part, its decrease must be taken into account, constant Mfl. The difference between the rated and max-
according to State Standard GOST 25.504-82 [2]. The
corresponding coefficient is
δ
Kσ
K = ⎛ --------
- + --------- – 1⎞ -------------,
1 1
⎝ K dσ K Fσ ⎠ K ν K A

where Kσ is the effective stress-concentration coeffi- d σmax

cient; the factor Kdσ takes account of the dimensions of F F
b
the part; KFσ takes account of the influence of the sur-
face roughness; Kν takes account of the influence of
surface hardening; KA is the anisotropy coefficient.
All these coefficients are determined for standard σrat = F/Aco
surface-treatment processes and for parts such as shafts
with grooves and recesses or plates with holes and cuts.
(Their values are presented in the specialist literature.) Fig. 1. Stress concentrations in a plate with a hole.

217
218 VIKHRENKO

174
1 2
161
148
135
123
1
110
97
84
72
59
46
33
21
8
2 –4
–16

Fig. 2. Stress distribution in the crankcase of an automobile driveshaft.

imum stresses is significant here. In that case, σmax = [1–4]. Software has been developed for calculating the
2.6σrat. In addition, it is impossible to determine the life, such as the Resurs program used at the MAZ plant;
most dangerous zone and the corresponding stress by this software was developed at the NIRUP Belavtotrak-
the strength method, taking account of the stress con- torostroenie Institute within the GNTP Belavtotraktor-
centration. Therefore, we need a general method of cal- ostroenie State Enterprise [5].
culating the fatigue strength based on determining the
stress–strain state in the zone of the stress concentrator However, this method does not permit the calcula-
by calculation (for instance, by the finite-element tion of parts with stress concentration, such as the
method), without calculating the rated stress and the crankcase of the driveshaft. In calculations based on the
coefficients for particular cases of stress concentration. maximum stress, the result obtained is an order of mag-
nitude less than the predicted life. Hence, an expedient
In that case, determination of the life by calculation method of taking the stress into account must be devel-
and experiment is expedient. The laboratory samples oped, so that the life may be calculated from the stress
employed are made by the same technology as the part rather than the rated stress.
in question, with a comparable characteristic dimen-
sion (the thickness, for crankcases of driveshafts and To this end, the rated stress in Eq. (2) is replaced by
frames; Fig. 3). Thus, the influence of the surface state the maximum value taking account of the stress-con-
and manufacturing technology on the fatigue limit is centration coefficient ασ = σmax/σrat and the fact that
taken into account. Practical experience confirms the only the effective stress-concentration coefficient Kσ
effectiveness of this approach in producing the samples has yet to be taken into account in the coefficient char-
acterizing the decrease in the fatigue limit

σmax = 1.17 σrat ασ σ R ⎞ m

N det = N 0 ⎛ -------------------
- . (3)
⎝ K σ σ Rmax⎠

We now eliminate the two unknown concentration

40
25

coefficients from Eq. (3). State Standard GOST 25.504-82

recommends the determination of the effective stress-
concentration coefficient Kσ by several methods. The
simplest is based on the sensitivity q of the material to
R50 stress concentration

Fig. 3. Stress distribution in a sample on extension. K σ = 1 + q ( α σ – 1 ). (4)

RUSSIAN ENGINEERING RESEARCH Vol. 28 No. 3 2008

 ESTIMATING THE LIFE OF COMPLEX PARTS WITH STRESS CONCENTRATORS 219

Despite its simplicity, this method does not elimi- type of load must also be taken into account: the crank-
nate the two unknown coefficients when Eq. (4) is sub- shaft is subjected to flexures, whereas the samples are
stituted into Eq. (3). In addition, this method of deter- tested in tension and compression.
mining Kσ is inaccurate.
The relation between the fatigue limits in tension
By the Siebel–Stieler method, the effective stress- and compression (σRt) and in flexure σRf is as follows
concentration coefficients for steel are calculated from [3]: σRt = 0.75σRf.
the following formula (to within 20%)
If we assume that the coefficient for conversion
K σ = α σ /n, (5) from flexure to tension–compression kft = 0.75, we may
write Eq. (3) in the following form, taking account of
σt ⎞
– ⎛ 0.33 + --------
- all the coefficients
⎝ 712⎠
where n = 1 + G × 10 ; here G =
α σsam σ Rsam m
1 dσ N det = N 0 ⎛ --------------------------------------------⎞ . (7)
---------- ------
σ max dx x = a
is the relative stress gradient [6]. ⎝ k ft F ( θ, ν σ )σ Rmaxdet⎠

