Investigating the effects of hard anodizing parameters ... - Springer Link

Int J Adv Manuf Technol (2013) 68:453–464 DOI 10.1007/s00170-013-4743-1

ORIGINAL ARTICLE

Investigating the effects of hard anodizing parameters on surface hardness of hard anodized aerospace AL7075-T6 alloy using fuzzy logic approach for fretting fatigue application E. Zalnezhad & Ahmed A. D. Sarhan & M. Hamdi

Received: 17 February 2012 / Accepted: 7 January 2013 / Published online: 31 January 2013 # Springer-Verlag London 2013

Abstract Aerospace applications and energy saving strategies in general raised the interest and study in the field of lightweight materials, especially on aluminum alloys. Aluminum alloy itself does not have suitable wear resistance. Therefore, improvements of surface properties are required in practical applications, especially surface hardness when aluminum is in contact with other parts. In this work, first Al7075-T6 was coated using hard anodizing technique in different parameters condition and the surfaces hardness of hard anodizing-coated specimens were measured using microhardness machine. Second, fretting fatigue life of AL7075-T6 was investigated for both uncoated and hard anodized specimens at the highest surface hardness obtained. Third, a fuzzy logic model was established to investigate the effect of hard anodizing parameters, voltage, temperature, solution concentration, and time on the anodized AL7075-T6. Four fuzzy membership functions are allocated to be connected with each input of E. Zalnezhad : A. A. D. Sarhan (*) : M. Hamdi Center of Advanced Manufacturing and Material Processing, Department of Engineering Design and Manufacture, Faculty of Engineering, University of Malaya, Kuala Lumpur 50603, Malaysia e-mail: [email protected] M. Hamdi e-mail: [email protected] E. Zalnezhad Faculty of engineering, Islamic Azad University, Chalous Branch, Iran e-mail: [email protected] A. A. D. Sarhan Department of Mechanical Engineering, Faculty of Engineering, Assiut University, Assiut 71516, Egypt

the model. The results achieved via fuzzy logic model were verified and compared with the experimental result. The result demonstrated settlement between the fuzzy model and experimental results with 95.032 % accuracy. The hardness of hard anodizing-coated specimens was increased up to 360 HV, while the hardness of uncoated specimens was 170 HV. The result shows that hard anodizing improved the fretting fatigue life of AL7075-T6 alloy 44 % in low-cycle fatigue. Keywords AL7075-T6 alloy . Hard anodizing coating . Surface hardness . Fuzzy logic model

1 Introduction Fretting fatigue is a phenomenon which occurs when the substrate is in contact with other parts while they are subjected to cyclic loads and sliding movements at the same time [1]. Fretting decreased fatigue life of materials drastically. The result of fretting in engineering components under cyclic load is the reduction of life by premature initiation and propagation of cracks within the contact area. Aluminum alloy, which has superior mechanical properties, low cost, light weight, and reliabile, has been widely used for aircraft engines, fuselage, and automobile parts. Aluminum 7075-T6 alloy which is used in this research work has low specific weight and high strength to weight ratio and also high electrical and thermal conductance. This alloy is widely used in industry and in particular in aircraft structure and pressure vessels [2]; however, it is always subject to different working conditions. Wear and fretting normally begin when the substrate is in contact with other

454

Int J Adv Manuf Technol (2013) 68:453–464

surfaces and rubbing each other under normal load, causing share force to act on the surface [3]. Aluminum alloy has problems of surface damage due to its softness and corrosion. Therefore, advances of surface properties are needed in practical applications. Hard anodize coating is widely used for this purpose. This coating is also promising from the standpoint of the possibility of achieving high hardness, strength, and simultaneously good protective and decorative surface properties [4]. Anodizing is an electrochemical process for producing stable oxide films on the surface of metals. Anodic coating can be produced on aluminum by using a wide variety of electrolytes with AC, DC, or a combination of both in order to increase the hardness of metals. All aluminum alloys do not accept hard anodize coatings equally well. Hard anodize coatings on alloys with high copper or silicon content tends to be porous and not very hard. Table 1 lists some of the aluminum alloys that are particularly troublesome; these should be avoided [3, 4]. Pure aluminum coating on the AL7075-T6 using magnetron sputtering technique is a method to create the possibility of hard anodizing performance [5]. To improve surface hardness, it requires investigating surface hardness of hard anodize coating at different parameter conditions. Hence, reliable systematic approaches to investigate the effects of hard anodize coating parameter for best surface hardness is thus required [5]. Soft computing techniques are useful when exact mathematical information is not available. The techniques differ from conventional computing in that it is tolerant of imprecision, uncertainty, partial truth, approximation, and metaheuristics [6, 7]. Components of soft computing include neural networks, fuzzy logics, evolutionary computation, chaos theory, and perceptron. Compared to other artificial intelligence methods, development of fuzzy logic is moderately easier and it does not need many software and hardware resources. Fuzzy logic is one of the soft computing techniques that play an important role in input–output parameter relationship modeling [8, 9]. The fuzzy modeling technique is used when subjective knowledge and suggestion by the expert are significant in defining objective function and decision variables. Fuzzy logic is preferred in investigating the coating performance based on the input variables due to nonlinear condition in the coating process [10]. In this research work, hard anodize coating on Al7075-T6 substrate was carried out in different parameter conditions. Each

