The Reliability Life of 7B04 Aluminum Alloy under Alternate Action of Corrosion and Fatigue

Experiment of alternate action of corrosion and fatigue on 7B04 aluminum alloy was carried out in this paper and fatigue life was obtained. One of the characteristics about fatigue life is that it is not always decreasing with the increasing of corrosion time. The surface splits caused by the corrosion of intruding and extruding slip steps on the surface and decreasing of stress concentration around corrosion pits were themain reasons. Single side allowance factor kwas proposed to calculate reliability fatigue life with both reliability and confidence. Reliability, confidence, and specimen number had affection on the value of k. The reliability influences more greatly the reliability fatigue life compared to the confidence. The safe fatigue life is smaller than median life and it was safe to administrate aircraft life using safe fatigue life.


Introduction
Service life is one of the most important indexes of aircraft structure.In order to ensure flight safety, the service life is actually a life limit with high reliability [1].The determination of the life limit has been widely discussed to ensure that the aircraft is used safely and economically.Safe fatigue life is one main index of aircraft which is determined by fatigue test [2].Meanwhile, the test data is obtained by fatigue experiments without corrosive environment.
Corrosion and fatigue are main damage to aircraft structure in service and many investigations have found the influence of corrosion and fatigue on the aircraft material [3][4][5][6].Most studies focus on the corrosive environment and fatigue load simultaneously or separately.Precorrosion fatigue and corrosion fatigue damage rules have been broadly reported and indicate a synergism between corrosion and fatigue [7][8][9].When the aircraft flies above 3 km, the corrosive environment is minimal, for example, salt, temperature, moisture, and so on, while the fatigue load due to lift force, maneuver, flutter, and so forth is maximal.And when the aircraft is on the ground, the corrosion progress is maximal, while flight loads and vibration are not existent [10].So fatigue progress and corrosion progress are alternate during the service progress of aircraft.Some research had been done to investigate the difference between alternate action of corrosion and fatigue and precorrosion fatigue.Menan and Henaff [11,12] found the difference between alternate immersion corrosion and precorrosion in research of the affection of frequency and salt solution on crack growth of 2024 aluminum alloy.Chen et al. [13] showed that the fatigue life of 2A12 aluminum alloy specimens in "fatigue + corrosion + fatigue" mode was longer than precorrosion specimens under the same corrosion condition.Jin and Yang [14] put forward a fatigue damage competitive rule.The experimental results under "fatigue + corrosion + fatigue" mode were in agreement with the rule.Fan [15] studied distribution law of the alternate action of corrosion and fatigue under "corrosion + fatigue + corrosion + fatigue + ⋅ ⋅ ⋅ " mode.Experimental results showed that it obeys the normal distribution.But these investigations failed to discuss the safe fatigue life and to predict the reliability fatigue life under the reliability and confidence.
As experiments of alternate action of fatigue and corrosion can account for aircraft service progress well, it is interesting to discuss the safe fatigue life under alternate state.In this study, the Al 7B04 alloy was subjected to alternate action of fatigue and corrosion in order "corrosion + fatigue in air  + corrosion + fatigue in air + ⋅ ⋅ ⋅ ."The surface characteristic of specimen was shown and life variation rule was discussed in this paper.Logarithm normal and three-parameter-(3-) Weibull distribution were applied to describe the fatigue life distribution and a better one was chosen to study the reliability fatigue life.Single side allowance factor was taken into consideration to describe the reliability fatigue life and the affection of reliability, confidence, and specimen number were discussed.

Experiments
2.1.Materials.In this paper, all the specimens are made of the 7B04-T6 aluminum alloy plate in L-T direction where the thickness of the plate is 2 mm.The shape and size of the specimens are shown in Figure 1.In order to study the environment's affection on the plane, all specimens are covered with protection lay.The specimen in direction S was polished with the emery paper in the order of #600, #800, and #1000.
The chemical composition and mechanical properties of 7B04-T6 aluminum alloy are shown in Tables 1 and 2, respectively.