It is simple to determine G —and hence n—from Equation (7) only includes parameters determined
the results of calculating the stress–strain coefficient by explicitly in calculating the stress–strain state of the
the finite-element method. Thus, substituting Eq. (5) part by the finite-element method and by tests of sam-
into Eq. (3), we finally obtain the life ples and does not require the determination of addi-
tional coefficients or parameters.
N det = N 0 ( nσ R /σ Rmax )
m

Determination of the effective stress-concentration EXAMPLE

coefficient taking account of a scale factor (the Kogaev
method) is regarded as more accurate [6] The parameters in Eq. (7) are found to be as follows:
σ0maxdet = 330 MPa (finite-element calculation in
K σ = α σ F ( θ, ν σ ). (6) teststand loading of a crankcase by a 190-kN force);
σ0sam = 193.5 MPa with a 90% probability of fault-free
2 operation; L = 28 mm; G = 0.11 mm–1; L0 = 16 mm;
Here F(θ, νσ) = -----------------
–ν
- is a similarity function; θ =
1+θ σ G 0 = 0.0356 mm–1; θ = 0.56; νσ = 0.129; F(θ, νσ) = 0.96.
L/G
----------------- is a dimensionless number; L = 2t is the perim- Thus, in teststand loading
( L/G ) 0
eter of the sample’s working cross section or the part of 1.17σ 0sam m
it subjected to maximum stress; t is the wall thickness N det = N 0 ⎛ --------------------------------------------
-⎞
⎝ 0.75 × 0.96σ 0maxdet⎠
of the crankcase and the sample; subscript 0 denotes the
sample parameters; νσ is the parameter of the similarity 1.625σ 0sam⎞ m
= N 0 ⎛ -------------------------
- ,
equation for the fatigue failure. (In the absence of ⎝ σ 0maxdet ⎠
experimental data, νσ = 0.19–0.000125σB when σB <
1200 MPa.) i.e., the calculated fatigue limit must be increased by a
In this case, substituting Eq. (6) into Eq. (3) elimi- factor of 1.625.
nates the unknown coefficients:
Calculation of the crankcase life with these parame-
σR m ters using Resurs software shows that its life is more
N det = N 0 ⎛ ---------------------------------
-⎞ . than 1 million loading cycles, as confirmed by teststand
⎝ F ( θ, ν σ )σ Rmax⎠
data. Disregarding the stress-concentration coeffi-
To calculate the dimensionless numbers for the sam- cients, the life is found to be 186000 cycles, i.e., less by
ple, its stress–strain state in extension is determined by a factor of 6.6, which is incorrect. Thus, the proposed
the finite-element method. The stress concentration in method permits reliable estimation of the life of
the sample is found to be ασsam = 1.17 (Fig. 3). machine parts.
The effective concentration coefficient for the sam-
ples is calculated by the Siebel–Stieler method. The REFERENCES
calculation results in fact agree with the theoretical val-
ues. (The difference is less than 1%.) Therefore, in cal- 1. Pochtennyi, E.K., Kinetika ustalosti mashinostroi-
culating the crankcase life, the fatigue limit determined tel’nykh konstruktsii (Fatigue Kinetics of Manufacturing
for a sample must be increased by the factor ασsam. The Structures), Minsk: Arti-Feks, 2002.

RUSSIAN ENGINEERING RESEARCH Vol. 28 No. 3 2008

220 VIKHRENKO

2. GOST (State Standard) 25.504-82: Strength Calcula- 5. Razrabotka programmy rascheta dolgovechnosti nai-
tions and Tests. Methods of Calculating the Fatigue bolee nagruzhennykh zon ramy po rezul’tatam modeliro-
Strength, 1982. vaniya uslovii nagruzheniya: Otchet (Developing Soft-
3. Mariev, P.L., Bernatskii, A.K., Rakitskii, A.A., et al., ware for Calculating the Life of the Most Loaded Zone
Povyshenie dolgovechnosti nesushchikh konstruktsii of the Chassis from the Results of Simulating the Load-
kar’ernykh avtosamosvalov (Extending the Life of Sup- ing Conditions: Research Report), Minsk: NIRUP
porting Structures in Pit Dump Trucks), Yakutsk: YaNTs Belavtotraktorostroenie, 2005.
SO RAN, 1991.
4. Pochtennyi, E.K., Rakitskii, A.A., Ryzhkov, E.P., et al., 6. Kogaev, V.P., Raschety na prochnost’ pri napryazhe-
Estimating the Fatigue Strength of a Stamped and niyakh, peremennykh vo vremeni (Strength Calculations
Welded Crankcase by Means of Local Models, Izv. AN with Time-Varying Loads), Moscow: Mashinostroenie,
BSSR, Ser. Fiz.-Tekhn. Nauk, 1988, no. 1, pp. 3–6. 1993.

RUSSIAN ENGINEERING RESEARCH Vol. 28 No. 3 2008