parameter has four levels which include voltage, temperature, solution concentration, and time. Fuzzy rule-based method was proposed to investigate surface hardness of hard anodize coating on AL7075-T6 alloy. The fretting fatigue test was performed on hard anodized specimens under the best parameter conditions that were achieved from the experimental results and fuzzy logic method. 1.1 Design of experiments The most important stage in the design of an experiment lies in the selection of parameters and identifying the experimental array. In this experiment, with four parameters and four levels each, the fractional factors design used is a standard L16 (44) experimental array. This array is chosen due to its capability to check the interactions among parameters. The parameters and levels are assigned as in Table 2. The 16 experiments with the details of combination of the experimental levels for each parameter (A–D) are shown in Table 3. 1.2 Experimental details 1.2.1 Material Aluminum 7075-T6 alloy was used in this investigation. The material’s composition was obtained using EDX apparatus as illustrated in Table 4. From a number of tensile tests, the yield stress and ultimate strength of Al7075-T6 were obtained as: σУ =520 MPa and σut =590 MPa, respectively. Two types of specimen (uncoated and hard anodized) for fretting fatigue test are used. 1.2.2 Specimen preparation and fretting parts fabrication Fretting fatigue test specimens were machined with initial surface roughness Ra =0.6±0.1 μm by lathe turning (CNC LATHE MACHINE, Miyano, BNC-42C5). The roundshaped specimens used in this work were prepared in accordance with ISO standard [11]. Fretting fatigue pads were fabricated from AISI 4140 steel plate with hardness of 346 HV. Substrate material (179 HV) is softer than the pads but hard anodize coating (360 HV) is harder than pads. The Table 2 Parameters and levels used in the experiment

Table 1 List aluminum alloys which should be avoided to hard anodizing

Difficult Al alloys for hard anodizing 2011 2017 2024 7075 Cast and wrought alloys with Cu>4 % or Si>7 %

Parameters

A B C D

Voltage (V) Temperature (°C) Solution concentration (%) Time (min)

Experimental condition levels 1

2

3

4

10 0 5 30

20 5 10 60

30 10 15 90

40 25 20 120


455

Table 3 Standard L16 (44) experimental array Experiment

Parameters combination A

B

C

D

1 2 3 4 5

1 1 1 1 2

1 2 3 4 1

1 2 3 4 2

1 2 3 4 3

6 7 8 9 10 11 12 13 14 15 16

2 2 2 3 3 3 3 4 4 4 4

2 1 4 1 2 3 4 3 4 1 4

3 3 2 3 4 1 2 4 3 2 2

2 3 2 4 3 2 1 2 1 4 3

fretting fatigue specimen and the friction pads drawings are illustrated in Fig. 1. The dimensions of drawings are given in millimeter. A ring-type load cell and bridge-type fretting pads was designed and manufactured, which can simulate fretting fatigue conditions. Figure 2 shows a schematic view of fretting fatigue test setup employed in this present study. 1.2.3 Surface treatment All samples for anodizing were initially coated by AL target with purity of 99.99 % using magnetron sputtering machine. The surface of all samples for aluminum coating were polished with SiC papers with grits of 800–2,000; after that, all samples were surface mirrored by diamond liquid and the substrate were ultrasonically cleaned in acetone for 14 min, thoroughly rinsed with distilled water, and dried using nitrogen gas to avoid contamination. An SG Control Engineering Pte Ltd series magnetron sputtering system was used to experimentally deposit thin films of metal. This system contained 600 W RF and 1,200 W DC generators with 4″×12″ electrodes 15 cm away from the target. To easily sputter metals, we designed DC generators. The substrate carrier was circular and was rotatable at various speeds for required cosputtering deposition. The chamber Table 4 Chemical composition of AL 7075-T6 Cu

