Accepted Manuscript

Effect of pre-deformation on the pre-corrosion multiaxial fatigue behaviors of

2024-T4 aluminum alloy

Yajun Chen, Jian Zhou, Chenchen Liu, Fusheng Wang

PII: S0142-1123(17)30428-0
DOI: https://doi.org/10.1016/j.ijfatigue.2017.11.008
Reference: JIJF 4503

To appear in: International Journal of Fatigue

Received Date: 18 July 2017

Revised Date: 13 October 2017
Accepted Date: 16 November 2017

Please cite this article as: Chen, Y., Zhou, J., Liu, C., Wang, F., Effect of pre-deformation on the pre-corrosion
multiaxial fatigue behaviors of 2024-T4 aluminum alloy, International Journal of Fatigue (2017), doi: https://
doi.org/10.1016/j.ijfatigue.2017.11.008

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Effect of pre-deformation on the pre-corrosion multiaxial fatigue
behaviors of 2024-T4 aluminum alloy
Yajun Chen, Jian Zhou, Chenchen Liu, Fusheng Wang

Sino-European Institute of Aviation Engineering, Civil Aviation University of China, Tianjin 300300, China.

Abstract
Pre-damages which contain pre-deformation and pre-corrosion have influences on multiaxial fatigue properties. In this

paper, the effect of pre-tension of 1%, 2%, 3% and 4% as well as pre-torsion of 10°, 20°, 30° and 40° on fatigue life of

2024-T4 aluminum alloy under corrosion conditions is evaluated. Multiaxial fatigue tests are performed under constant

amplitude sinusoidal wave loading with tension-torsion ratio of in air at room temperature. The results indicate a

reduction tendency of pre-corrosion multiaxial fatigue life with the pre-deformation level in the same pre-corrosion time.

With the same pre-deformation level, the pre-corrosion multiaxial fatigue life decreases as pre-corrosion time increases.

Pitting caused by corrosion is observed on the surface of specimens. The observed mechanical behavior and associated

phenomena are directly linked to microstructure characteristics such as surface defects, corrosion pits and micro-cracks. The

analysis of mechanism reveals that the synergistic combination of pre-deformation and pre-corrosion is more detrimental

than that of either one acting separately. The alternate cyclic hardening and softening appear in the axial direction. The

cyclic softening in the tangential direction becomes more serious with the increase of pre-deformation. The typical

morphology of corrosion fatigue can be observed on the fracture. Involving the damage factors of single pre-damage

conditions, which include pre-deformation and pre-corrosion effects, the damage factor for pre-corroded multiaxial fatigue

after pre-deformation is deduced. Fine results are achieved according to the above methods with most predicted life in 2x

scatter band.

Keywords: 2024-T4 aluminum alloy; Multiaxial fatigue; Pre-deformation; Pre-corrosion; Life prediction


Corresponding author.
E-mail addresses: yjchen@cauc.edu.cn (Y. Chen)
1
1. Introduction
Aluminum alloys are widely used in aeronautical structures. With the increase of aircraft service time, numerous

factors such as the unexpected deformation, complicated loading and corrosion damages can lead to accidental failures of

aluminum alloy components [1-3]. Besides, failure of airplane components, for instance, the landing gears, the joints and

the rotors [4-6] are always attributed to the multiaxial loads.

Steam and acid ions in the air usually cause corrosion problems for components. For example, the airplanes locate at

the coastal airports are exposed to corrosive environment on ground. In recent years, some studies were conducted to

investigate the corrosion fatigue performance of metal. Li et al. [7] used SEM (scanning electron microscope) to examine

the micro-cracks initiation and early expansion at corrosion pits of 6151-T6 aluminum alloy. The results showed a number

of crack origins around the pits and that the early expansion of fatigue crack was mixed mode. Frédéric et al. [8]

carried out corrosion fatigue tests on 2024 aluminum alloy in different corrosion environments. It was demonstrated that

fatigue and corrosion had a combined effect on crack growth. Hu et al. [9] proposed a pit growth model of aluminum alloy

with consideration of multiaxial fatigue damage. Different degrees of corrosion affected the size of pits and fatigue crack

initiation. Chen et al. [10] conducted researches on the multiaxial fatigue behavior of 2024-T4 aluminum alloy with

different pre-corrosion time and alternate corrosion modes. Suresh et al. [11] studied the fatigue crack growth properties of

pressure vessel steels under different environments. It indicated that compared with the condition of a dry environment, the

near-threshold crack propagation rate was much slower in moist environment. Chen et al. [12] carried out researches on the

changeover from pitting to fatigue crack growth by conducting fatigue experiments with different loading frequencies. The

results showed that the pitting was responsible for the fatigue crack initiation and the pit could be simulated by the

semi-elliptical model. As frequencies exceeded 5Hz, the parameters of pits were insensitive to the frequencies. Conversely,

the pit dimension and stress intensity coefficient increased with the decrease of frequencies which were lower than 5Hz.

During the process of manufacture, maintenance and operation, components are possible to be slightly deformed.

