E F F E C T OF C O R R O S I V E A N D S U R F A C E - A C T I V E S T R E N G T H OF A L U M I N U M A L L O Y S

MEDIA

ON T H E F A T I G U E

A. V. Karlashov, A. D. Gnatyuk, and V. P. Tokarev Fiziko-khimicheskaya mekhanika materialov, Vol. 1, No. 1, pp. 7-11, 1965 The lack of published data on the effect of working media on the fatigue strength of a l u m i n u m alloys, and of data based on likely theories of the corrosion resistance of these materials, demands a broad investigation of the problem. Here we present the results of experimental studies on the effect of corrosive and surface-active media on the fatigue strength of a l u m i n u m altoys D16 and V95, which are widely used in aircraft construction. As corrosive media we chose fresh water and a 3% NaC1 solution, to simulate seawater. As the surface-active medium we chose liquid AMG-10, which is used in the hydraulic systems of aircraft, activated with 2% oleic acid (CI7HaaCOOH), TABLE

1

Chemical composition, % Material

Cu

big

DI6

4.0

1,29

V95

1.5

2.1

Zn

bin

Cr

0.76 5,8

0.3

0.2

Fe

Si

Ni

6.a

0.19

0.04

0.2

0.07

0,07

The above materials were tested on Soviet MUI-6000 testing machines at 6000 stress reversals per m i n u t e with a test base of 20 9 166 cycles. For these tests we used smooth, round test pieces prepared to industrial specifications in accordance with Soviet Standard GOST 2860-48 from bars of D16 and V95 from the same m e l t . The c h e m i c a l composition and m e c h a n i c a l properties of the materials tested are given in Tables 1 and 2. TABLE

2

Mechanical properties Material Ou kg/mm 2 ~ kg/mm 2 6 % D16

I

46.5

32.1

50,1

41.2

18

Since the tests were to be performed in liquid media, the length of the working part of the test pieces was shortened to 66 ram, so as to bring the sealing sleeves of the liquidmedia chamber proposed by Yu. I. Babel [1] to the ends of the test piece

Typical of fatigue tests on a l u m i n u m alloys is a large scatter in the endurance values at the same stress level, which in some cases reaches a factor of several tens [2]. This makes a correct estimate of the carrying capacity of the material under alternating stresses more difficult. Hence all our data for the alloys D16 and V95 were analyzed by a statistical method, using a logarithmic normal distribution law in accordance with [2, 3, 4]. This method of analysis requires a large number of tests in order to construct the complete fatigue strength probability diagram characterizing the relationship between stress, number of cycles to failure, and probability of failure. V95

10

According to the data given in [5], the optimum n u m ber of tests necessary to construct the complete fatigue strength probability diagram for a l u m i n u m alloys is not less than 15-20 for a single stress level. Hence, in constructing a single fatigue curve for 5-6 levels, we had to test 80-120 test pieces. The data from this number of tests were treated statistically, and this gave a statistical estimate of the probability of failure P(N)O 9 100%. The processed test data are given below. We arbitrarily adopted an allowable failure probability of 0.5%. This m e a n t that, when 200 identical test pieces were tested at one stress level, 199 of them lasted out the base number of cycles and only failed when the number of cycles indicated by the fatigue curve corresponding to P = = 0.5% was exceeded.

Okg/mm 2 ~L

Jill I I fII111f

....

'|

T I IIliltl t

?015 10 5 0

105

fOe

I0 r &lO;" N

Fig. 1. Fatigue curves for D16 with failure probability P = 0.5~ for tests in air, water, and 3% NaC1 solution. 1) Air, 2) Water, 3) 3% NaC1.

Figures 1 and 2 show the fatigue curves for alloys D16 and V95 with 0.5~ failure probability in air, water, and 3% NaC1 solution. Alloy V95 was also tested by dripping AMG-10 liquid activated with 2% oleic acid onto the test piece. The diagrams show that the effect of corrosive m e d i a on the materials is greater, the higher the aggressiveness of the

m e d i u m and the longer the time during which it acts. It should be noted, however, that different alloys respond differently to an increase in the aggressiveness of the m e d i u m . For e x a m p l e , in ordinary water, as m a y be seen in Figs. 1 and 2, the fatigue strength of alloy V95 falls off more than that of D16. In 3% NaC1 the opposite holds: the greater fall in strength is displayed by D16. For a failure probability of 0 . 5 % the reduction in the end~lrance limits (based on 2 9 107 cycles) in corrosive m e d i a was" in water - 30% for D16, 46% for V95; in 3% NaC1 solution - 80% for D16, 7a07o for v95. ~. kg/mm 2