Si

Mg

Cr

Zn

Mn

1.85

0.47

1.8

0.28

4.6

0.06

(a) Fretting fatigue specimen 5

(b) Drawing of fretting pad Fig. 1 The fretting fatigue specimen and the fretting pad

was evacuated to below 2×10−5 Torr before the argon gas for sputtering was introduced. Here, we used a constant sputtering pressure of 5.2×10−3 Torr. The pure aluminum coating process parameters employed in the present work is shown in Table 5. Furthermore, all pure aluminum-coated specimens are subject to electrochemical conversion of hard anodizing process using the conditions shown in Tables 2 and 3. At the beginning, substrate cleaning are required to remove all unwanted surface contamination to prepare the surface for further processing. Substrate surface finish is created by etching with hot solutions of sodium hydroxide to remove minor surface imperfections. To remove surface oxides, “smut”—which is a combination of intermetallic, metal, and metal oxides remaining on the surface after cleaning/etching, an aqueous solution containing an oxidizing inorganic acid, phosphoric and sulfuric acids, simple and complex fluoride ions, an organic carboxylic acid having 1–10 carbon atoms, and manganese in its oxidation state is used. Finally, a near-mirror finish is created with a concentrated mixture of phosphoric and nitric acids which chemically smooth the surface. After cleaning, aluminum-coated specimens substrate is suspended in electrolytic bath (sulfuric acid at different concentration) as an anode; hence, the current is passed through the bath and oxygen is produced at the anode surface. The equipment used in anodizing process are power supply, electrolytic solution, anode (substrate material), and cathode (stainless steel) as shown in Fig. 3. The oxygen reacts with the substrate to form a thin oxide layer of durable and abrasion-resistant hard anodizing coating; at the same time, hydrogen is formed at the cathode. The anode and cathode chemical reaction are as follows;

456


Acid sulfuric solution

DC power supply

Fig. 3 Schematic of hard anodizing process

natural oxidation. In addition, the hard anodizing coating layers increases the melting point of the substrate surface from approximately 650 °C to approximately 2,000 °C, which is good enough to ensure maintaining the mechanical properties at higher temperature.

(a) Fretting ring A pair of springs to apply the load

Fretting ring Chuck

Friction pad

1.2.4 Surface measurement The layers were characterized using scanning electron microscopy (FE/scanning electron microscope (SEM)-FEG) and focused ion beam techniques (Quanta FEG250). The hardness of the layers was determined using microhardness equipment (HMV Micro Hardness Tester, Shimadzu). The roughness of uncoated samples was characterized with roughness tester machine. The adhesion of the pure aluminum films was determined using a Micro Material Ltd, Wrexham, UK. The data are stored in a digital computer and can be displayed on a screen. 1.2.5 Fretting fatigue test

(b) Rotating bending fretting fatiguetest machine

The specimens were gripped and loaded rotationally in a rotating bending fretting fatigue test apparatus (Fig. 2b). By

Fig. 2 Schematic of fretting fatigue test rig

Anode reaction: 2AL ! 2AL3þ þ 6e 2Al3þ þ ðmetalÞ þ 3H2 O ! Al2 O3 þ 6Hþ

Oxide AL

Electrolyte

Cathode reaction:

O2-

6Hþ þ 6e ! 3H2 ðgasÞ

The schematic of the anodic oxide layer is shown in Fig. 4. The anodizing process produces an oxide layer (coating) which is uniform, much harder, and denser than

AL3+

H 2O

H 2O

Table 5 Pure aluminum coating parameters DC power (W)

Temperature (°C)

DC bias voltage

Time (h)

350

200

75

6

Fig. 4 Schematic of the anodic oxide layer


457

adjusting the load screw on a proving ring with a torque driver, the normal contact load between the contact pads and specimen was controlled. The fretting fatigue tests were carried out at constant average contact pressure of 100 MPa. When a fatigue specimen is subjected to cyclic stresses, fretting between the contact pads and specimen is generated. The samples which were used for fretting fatigue test were uncoated and hard anodized AL7075-T6. Plain and fretting fatigue testing were carried out at room temperature in a two-point loading rotating bending machine (R=−1) under constant stress amplitude at a rotational speed of 2,940 rpm. The nominal maximum cyclic stress was set at value that was expected to result in a fatigue life of between 104 and 107 cycles and test were stopped if the specimen did not fail at 1×107 cycles. The friction force, created by normal force and sliding movement between the specimen and pads, and the friction coefficient were measured by a friction test machine. The amount of friction coefficient between pads (AISI 4140 steel) and AL7075-T6 is calculated at around 0.607. The friction force can be determined from the relation F=μ. In which P is the contact load calculated by ring shape load cell (Fig. 2a) and F is the friction force measured from the friction test machine.