2
However, the induced damages can’t be detected easily and affect the material properties. Cuesta et al. [13] carried out

experiments with pre-notched small punch specimens to estimate the impact of biaxial pre-deformation on fracture

properties. The results demonstrated that for the plane stress condition, the biaxial pre-tension reduced the fracture

toughness of specimens linearly. Chiou et al. [14] studied the fatigue behavior of pre-stretched SUS 430 stainless steel

specimens at three levels. The study found an evident decrease of fatigue life caused by the pre-deformation. Significant

changes in cyclic characteristic response were observed. Park et al. [15] investigated the influence of pre-tension on the

uniaxial fatigue performance of a rolled Mg-3Al-1Zn alloy by conducting low-cycle fatigue tests after pre-tension. The

results demonstrated that the applied pre-tension increased the dislocation density and introduced a small number of twins

into the material. The pre-tension process could improve the yield strength of rolled Mg alloys without a loss of their

low-cycle fatigue resistance. Huang et al. [16] conducted the stress-controlled tests to investigate the effect of

pre-compression deformation on the low-cycle fatigue behavior of as-rolled AZ31 alloy, and the fatigue life of

pre-compressed samples was superior to that of the as-rolled samples under different stress amplitudes. Wang et al. [17]

conducted tensile and high-cycle tests of twinning-induced plasticity steel after pre-strain. The results indicated an

improvement in fatigue properties of pre-strained specimens due to the change of fatigue strength coefficient and exponent.

Wang et al. [18] investigated the impact of pre-tension on the evolution of microstructure and stress corrosion cracks of

over-aged 7050 aluminum alloy. A decline in corrosion cracking resistance was proved with increasing the pre-strain

percent. He et al. [19] verified that the pre-tension and annealing process induced beneficial effects on the enhancement of

mechanical performances and corrosion resistance of AZ31 sheet. Al-Duheisat et al. [20] conducted different degrees of

pre-deformation to structural steel girders type A36 and monitored by accelerated corrosion tests. The results showed a

beneficial effect on the initiation and development of cracks. The corrosion was more serious with the increase of

pre-deformation. Lv et al. [21] studied the effect caused by pre-strain on the corrosion resistance of 2205 duplex stainless

steel through some measurement methods. The results showed a reduction trend in corrosion resistance with the increase of

pre-deformation. Zhang et al. [22] proved that the pitting corrosion resistance could be improved by the slight

3
pre-deformation.

During last decade, prediction models for multiaxial fatigue had been well developed with consideration of damage

parameters, material response and loading mode [23]. Al-Rubaie et al. [24] carried out fatigue tests of 7050-T7451

aluminum alloy under different levels of pre-strain. It was shown that the specimen’s life had a decreasing trend with the

increase of pre-strain. A bootstrap statistical analysis was induced to build a prediction model. Gerin et al. [25] investigated

multiaxial fatigue behavior with different pre-strains caused by cold extrusion. Several mechanical and microstructural

properties were observed. With consideration of these effects, a fatigue criterion was developed to meet with the forward

process. Huang et al. [26] conducted pre-corrosion fatigue tests on 7075-T6 aluminum alloy and analyzed correlated

coefficients for corrosion cracks and cracks according to Pearson method. ECS (equivalent crack size) model was developed

for both single crack and multi-crack initiations with different stress levels. Ishihara et al. [27] proposed a pit depth function

of both stress and corrosion time. It was demonstrated that a threshold stress could divide the specimen life into crack

initiation stage and crack propagation stage. Pit growth rule could help to predict the fatigue life. However, few reports

focused on multiaxial fatigue life prediction with synergistic action of pre-deformation and pre-corrosion.

Researches of the influences of pre-deformation and pre-corrosion on fatigue properties were focused on uniaxial

loading while there few reports on multiaxial fatigue. Since the aircraft components are possible to be slightly deformed

with corrosion environment on ground before being subjected to fatigue load in air, an attempt is made to research the effect

of the pre-deformation on pre-corrosion multiaxial fatigue life of 2024-T4 aluminum alloy in this paper. Cyclic loading

curves are analyzed to point out the mechanism of the fracture after pre-deformation. The fracture morphologies are

observed. A criterion is proposed to predict the pre-corrosion multiaxial fatigue life after pre-deformation.

2. Experimental procedures
2.1. Material and specimens

In this study, 2024-T4 aluminum alloy was used as specimen material with the normal chemical composition of 0.1%

Cr, 3.8%-4.9% Cu, 0.5% Fe, 1.2%-1.8% Mg, 0.3%-0.9% Mn, 0.5% Si, 0.15% Ti, 0.25% Zn (mass fraction) and balance is

Al. The material had a tensile yield stress of 373.7MPa, a tensile ultimate stress of 526.8MPa and an elastic modulus of

4
72GPa. Specimens for multiaxial fatigue test are in form of funnel with a maximum diameter of 24mm and a minimum

diameter of 12mm. Its specific geometry is indicated in Fig.1.

Fig. 1. Fatigue specimen geometry (Unit: mm).

2.2. Multiaxial fatigue tests with pre-corrosion and pre-deformation

The whole process of tests included 2 stages. Firstly, the specimens were under certain pre-damages which included

single pre-corrosion or pre-deformation as well as pre-deformation + pre-corrosion in sequence. Secondly, the specimens

after pre-damages were subjected to multiaxial fatigue loading. For each condition, 5 specimens were tested to obtain the

distribution of fatigue life.

2.2.1 Pre-corrosion and pre-deformation methods

For the single pre-corrosion treatment, the specimens were soaked in the 3.5% NaCl solution with silicone protection

on both grip ends and the pre-corrosion time varied from 0 to 8 days in 2 days increments as illustrated in reference [10].

For the single pre-deformation treatment, specimens were pre-twisted for 10°, 20°, 30° and 40° at an angular rate of 2°

/min or pre-stretched for 1%, 2%, 3% and 4% at a rate of 0.2mm/min at room temperature.

Concerning the sequential pre-deformation + pre-corrosion treatment, it could be divided into 2 patterns. For the first

pattern, after a fixed pre-deformation level which referred to 30° of pre-torsion angle or 2% of pre-tension strain,

specimens were pre-corroded under different time of 2, 4, 6 and 8 days. For the second pattern, before the fixed

pre-corrosion time of 4 days, different pre-deformation levels were performed on specimens which were same to the above

single pre-deformation treatment.