Experiment Methodology.
Completely reversed tension and tension fatigue test were conducted under a sine load waveform at constant cyclic stress frequency ( = 20 Hz) on the MTS 810 servohydraulic test machine (MTS Systems Co., USA).The stress amplitude applied in fatigue test was 190 MPa and stress ratio  was 0.06.The alternate load of corrosion and fatigue test specimens was divided into six sets, each with different load cycles and corrosion time, as shown in Table 3.
The solution used for the corrosion investigations was made of 3.5% NaCl + 5 × 10 −5 M H 2 SO 4 (PH = 4) using  Surface damage by corrosion and fatigue is the main failure factor of aircraft structure.Corrosion pitting has been found to be responsible for the nucleation of fatigue cracks in a wide range of aluminum alloys [20].The examination of the specimens' surface indicated that cracks essentially nucleated and grew from corrosion pits [16].The surface of specimens under precorrosion fatigue and corrosion fatigue always has one or two main large corrosion pits and the biggest pits are critical for pit-to-surface crack, as shown in Figure 3.However, there is much small proximity of corrosion pits growing and combined into a unit on the surface of alternate experiment specimen, as shown in Figure 3.Moreover, optical observations also prove that there are many surface splits on the alternate specimens, as shown in Figures 4(a During the alternate progress, fatigue loads produce intruding and extruding slip steps on the surface, as shown in Figure 5.However, if the material is subjected to a corrosive environment at this stage, corrosion tends to remove intrusions and extrusions and forms many small corrosion pits and the surface roughness becomes weak and tends to form split on the surface.When the material fatigue is after corrosion again, the slips form perpendicular to stress direction.Many investigations have found that the fatigue life decreases when the corrosion time increases [15].Figure 6 shows that specimens of 1-4 and 1-5 have longer fatigue cycles than 1-3, even though specimens of 1-4 and 1-5 are subject to longer corrosion time.When the specimens are subject to longer corrosion time, more intruding and extruding slip steps are dissolved.While the material is subject to fatigue loads again, more surface slips are formed; thus, more fatigue damage energy is dispersed.The corrosion pits of 1-4 consist of more small pits at the same size, and pits shape is not as sharp as 1-3; reducing the stress concentration leads to a longer fatigue life.

Distribution Test of Fatigue Life
Many investigations have found that logarithm normal distribution and three-parameter-(3-) Weibull distribution were applied to describe the fatigue life data [21,22].In this paper, both the distributions were discussed and the best  (1) Logarithm Normal Distribution.The PDF and CDF of logarithm normal distribution can be expressed by ( 1) and ( 2): lg  is the random variable,  is the mathematical expectation, and  is the variance; (2) can be replaced by By assuming  = lg ,  = Φ −1 [(lg )],  = −(/), and  = 1/, the standard linear equation can be obtained as follows: The evaluating values of ,  were calculated by using maximum likelihood estimation method in Table 4.
(2) 3P-Weibull Distribution.The PDF and CDF of 3-Weibull distribution are expressed by where  is the random variable,  is the lower limit parameter of random variable , and  is the shape parameter. is the feature parameter which is independent of other constant parameters because () = 1 −  −1 : ) .Using ( 6) into ( 7), ( 7) can be rewritten as ( 8) by taking the logarithm of both sides: By assuming  = ln( − ),  = ln ln[1/(1 − ())],  = − ln( − ), and  = , the standard linear regression equation can be obtained as follows: The empirical frequency function is defined as follows: is the sample size and  is the serials number of fatigue life data in ascending order at the same given stress level.
The values of , , and  were calculated by using the least square method [23,24] in Table 4.

Statistical Test.
To determine which distribution is better to describe the fatigue life data, the correlation coefficient  is applied.The correlation coefficient  is defined as follows: where There exists the equation || ≤ 1.A good correlation can be affirmed between variables  and  if || is close to 1.So the better fitting of distribution can be determined by comparing  value.However, only total eight specimens are used in fatigue experiment.So it is necessary to determine a threshold value for  under different sample size and confidence.The statistical test was carried out by using (13) [25].Consider ( − 2) is Student's distribution with  − 2 degrees of freedom To guarantee that the fitting is precise,  should satisfy the following condition: Well,   is 0.621 under the sample size  = 8 and  = 0.05.The correlation coefficient  and distribution parameter of two distributions are listed in Table 6.It is obvious that the fatigue life data fits both logarithm normal distribution and 3-Weibull distribution and the correlation coefficient of both distributions is larger than 0.621.However,  of 3-Weibull distribution is smaller than logarithm normal distribution, indicating that logarithm normal distribution is a better description for the fatigue data.