II 1

25

-, -.1~-,-

/

~

Fig. 2. Fatigue curves for a l l o y V95 with failure proba b i l i t y P = 0.5% for tests in air, water, 8%NaCI so lution, and AMG-10 a c t i v a t e d with 2070oleic acid. 1)Air, 2) AMG X0+ 207oClTHasCOOH, 3)Water, 4) 8070NaC1.

2

~.~--~" ~

20

,,

I tilttli

, I

}

0.10 3 I0 ~

I0 5

fO e

/01 ~.I0 r N

Thus, parts m a d e of alloy D16 are better suited for fresh water, and those m a d e of V95 for sea water. The s u r f a c e - a c t i v e m e d i u m has less influence on the loss of strength than the corrosive m e d i u m , and for alloy V95 with P = 0.5% the difference is 8-10%. The data, however, show that one must allow for this effect, since in aircraft m a n y of the structural elements ( e s p e c i a l l y in the lower part of the fuselage) are in contact with liquid AMG-10 which has fallen there as a result of carelessness in servicing the hydraulic system (spills during filling operations, leaks in pipe joints, e t c . ) .

6-1C,,V

tic= 6_18N

I,o

illlJl

I |If

",.

o.2

o

]111 2"I0"

lOs

106 a

10~ 2-10;" N

o

/11 2'10*

10s

106

[ll]Ilti

10~ 2.107 N

b

Fig. 8. Variation of coefficient ~c with test base N for failure probabilities P = 80~ and 0. 5070. a) Alloy D16: 1) Water, P = 50070, 2) Water, P = 0.5%, 3) 3070NaC1, P = 50070, 4) 8~ NaC1, P = 0. 5070. b) A l l o y V95: 1) AMG-10 + 207ooleic acid, P = 50070, 2) AMG-10 + 2070 oleic acid, P = 0.5070, 3) Water, P = 50070, 4) Water, P = 0.5070, 5) 8070 NaC1, P = 50070, 6) 8 9 NaC1, P = 0 . 5 9 . As criteria of the effect of the medium on the strength and endurance of the m a t e r i a l s studied, we chose the c o efficients Bc and 7 c , defined as follows. The coefficient for the effect of the m e d i u m on the endurance of the m a t e r i a l is $c = O-lcN/O_iBN; where O-lcN is the fatigue strength of the m a t e r i a l in the m e d i u m for N cycles and O-lBN is the corresponding value for air. The coefficient for the effect of the m e d i u m on the life (in cycles) is l c = N o - l C / N o - t B, where No.1 c is the life of the m a t e r i a l (in cycles) in the m e d i u m at the stress level o, and No_zB is the corresponding value for air. Figure 8 (a, b) shows the variation of the coefficient Bc with the test base N for failure probabilities of 50 and 0.507~ respectively. As m a y be seen from the graphs, the choice of failure probability has a considerable effect on the d e g r e e of influence of the corrosive and s u r f a c e - a c t i v e m e d i a . Reducing the failure probability leads to a lower value of the coefficient 13c, This confirms the previously mentioned notion of the importance of determining the l o a d - b e a r i n g c h a r a c teristics of the m a t e r i a l in corrosive and s u r f a c e - a c t i v e m e d i a on the basis of probability concepts of fatigue strength; unfortunately, few authors have yet done so. The variation of the coefficient 7c with stress level is shown in Fig. 4 (a, b). It follows from the graphs that the effect of corrosive and s u r f a c e - a c t i v e m e d i a on the life of the m a t e r i a l s tested is even more m a r k e d . At stress levels

close to the arbitrary endura]~ce limits in air, for a failure probability of P = 0.5%, the ]ife of the materials tested fa]h by a factor of several tern. Thus, for example, for alloy V95 in 3% NaCl soiution the factor is 70-80, and for I)lg 50-~0. When the failure probability is reduced, the effect of the media ou the life in cycles diminishes, as shown by the higher values of Tc for P = 0.5% (see Fig. 4 a , b ) . This may be because the time For which the medium acts diminishes with diminishing failure probability. NcO *'c = -~-d

NCO

0.6 -.A.~ 3

0.2

0.6

I

0,+

?