5
2.2.2 Multiaxial fatigue test

Multiaxial fatigue tests were performed with SDN 100/1000 electro-hydraulic servo fatigue testing machine that

connected with a computer for machine control and data collection. The dynamic loading capacity is in axial

direction and in torsional direction. Sine wave, square wave and triangular wave can be loaded in both axial

and torsional directions. The experiments used von Mises equivalent stress criterion to calculate the axial and tangential

loads. The fatigue tests parameters are listed in the Table 1. Surface conditions were analyzed in a Zeiss microscope.

Microstructure observations were performed with SEM.

Table 1

Loading parameters of multiaxial fatigue test.

Loading wave Equivalent stress Stress amplitude Tension-torsion Loading Environment

form σ ratio phase frequency temperature

Sinusoidal 330MPa 0° 5Hz Room temperature

3. Results and discussion

3.1 Experimental results
3.1.1 Results of pre-corroded multiaxial fatigue

Table 2 shows the results of pre-corroded multiaxial fatigue tests [10]. The fatigue life reduces from 32665 cycles to

13595 cycles with the pre-corrosion time increases from 0 to 8 days.

In this paper, the dispersion of 5 specimens’ fatigue life for each condition was obtained according to Eq. (1).

(1)

In which, D refers to the dispersion, s means the standard deviation of 5 specimens’ fatigue life which can be computed

with Eq. (2) and is the average fatigue life for the specific condition above.

(2)

In the Eq. (2), is the ith specimen’s fatigue life and n is the number of specimens which values 5 in this paper.

Table 2

6
Pre-corroded multiaxial fatigue life results [10].

Pre-corrosion time (days) 0 2 4 6 8

Average life (cycles) 32,665 23,007 17,023 14,648 13,595

Dispersion 8.81% 7.13% 11.91% 12.70% 3.97%

Fig. 2. The average multiaxial fatigue life under pre-corrosion conditions [10].

Fig.2 shows the variation of multiaxial fatigue life with the pre-corrosion time. Multiaxial fatigue life decreases

gradually with the increase of pre-corrosion time. It reveals an adverse effect on the fatigue performance of the material.

3.1.2 Results of pre-deformation multiaxial fatigue

Table 3 shows the average multiaxial fatigue life of the specimens after different levels of pre-deformation. The fatigue

life reduces from 32665 cycles to 13742 cycles with the increase of pre-torsion angle from 0° to 40°. As the pre-tension

strain increases from 0% to 4%, the fatigue life reduces from 32665 cycles to 10700 cycles.

Table 3

Pre-deformed multiaxial fatigue life results.

Non-deformation Pre-deformation

Pre-torsion angle (°) Pre-tension strain (%)

10° 20° 30° 40° 1% 2% 3% 4%

Average life 32,665 25,637 22,894 15,758 13,742 24,621 15,404 12,061 10,700
(cycles)
Dispersion 8.81% 12.49% 10.52% 9.91% 15.26% 14.94% 14.06% 17.35% 17.34%

7
 Fig. 3a and b show the average multiaxial fatigue life of specimens. A gradual decrease trend of multiaxial fatigue life

with the increase of pre-deformation level can be observed. It demonstrates that pre-deformation induces a reduction of

material’s fatigue properties.

(a) (b)

Fig. 3. The average multiaxial fatigue life under pre-deformation conditions. (a) Pre-torsion condition, (b) Pre-tension

condition.

3.1.3 Results of pre-corroded multiaxial fatigue after pre-deformation

The multiaxial fatigue life of pre-deformation specimen after 4 days of pre-corrosion is listed in Table 4. It reveals that

the multiaxial fatigue life decreases with the pre-deformation level. Compared with the results which are listed in Table 2,

with the increase of pre-torsion angle from 0° to 40°, the multiaxial fatigue life reduces from 17023 cycles to 9929 cycles.

Besides, with the increase of pre-tension strain from 0% to 4%, the multiaxial fatigue life reduces from 17023 cycles to

7759 cycles. The dispersion of fatigue life gradually decreases with the pre-deformation level.

Table 4

Pre-deformed multiaxial fatigue life results under 4 days of pre-corrosion condition.

4d of pre-corroded fatigue after pre-deformation

Pre-torsion angle (°) Pre-tension strain (%)

10° 20° 30° 40° 1% 2% 3% 4%

#1 10,266 14,839 13,496 8218 18,184 12,974 7562 6384

#2 12,972 12,626 11,641 11,021 17,739 8302 7341 7216

8
 Specimen #3 19,083 11,769 9271 10,652 10,371 11,095 10,517 9438
life(cycles)
#4 18,285 15,673 10,932 9876 10,681 9483 8035 8471

#5 14,864 10,008 10,895 9878 13,185 6796 7865 7286

Average 15,094 12,983 11,247 9929 14,032 9730 8264 7759
life(cycles)
Dispersion 21.76% 15.98% 13.59% 10.86% 26.74% 24.70% 15.58% 15.43%

Fig. 4a and b show the variation of average fatigue life of pre-deformation specimen under 4 days of pre-corrosion

condition. It demonstrates a decrease tendency of the multiaxial fatigue life of the specimen with the increase of the

pre-deformation level. The fatigue properties of the material decline with the pre-deformation level.

(a) (b)

Fig. 4. Average fatigue life of pre-formed specimens under 4 days of pre-corrosion. (a) Pre-torsion condition, (b)

Pre-tension condition.