Single Side Allowance Factor. As mentioned above in
Introduction, many investigations about fatigue life under corrosive environment focus on the scatter factor and distribution.However, confidence was not taken into consideration, especially in alternate action of corrosion and fatigue.In many fatigue experiments the specimens are not sufficient because of the high cost, so it is necessary to explore the reliability fatigue life at various reliability  and confidence .Here, the single side allowance factor is proposed.
In previous results presented by this paper, the fatigue life data obeys logarithm normal distribution better.The logarithm reliability life   with reliability  can be presented as is the mathematical expectation,   is the upper-side factor of standard normal distribution, and  is the standard deviation.
Because the values of  and  are unknown, the logarithm reliability life   can be rewritten as follows:   is the mean value of logarithm fatigue life data from the sample, σ is the estimated value of ,   is the standard deviation of the sample, and k is the correction factor of σ.
k can be presented as in ( 18) [24][25][26].Consider is sample size and Γ is gamma function.  in ( 17) can be replaced by the single side allowance factor .The probability that +  kŝ  is smaller than +   is  (see Figure 7).So The random variable  is defined as Assuming that  obey the normal distribution, ( 16) can be expressed as (20) according to Figure 7: Based on the statistical theory, the mathematical expectation () is defined and calculated as It is obvious that () = , ( k  ) = (σ) = , so ( 22) can be replaced by According to (20), the variance Var () can be written as Here Var (  ) is given by ( 26) [26].Consider Take ( 18), (25), and ( 26) into (24).So Var() can be rewritten as follows: If  ≥ 5, (28) can be obtained as follows: Thus ( 27) can be simplified into ] .
Taking ( 23) and ( 29) into ( 21), (30) can be obtained as follows: Then So the single side allowance factor  can be presented as follows: .
With given reliability  and confidence  which can be calculated through standard normal distribution, the logreliability fatigue life x can be calculated by x =  + σ.
(33) Finally the reliability fatigue life  , with reliability  and confidence  can be calculated by The value of  with various reliability from 50% to 95% (increasing in step of 5%), and confidence from 50% to 95% (increasing in step of 5%) was calculated based on (32).As shown in Figure 8 ( = 8), the value of  decreases when reliability  and confidence  increase at the same  state.
The value of  tends to increase slowly and reach a stable constant with specimens number  increasing, as shown in Figures 9 and 10 (especially  = 50%, the value of  is constant,  = −1.6499). = 8 is the critical point of increasing velocity of ; when  < 8, the value of  rises rapidly with the increasing of specimen number .While  > 9,  rises smoothly with the increasing of .So in order to achieve a higher reliability and stable fatigue life with a given reliability and confidence more specimens are needed and the amount should be larger than eight.

Reliability Fatigue Life.
Reliability fatigue life  , with reliability  = 95% and confidence  = 50%, 75%, 85%, and 95% was listed in Table 7. Reliability fatigue life increases with the confidence decreasing. , at confidence  = 95% and reliability  = 50%, 75%, 85%, and 95% was calculated and shown in Table 8.Safe fatigue life of a structure is that number of events such as flights, landings, or flight hours, during which there is a low probability that the strength will degrade below its design ultimate value due to fatigue cracking [27].The safe fatigue life is one main and important index for administrating aircraft service life and ensuring safety which is determined by fatigue test.When the test data obeys logarithm normal distribution, the safe fatigue life is fatigue cycles at 99.9% reliability and 90% confidence [2].So the safe fatigue life is  , at 99.9% reliability and 90% confidence, listed in Table 9.The median fatigue life has larger scale than safe fatigue life in each set, as shown in Figure 11.It indicates that it is safe to administrate the aircraft life using safe fatigue life.Figure 11 also shows that the fatigue life of the same set at reliability  = 0.95 and confidence  = 0.85 is longer than that at reliability  = 0.85 and confidence  = 0.95.So the reliability has more serious effect on the value of fatigue life than confidence.

Conclusion
Alternate action of corrosion and fatigue experimentation of AA7B04 was carried out to investigate the fatigue performances.The different characteristics between precorrosion and alternate action of corrosion and fatigue surface were studied.Special attention was given to the reliability fatigue life of using single side allowance factor.The major results and conclusions are summarized as follows.
(1) The fatigue life of alternate experimentation does not always decrease with the increasing of total corrosion time as the precorrosion test.The dispersion of fatigue under alternate experiment has obvious rule as precorrosion.No big-main corrosion pit and surface slips are distinct characteristic of alternate experiment.
(2) Fatigue life data of alternate experiment follows both logarithm normal distribution and 3-Weibull distribution but logarithm normal distribution has a better agreement.
(3) One-side allowance factor  is proposed to predict the reliability fatigue life  , based on logarithm normal distribution.The value of  increases with the decreasing of reliability and confidence.
The specimen number also had affection on the one-side allowance factor.The one-side allowance factor  increases  with increasing of specimen number and  = 8 is the critical point of increasing velocity.
(4) Safe fatigue life is smaller than median fatigue life and it is safe to administrate aircraft life using safe fatigue life.

Figure 1 :
Figure 1: The dimension of specimens.

Figure 11 :
Figure 11: Reliability fatigue life and median fatigue life of different specimen set at various reliability and confidence.

Table 2 :
Mechanical properties of the material.

Table 3 :
The corroded time and fatigue cycles in each alternate progress.

Table 4 :
The test data of alternate action of corrosion and fatigue experiment.Specimen set Specimen number Cycles/ Stress level Specimen number Cycles/ Stress level Specimen number Cycles/

Table 5 :
The total corroded time and alternate times of the test.

Table 6 :
The evaluating value of , , , , , and  for log-normal distribution and 3P-Weibull distribution.

Table 9 :
Safe fatigue life of different specimen set.