~2

'_-r_ 5 [i .........;........

:or"

l

0

0 5

to

t5 a

5

20 25 30 o kg/mm2

/ ! ,.i

l!

IO

!

r

20

30 25 o kg/mm z

Fig. 4. Variation of coefficient 7c with stress level o for failure probabilities P = 50% and 0.5%. a) Alloy DI6: I) Water, P ~ 0. 5070. 2) Water, P = 50070, 3) 3 % NaCI, P = 3070NaCI, P = 0.507o, 4) 3~ NaCI, P - 50070.b)Alloy V95: I) A M G - 1 0 § 2070oleic acid, P = 0.5~ 2) A M G - 1 0 + 2~ oleie acid, P = 50%, 3) Water, P = 0.507o, 4) Water, P : 50070, 5) 3070NaC1, P : 0.507o, 6) 3%NaCI, P= 50%. Table 3 shows numerical values of the endurance limits and values of the coefficients ~c and 7e for various failure probabilities obtained for a base of 2 9 107 cycles. TABLE 3 Endurance Failure probabil- l i m i t in air O-91 ity, P 070 kg/mm 2

D16

P=50~ P=0.5%

V95

P=50% P=0.5 %

11.5 10.0 15.0 13.0

Value of coeff.

Endurance limit in media o-m, k g / m m ~

15c

A M G- i0 Water

8.5 7.0 9.5 7.0

s9 NaCI

+

C17Ha3 COOH

0.74

3.7 2.0 5.t 3.5

2070 : Water 3% NaCl

14 12

oL 0.54

0,32 0.2 0.34 0,27

Value of coeff. Ye AMG- 10

AM G - i0 + 207o Water C17Haa COOH

--

0,25 0.22

0.93

I 0,028

-

-

0.92

io.o2

3070

+ 2070

NaCI

C17Hsa COOH

0.037 0.017 0.016 0.012

--0.215 0.213

SUMMARY I. Analysis of fatigue tests on alloys DI6 and V95 exposed to corrosive and surface-active media indicates that these media (especially the corrosive) do affect the fatigue strength of the materials in question. The degree of influence depends on the aggressiveness of the media and the test base. Thus, for example, in testing DI6 and V95 in ordinary water, the endurance limit for P = 0.5~ falls by 30-4607o as compared with that obtained in air. On testing these materials in a more aggressive medium, 307oNaCI solution, the limit fails by 4-5 times. 2. Different alloys behave differently when the aggressiveness of the m e d i u m changes. For example, in ordinary water, the fatigue strength of V95 falls more than that of D16, but in 3~ NaC1 solution the reverse is true. 3. Reducing the failure probability leads to an increase in the effect of the m e d i u m on the fatigue strength of the m a t e r i a l . 4. The action of corrosive and surface-active m e d i a results in a very considerable shortening of the life measured in cycles. At stress levels close to the arbitrary endurance l i m i t in air, when P = 0.8% the life in cycles is reduced by a factor of several tens for the materials studied. This loss of endurance is an extremely important factor in determining the service life of aircraft on the basis of accumulated fatigue damage over a wide range of loadings with respect to both amplitude and frequency.

REFERENCES 1. Yu. I. Babel, "The IMA-30 machine for testing the fatigue of metals in working media," Machines and Apparatus for Testing Metals [in Russian], lzd. AN UkrSSR, 1961. 2. S. V. Serensen, V. P. Kogaev, and E. V. Giatsintov, "Stability of lifetime distribution functions in fatigue tests on aluminum alloys," Trudy MATI, no. 37, 1959. 3. S. V. Serensen, V. P. Kogaev, and E. V. Giatsintov, "Study of the scatter of fatigue strength characteristics of structural aluminum alloys in connection with their production technology," Tmdy MATI, no. 35, 1958. 4. S. V. Serensen, V. P. Kogaev, and E. V. Giatsintov, "Endurance distribution law in fatigue tests," Zavodskaya laboratoriya, no. 3, 1958. 5. M. N. Stepnov, "Determination of the cycle sensitivity threshold in fatigue tests on aluminum alloys," Zavodskaya laboratoriya, no. 7, 1962. 12 August 1964

Kiev Institute of Civil Aviation,