The pre-corroded multiaxial fatigue life of 2024-T4 aluminum alloy after pre-deformation is listed in Table 5. It shows

that the multiaxial fatigue life decreases with the pre-corrosion time. Compared with the results which are listed in Table 3,

for the 2% of pre-tension, the multiaxial fatigue life reduces from 15404 cycles to 7946 cycles with the increase of

pre-corrosion time from 0 day to 8 days. Furthermore, for the 30° of pre-torsion, the multiaxial fatigue life reduces from

15758 cycles to 10041 cycles with the increase of pre-corrosion time from 0 day to 8 days. In addition, the dispersion of

fatigue life gradually decreases with the increase of pre-corrosion time.

9
Table 5

Pre-corroded multiaxial fatigue life results under different pre-deformation conditions.

Pre-corroded fatigue after 2% of pre-tension Pre-corroded fatigue after 30° of pre-torsion

2d 4d 6d 8d 2d 4d 6d 8d

#1 9703 12,974 9084 9583 17,397 13,496 10,573 11,471

Specimen #2 14,059 8302 8573 7624 12,832 11,641 12,231 9482

life(cycles) #3 8250 11,095 11,794 6716 11,983 9271 11,748 10,725

#4 10,924 9483 7271 8093 15,460 10,932 9354 9476

#5 13,904 6796 7528 7714 15,753 10,895 9834 9051

Average 11,368 9730 8850 7946 14,685 11,247 10,748 10,041
life(cycles)
Dispersion 22.58% 24.70% 20.40% 13.16% 15.16% 13.59% 11.40% 10.11%

Fig. 5a and b show the variation of pre-deformed specimens’ average fatigue life under pre-corrosion condition. It’s

indicated that the average fatigue life decreases with the increase of pre-corrosion time.

(a) (b)

Fig. 5. Average fatigue life of pre-deformation specimen under pre-corrosion condition. (a) 2% of pre-tension condition, (b)

30° of pre-torsion condition.

During the process of deformation, the specimens receive energy generated by the external force. Due to the

temperature difference between the around environment and the specimen, the majority of energy dissipates into outside

through conduction. However, a small part of the induced energy is stored in the material in form of residual internal stress

10
and strain. The uneven deformation of macroscopic and crystal particle scale, even the metastable force neighboring the

grain boundary or slip plane which is caused by dislocation or point defect are the origins of residual internal stress. The

metastable force, which is called static distortion, is generally considered as the main form of energy storage. According to

the Gutman model of mechanical-chemical interaction in metal corrosion [28-30], the existence of energy storage increases

the internal energy of the metal, which is favorable for the oxidation of metal, manifesting as the higher anodic dissolution

rate of the pre-deformation specimens. Zhang et al. [31] investigated the impact of pre-deformation on stress corrosion of

aluminum alloy 2519A plate. It indicated that pre-deformation both accelerated the anodic dissolution rate and decreased

the electrochemical impedance. Defects which were active places in electrochemical corrosion could lead to the decrease of

corrosion potential . He et al. [19] proved that in grain and twin boundaries, the coupling of deformation twinning and

stress concentration increased the dissolution trend, and the twinned area might prefer to be corroded. Pre-deformation

increases the dislocation density and forms the dislocation tangle around the sub-grain to improve the corrosion driving

force, which reduces the corrosion resistance of metal. Therefore, the probability of pitting increases greatly.

Pitting, which is mainly caused by localized corrosion, can lead to the formation of tiny holes on the metal surface.

The driving force for pitting corrosion is caused by the local potential difference. In which, a small area of bare-metal

becomes anodic and the surrounding area behaves as the cathode. A local galvanic cell is formed and metal starts to dissolve

into the potential corrosive environment. The formation of corrosion pits increases the stress concentration which is

beneficial for the fatigue crack initiation.

Aluminum alloys are sensible to NaCl solution. The presence of anions aggravates the conditions for the

formation and growth of pits which deteriorate the fatigue performance. In the process of fatigue loading, the mechanism is

always the same as anodic dissolution. The only difference is that now it takes fatigue cycles to activate the anodic

dissolution of the crack tip. The stress loading breaks the passive film in the bottom of pit where the crack initiates. The

bare-metal and surrounding area form a galvanic cell to conduct future dissolution. This process is done at each cycle.

Therefore, the material is not given the chance to re-passivize permanently which results in the increased crack growth rate

11
of corrosion fatigue [32].

3.2 Cyclic loading curve analysis

The cyclic hardening and softening are caused by the variation of material’s flow plasticity in the multiaxial loading

process. These phenomena reflect the changes in the material’s resistance to deformation. Under the same loading

conditions, axial displacement amplitude and sheering torsional angle amplitude decreases with time, which indicates cyclic

hardening and the opposite tendency is cyclic softening. Taking specimens with 0°, 20° and 40°of pre-torsion and under 4

days of pre-corrosion condition as example, axial displacement amplitude variations with time are shown in Fig. 6a-c. For

the specimens without pre-torsion, displacement amplitude decreases at the initial loading stage as shown in Fig. 6a, which

indicates axial cyclic hardening. Then alternate hardening and softening can be observed in the axial direction. Fig. 6b and c

show the axial displacement amplitude variations of pre-deformed specimens with pre-torsion angles of 20° and 40°. At the

beginning stage, cyclic softening occurs immediately and then alternate hardening and softening follows. Finally, the

specimen softens to rupture.

In the process of multiaxial loading, the increase of dislocation density will promote the dislocation tangle to increase

the movement resistance which leads to the cyclic hardening phenomenon. Besides, the stress concentration occurs at the

points of dislocation tangle and it will be released with the initiation of micro-cracks, which can result in the cyclic

softening. Therefore, alternate hardening and softening occur during the process of multiaxial loading. The surface defects

and micro-cracks caused by pre-torsion can be channels for erosive ions like contained in the corrosion solution,

which lead to more serious damages. Thus, for the pre-deformation conditions shown in Fig. 6b and c, larger displacement

amplitude accompanying with the constant loading in the beginning stage demonstrates cyclic softening.

12
 (a) (b) (c)

Fig. 6. Axial displacement amplitude variation under different pre-torsion angles and 4 days of pre-corrosion. (a) 0°of

pre-torsion angle, (b) 20° of pre-torsion angle, (c) 40° of pre-torsion angle.

(a) (b) (c)

Fig. 7. Sheering torsional angle amplitude variation under different pre-torsion angles and 4 days of pre-corrosion. (a) 0° of

pre-torsion angle, (b) 20° of pre-torsion angle, (c) 40° of pre-torsion angle.

Fig. 7a-c show the variation of sheering torsional angle amplitude with time under 0°, 20° and 40° of pre-torsion for

pre-corrosion fatigue specimens. The platform holding area where the torsional angle amplitudes remain generally constant

can be observed in the figures. The amplitude of the torsion angle in the platform stage is gradually raised with the increase

of pre-torsion angle. It indicates that under the same loading condition, the damages caused by the tangential load are

enhanced. In the process of multiaxial loading, the tangential load plays an important role in the fatigue cracks initiation.

With the increase of the pre-torsion angle, the defects on the surface of specimens are more serious and the stress

concentration is enhanced, which shortens the crack initiation life.

Fig. 8a-c show the variation of axial displacement amplitude with time for specimens after 0%, 2% and 4% of

pre-tension strain as well as 4 days of pre-corrosion. Similar with the condition of pre-torsion, cyclic hardening and

softening occur alternately. For the conditions of 0% and 2% of pre-tension, cyclic hardening appears at the initial loading

stage while cyclic softening occurs in the condition of 4% of pre-tension.

13
 (a) (b) (c)

Fig. 8. Axial displacement amplitude variation under different pre-tension strain and 4 days of pre-corrosion. (a) 0% of

pre-tension, (b) 2% of pre-tension, (c) 4% of pre-tension.

(a) (b) (c)

Fig. 9. Sheering torsional angle amplitude variation under different pre-tension strain and 4 days of pre-corrosion. (a) 0% of

pre-tension, (b) 2% of pre-tension, (c) 4% of pre-tension.

The variations of sheering torsional angle amplitude with time are shown in Fig. 9a-c for pre-corrosion fatigue

specimens after different levels of pre-tension strain. Similar with the feature of pre-twisted specimens, with the increase of

the pre-tension strain, the torsional angle amplitude is raised, resulting in the enhancement of the destructive effect of the

tangential load. Therefore, a decrease of the multiaxial fatigue life is generated.

3.3 Morphology analysis

3.3.1 Microscopic morphology of the specimen surface

The effect of pre-deformation on the specimens’ surface was estimated by the microscopic observation before

corrosion test. Surface conditions of as-received specimens and pre-deformed specimens are shown in Fig. 10a and b,

respectively. Obviously, the pre-deformation has a notable influence on surface state. It can be seen from Fig. 10a that stria

of the specimen are along the “L” axial direction. However, the pre-deformed specimen to a pre-torsion angle of 30°

14
exhibits a certain angle to the “L” axial direction and surface becomes quite rugged as presented in Fig. 10b. It indicates that

the specimen undergoes plastic deformation. Also, the surface conditions are important. In general, a rough surface is more

susceptible to pitting and exhibits a lower pitting potential [33]. Rougher surface, in fact, has more occluded sites, which

can sustain the required conditions to activate metal dissolution.

Fig. 10. Surface morphology of specimens without corrosion. (a) As-received, (b) Pre-deformed to an angle of 30°.

Surface morphologies of specimens with 30°of pre-torsion under different corrosion times are presented in Fig. 11a-d.

With 2 days of pre-corrosion, the surface is slightly corroded with corrosion pits of small sizes disperse as shown in Fig. 11a.

For the 4 days of pre-corrosion condition, the number of corrosion pits increases and the distribution becomes denser as

shown in Fig. 11b. The pits also have a tendency of connection. Under 6 days of pre-corrosion, corrosion conditions

continuously deteriorate as shown in Fig. 11c. Corrosion pits start to connect to form larger pits, which reduce the fatigue

performance of material seriously. Fig. 11d shows the condition with 8 days of pre-corrosion. The surface is badly corroded

with lots of large and wide defects caused by the pit connection. Besides, from the Fig. 11, it can be seen that the corrosion

pits are produced along the stria direction. The plastic deformation is beneficial for the production of corrosion pits. Once a

pit forms, it is very possible to grow and become a crack in the process of multiaxial loading since stress concentration will

occur around the pit.

15
Fig. 11. Surface morphology of pre-deformed specimens to an angle of 30° under different pre-corrosion time. (a) 2 days of

pre-corrosion, (b) 4 days of pre-corrosion, (c) 6 days of pre-corrosion, (d) 8 days of pre-corrosion.

3.3.2 Microscopic morphology of the fracture

Fig. 12a-d show the typical microscopic fracture morphology of pre-corrosion multiaxial fatigue under different

pre-deformation condition. Fig. 12a and b are the fractures of pre-corroded specimens after 2% of pre-tension. As shown in

Fig. 12a, the characteristic of multiple cracks source around the circumference is observed which is indicated by arrows.

Surface defects caused by pre-deformation and pre-corrosion induce more stress concentration, which is beneficial for the

fatigue crack initiation. In Fig. 12b, the “mud crack” morphology circled by the white dotted line is observed. Fatigue

striations also exist, which are caused by the open and close function of the crack tip during the loading process. Cracks are

possible to be observed between striations. Fig. 12c and d indicate the fractures of pre-corroded specimens after 30° of

pre-torsion. Fig. 12c shows a morphology which contains torsional and tensional features. The left side of the white line is a

16
flat morphology due to friction effects caused by the shear stress and the right side is the ductile dimple features caused by

the tension-compression stress which are pointed by white arrows. Under these circumstances, tension stress would open the

micro-crack and release the friction which allows the micro-crack to grow easier. It’s indicated in Fig. 12d that two fracture

platforms appear which both have regular sections with sliding traces. Torsional stress induces local brittle fracture and

corresponding parts are pulled out along the sheer direction. In this area, some fuzzy fatigue striations and tire track signs

also exist. The formation of tire characteristic is caused by two reasons. On the one hand, the relative movement of the

mating surface makes mismatching between elevations and valleys during crack propagations. It may lead to continuous

hammering between pikes during the unloading phase. On the other hand, hard particles that appear in the material will

hammer opposite side due to surfaces offset. However, the fatigue striations and tire characteristics are worn out gradually

by the torsional stress.

Fig. 12. Micro-morphology of multiaxial fatigue fracture under 4 days of pre-corrosion after different pre-deformation. (a)

17
and (b) 2% of pre-tension, (c) and (d) 30° of pre-torsion.

Fig. 13a and b show the crack source morphology. A corrosion pit circled by the white dotted line is shown in Fig. 13a.

The pit caused by chemical corrosion is very severe and about 0.3mm deep which makes it be a real crack. The penetrating

ion erodes continuously through the corrosion pit. Fig. 13b is the magnified view of the area in the small rectangle

marked (1) in Fig. 13a, and a “mud crack” morphology appears which is similar with the part circled by the white dotted

line in Fig. 12b. Actually, this kind of feature is caused by the combined effect of chemical corrosion and stress loading. In

Fig. 14, ion can be detected in this area which shows the influence of corrosion. While striations that prove the role of

fatigue load can be observed in Fig. 12b.

Fig. 13. Morphology of crack source. (a) Morphology of corrosion pit, (b) The morphology of the area (1).

Fig. 14 shows the energy spectrum analysis result of the area circled by the frame in Fig. 13b. The mass fraction is

56.86% for oxygen and 30.26% for aluminum. It indicates that the component is oxide of aluminum. Besides, ion

appears in this area for the reason of its strong penetration. Therefore, this area is constituted by the corrosion products.

18
 Fig. 14. Energy spectrum analysis result of the corresponding area in Fig. 13(b)

4. Multiaxial pre-corrosion fatigue life prediction with pre-deformation

From results of multiaxial fatigue tests, it can be referred that all pre-damages mentioned in experimental results have a

negative effect on specimen life. However, these pre-processes would occur simultaneously and randomly. In order to figure

out the combined influence of both pre-corrosion and pre-deformation, it’s necessary to define the damage factors for

specimens under multiaxial fatigue after pre-corrosion and pre-deformation separately.

In this paper, the damage factor is defined as logarithm of the ratio between the multiaxial fatigue life with pre-damage

and without pre-damage. In particular, for pre-corroded multiaxial fatigue, the damage factor of pre-corrosion is

defined as:

(3)

In which, is the multiaxial fatigue life after t days of pre-corrosion, and is the multiaxial fatigue life with

no pre-damage under the same equivalent stress. Based on experimental results in previous work [10], the damage factors of

pre-corrosion are computed in Table. 6 and Fig. 15 shows the variation of with pre-corrosion time t. With the

augmentation of pre-corrosion time, increases, which corresponds the experimental results.

Table 6

Variation of with pre-corrosion time t.

Pre-corrosion time (days) 0 2 4 6 8

19
 0 0.1522 0.2830 0.3483 0.3807

Fig. 15. Relation between and pre-corrosion time t.

Besides, for pre-deformed multiaxial fatigue, the damage factor of pre-deformation is defined as:

(4)

In which, is the multiaxial fatigue life after different levels of pre-deformation x, and x refers to pre-tension

strain ε or pre-torsion angle θ in this paper. is the multiaxial fatigue life with no pre-damage under the same equivalent

stress. Based on experimental results, the damage factors of pre-deformation are listed in Table 7 and with different

pre-deformation levels are shown in Fig. 16a and b. As the pre-deformation level raises, the damage factor tends to be

larger.

Table 7

Damage factors of pre-deformation.

Pre-deformation Pre-torsion angle θ (°) Pre-tension strain ε (%)

x 10° 20° 30° 40° 1% 2% 3% 4%

0.1052 0.1544 0.3166 0.3760 0.1228 0.3264 0.4327 0.4847

20
 (a) (b)

Fig. 16. Relation between and pre-deformation x. (a) Pre-tension strain, (b) Pre-torsion angle.

As the damage factors for pre-damage of single condition are obtained, for instance, and . The damage

factor for pre-corroded multiaxial fatigue after pre-deformation can be described as the accumulation of damage factors of

each stage of treatment. Considering the combined effect of pre-corrosion and pre-deformation, an interactive parameter α is

induced. For interactive mode ‘pre-tension + pre-corrosion’, the parameter is -0.3 and for interactive mode ‘pre-torsion +

pre-corrosion’, the parameter is -0.4. Finally, the damage factor of pre-deformation and pre-corrosion can be

derived as:

(5)

And according to the same definition in Eq. (3) and (4), is defined as:

(6)

In which, is the multiaxial fatigue life after certain pre-corrosion time t and pre-deformation level x. From

Eq. (5), the damage factor of pre-deformation and pre-corrosion can be obtained and the results are shown in

Table 8 and Table 9. Once these parameters are calculated, the specimen life can be predicted according to Eq. (6), which is

listed in Table 10 and Table 11 with the comparison of average experimental life.

Table 8

Variation of with pre-corrosion time t.

Pre-corrosion time (days) 2 4 6 8

0.4117 0.5182 0.5735 0.6013

21
 0.3810 0.4797 0.5321 0.5584

Table 9

Variation of with pre-deformation level x.

Pre-deformation Pre-torsion angle θ (°) Pre-tension strain ε (%)

x 10° 20° 30° 40° 1% 2% 3% 4%

0.3192 0.3538 0.4799 0.5285 0.3499 0.5182 0.6107 0.6566

The prediction results are shown in Fig. 17 and Fig. 18a and b. From Fig. 17, it can be seen that with the same 4 days

of pre-corroded time, the predicted results are almost in the 2x scatter band. Besides, the comparisons between predicted life

and experimental life are shown in Fig. 18a and b for different pre-corrosion time. Under the same pre-deformation level,

for instance, 2% of pre-tension strain in Fig. 18a and 30o of pre-torsion angle in Fig. 18b, the predicted results are almost in

the 2x scatter band. Considering the dispersion of multiaxial fatigue data, it’s believed that the prediction method of damage

factor gives fine consequences.

Table 10

Prediction results under the same pre-corrosion time.

Pre-deformation Pre-torsion angle θ (°) Pre-tension strain ε (%)

x 10° 20° 30° 40° 1% 2% 3% 4%

(cycles) 15,094 12,983 11,247 9929 14,032 9730 8264 7759

(cycles) 15,664 14,464 10,820 9673 14595 9905 8005 7203

Table 11

Prediction results under the same pre-deformation level.

Pre-corrosion time (days) 2 4 6 8

(cycles) 11,368 9730 8850 7946

(cycles) 12,658 9905 8720 8180

(cycles) 14,685 11,247 10,748 10,041

(cycles) 13,586 10,820 9594 9029

22
 Fig. 17. Prediction results under the same pre-corrosion time

(a) (b)

Fig. 18. Prediction results under the same pre-deformation level. (a) 2% of pre-tension strain, (b) 30o of pre-torsion angle.

5. Conclusions
(1) The pre-corroded multiaxial fatigue life decreases with the increase of pre-deformation level. Under the same

equivalent stress of 330MPa, while the pre-torsion angle increases from 0° to 40°, the pre-corroded multiaxial fatigue

life reduces from 17023 cycles to 9929 cycles. Moreover, the pre-tension strain increases from 0% to 4%, the

multiaxial fatigue life reduces from 17023 cycles to 7759 cycles correspondently. The multiaxial fatigue life after

pre-deformation decreases with pre-corrosion time. For the same loading condition, as the pre-corrosion time increases

from 0 day to 8 days, the multiaxial fatigue life reduces from 15404 cycles to 7946 cycles after 2% of pre-tension and

23
 reduces from 15758 cycles to 10041 cycles after 30° of pre-torsion. For each condition, the dispersion of fatigue life

has a general decreasing tendency.

(2) Pre-deformation changes the micro-structure of material and increases the corrosion driving force as well as anodic

dissolution rate. Therefore, the probability of pitting raises greatly during the pre-corrosion for the pre-deformed

specimens compared with the as-received specimens. Corrosion pits induce serious stress concentration which has a

great opportunity to initiate fatigue cracks. Pre-deformation and pre-corrosion have a synergistic effect on multiaxial

fatigue damage. The “mud cracks” morphology caused by the combined effect of chemical corrosion and stress loading

appears near the fatigue crack source.

(3) With consideration of both pre-deformation and pre-corrosion effects, the damage factors for each stage of pre-damage

are calculated. Involving the interactive parameter, the damage factor for pre-corroded multiaxial fatigue after

pre-deformation is obtained and is used for life prediction. Fine results are acquired for all experiments in this paper.

Under the same pre-corrosion time, the predicted life with different pre-deformation conditions is almost within 2x

scatter band. In addition, under the same level of pre-deformation, predicted results are close to experimental results

and almost in 2x scatter band in condition of various pre-corrosion time.

Acknowledgement

The authors gratefully acknowledge the financial support of National Natural Science Foundation of China (Grant No.

11502285) and the Fundamental Research Funds for the Central Universities (Grant No. 3122017112).

References
[1] Szusta J, Seweryn A. Experimental study of the low-cycle fatigue life under multiaxial loading of aluminum alloy EN

AW-2024-T3 at elevated temperatures. Int J Fatigue 2017; 96: 28-42.

[2] Jones K, Hoeppner DW. Prior corrosion and fatigue of 2024-T3 aluminum alloy. Corros Sci 2006; 48: 3109-3122.

24
[3] Menan F, Hénaff G. Influence of frequency and exposure to a saline solution on the corrosion fatigue crack growth

behavior of the aluminum alloy 2024. Int J Fatigue 2009; 31: 1684-1695.

[4] Lok S K, Paul J M, Upendranath V. Prescience life of landing gear using multiaxial fatigue numerical analysis.

Procedia Eng 2014; 86:775-779.

[5] Esmaeili F, Chakherlou T N, Zehsaz M. Prediction of fatigue life in aircraft double lap bolted joints using several

multiaxial fatigue criteria. Mater Des 2014; 59(7):430-438.

[6] Itoh T, Nakamura H, Takanashi M, et al. Multiaxial low cycle fatigue life of Ti-6Al-4V under non-proportional loading

with mean strain. Theor Appl Fract Mec 2017; 90: 165-173.

[7] Li XD, Wang XS, Ren HH, Chen YL, Mu ZT. Effect of prior corrosion state on the fatigue small cracking behavior of

6151-T6 aluminum alloy. Corros Sci 2012; 55: 26-33.

[8] Menan F, Hénaff G. Synergistic action of fatigue and corrosion during crack growth in the 2024 aluminum alloy.

Procedia Eng 2010; 2: 1441-1450.

[9] Hu P, Meng Q, Hu W, Shen F, ZhanZ, Sun L. A continuum damage mechanics approach coupled with an improved pit

evolution model for the corrosion fatigue of aluminum alloy. Corros Sci 2016; 113: 78-90.

[10] Chen YJ, Liu CC, Zhou J, Wang XC. Multiaxial fatigue behaviors of 2024-T4 aluminum alloy under different

corrosion conditions. Int J Fatigue 2017; 98: 269-278.

[11] Suresh S, Zamiski GF, Ritchie RO. Oxide-induced crack closure: An explanation for near-threshold corrosion fatigue

crack growth behavior. Metall Mater Trans A 1981; 12A: 1435-1443.

[12] Chen GS, Wan KC, Gao M, Wei RP, Flournoy TH. Transition from pitting to fatigue crack growth—modeling of

corrosion fatigue crack nucleation in a 2024-T3 aluminum alloy. Mater Sci Eng A 1996; 219: 126-132.

[13] Cuesta II, Alegre JM. Influence of biaxial pre-deformation on fracture properties using pre-notched small punch

specimens. Eng Fract Mech 2014; 131: 1-8.

[14] Chiou YC, Yang JK. The effects of pre-deformation on the subsequent fatigue behaviors of SUS 430 stainless steel in

25
 load-control. Int J Solids Struct 2012; 49: 3263-3268.

[15] Park SH, Hong SG, Lee JH, Kim SH, Cho YR, Yoon J, et al. Effects of pre-tension on fatigue behavior of rolled

magnesium alloy. Mater Sci Eng A 2017; 680: 351-358.

[16] Huang GS, Li JH, Han TZ, Zhang H, Pan FS. Improving low-cycle fatigue properties of rolled AZ31 magnesium

alloy by pre-compression deformation. Mater Des 2014; 58: 439-444.

[17] Wang B, Zhang P, Duan QQ, Zhang ZJ, Yang HJ, Pang JC, et al. High-cycle fatigue properties and damage

mechanisms of pre-strained Fe-30Mn-0.9C twinning-induced plasticity steel. Mater Sci Eng A 2017; 679: 258-271.

[18] Wang D, Ma ZY. Effect of pre-strain on microstructure and stress corrosion cracking of over-aged 7050 aluminum

alloy. J Alloys Compd 2009; 469: 445–450.

[19] He JJ, Jiang B, Zhang JY, Xiang Q, Xia XS, Pan FS. Enhancement of mechanical properties and corrosion resistance

of magnesium alloy sheet by pre-straining and annealing. Mater Sci Eng A 2015; 647: 216–221.

[20] Al-Duheisat SA, El-Amoush AS. Effect of deformation conditions on the corrosion behavior of the low alloy

structural steel girders. Mater Des 2016; 89: 342–347.

[21] Lv JL, Guo WL, Liang TX. The effect of pre-deformation on corrosion resistance of the passive film formed on 2205

duplex stainless steel. J Alloys Compd 2016; 686: 176-183.

[22] Zhang ZX, Lin G, Xu Z. Effects of light pre-deformation on pitting corrosion resistance of copper-bearing ferrite

antibacterial stainless steel. J Mater Process Technol 2008; 205: 419-424.

[23] Fatemi A, Shamsaei N. Multiaxial fatigue: An overview and some approximation models for life estimation. Int J

Fatigue 2011; 33: 948-958.

[24] Al-Rubaie KS, Grande MAD, Travessa ND, Cardoso KR. Effect of pre-strain on the fatigue life of 7050-T7451

aluminum alloy. Mater Sci Eng A 2007; 464: 141-150.

[25] Gerin B, Pessard E, Morel F, Verdu C, Mary A. Beneficial effect of pre-strain due to cold extrusion on the multiaxial

fatigue strength of a 27MnCr5 steel. Int J Fatigue 2016; 92: 345-359.

26
[26] Huang Y, Ye X, Hu B, Chen L. Equivalent crack size model for pre-corrosion fatigue life prediction of aluminum

alloy 7075-T6. Int J Fatigue 2016; 88: 217-226.

[27] Ishihara S, Saka S, Nan ZY, Goshima T, Sunada S. Prediction of corrosion fatigue lives of aluminum alloy on the

basis of corrosion pit growth law. Fatigue Fract Eng Mater Struct 2006; 29: 472-480.

[28] Gutman E M. Mechanochemistry of solid surfaces. World Scientific Pub Co Inc, Singapore, 1994.

[29] Gutman E M. An inconsistency in “film rupture model” of stress corrosion cracking. Corros Sci 2007; 49(5):

2289-2302.

[30] Gutman E M. Metal surface charge change due to deformation. J Mater Sci 1996; 31(6):1449-1454.

[31] Zhang XM, Liu L, Ye LY, Liu J, Lei Z, Song JC. Effect of pre-deformation of rolling combined with stretching on

stress corrosion of aluminum alloy 2519A plate. T Nonferr Metal Soc 2012; 22: 8-15.

[32] Milella PP. Fatigue and corrosion in metals. Springer, Italia, 2013.

[33] Sasaki K, Burstein GT. The generation of surface roughness during slurry erosion-corrosion and its effect on the

pitting potential. Corros Sci 1996; 38: 2111-2120.

27
 Highlights

 Pre-corroded multiaxial fatigue life reduces with increase of pre-deformation level

 Introducing a synergistic combination of pre-deformation and pre-corrosion

 Typical fracture surfaces and microstructure morphology can be observed

 A criterion is proposed to predict the pre-damaged multiaxial fatigue life

Graphical abstract

28