Failure point

559.23µm

826.21µm

Failure point

1.3 Experimental result The film-to-substrate adhesion strength was measured quantitatively using a scratch tester. A diamond indenter (Rockwell type) of 25 μm radius applied an initial load zero onto a sample. The sliding velocity was 5 μm/s. The load was increased gradually by 9.2 mN/s. The scratch’s length during scratch test was 1,186.36 μm. In the scratch test, critical load, Lc, could be used to calculate the adhesion strength. In order to obtain the magnitude of the critical load, acoustic signal, friction curve, and microscope observation were utilized. Acoustic signal produced by the delamination of the film could be used to characterize Lc. Scratch adhesion testing was performed on a coated sample to measure Lc. Scratch force (adhesion) test on a coated sample and the critical load accompany with their force and depth versus distance graphs are shown in Fig. 5. These images basically show the character of failure of pure aluminum coating on AL7075-T6. The hardness of the hard anodized AL7075-T6 surface layers was measured using microhardness equipment. Each measurement is repeated three times and the averages are calculated and summarized in Table 6. Figure 6 shows a typical example of a hard anodize coating at a voltage of 25 V, temperature of 0 °C, solution concentration of 13 %, and time of 80 min; it can be seen under SEM that the coating structure is columnar. There are two types of coating, one pure aluminum on the substrate and hard anodizing coating in two directions (inside and outside of pure aluminum coating). In order to investigate the fretting fatigue life

Failure point

Fig. 5 Scratch force (adhesion) testing on a coated sample and the critical load accompany with their force and depth versus distance graphs

of hard anodized AL7075-T6 alloy in the best parameter condition, some experiments were carried out and the results are shown in Fig. 7b. The experiments were conducted for stress ratio of R=−1 (49 Hz) at a constant contact force of 100 MPa, and working stress amplitudes of 150–300 MPa. The relationship between the stress amplitude and the number of cycles to failure for the all the condition analyzed is defined by Eq. 1 [12]. S ¼ AN

b f

ð1Þ

Where, S is stress amplitude, A is fatigue strength coefficient, b is fatigue strength exponent, and Nf is number of

458


Table 6 Measured surface hardness

2nd

3rd

1 2 3 4 5

241 266 209 185 309

236 273 297 179 297

218 237 243 180 248

230 245 250 181 285

6 7 8 9 10 11 12 13 14 15 16

266 336 181 308 179 210 198 201 204 209 185

273 343 200 383 165 255 183 181 190 221 212

255 400 188 328 204 246 199 220 259 283 166

265 360 190 340 183 237 193 201 218 238 187

cycles to failure. S–Nf curve was obtained by least square fitting relationship in Eq. 1 [12]. Each data point on S/N curve represents the average of five specimens tested under identical conditions. Figure 7a and b show the number of cycle to failure versus stress for plain fatigue and fretting fatigue (uncoated and hard anodized specimens). As it can be seen in Fig. 7b, the fretting fatigue life of hard anodized specimens are improved in comparison to uncoated specimens. Fracture surfaces of tested specimens were examined using optical microscopy. Two typical results fractured surface and cross-section for uncoated and hard anodized specimen are illustrated in Figs. 8 and 9. The figure clearly indicates that the fracture surface consists of two quite distinct regions; a fatigue zone created by crack propagation and a tensile region which gives rise to fracture of specimen when it is sufficiently weakened by the crack zone development. The striations due to each stress cycle can be seen as in Figs. 8b and 9b which show the crack surface of a failed aluminum 7075-T6 at ×40 magnification along with a representation of the stress–cycle pattern that failed it. The occasional large amplitude stress cycles show up as larger striations than the more frequent small amplitude ones, indicating that higher stress amplitudes cause larger crack growth per cycle [13]. 1.4 Fuzzy logic-based model to investigate the effects of hard anodizing parameters on substrate surface hardness The relationship between input parameters which are voltage, temperature, solution concentration, and time with the output

(a) SEM micrograph of hard anodizing on pure aluminum Hard Anodizing

1st

Average surface hardness (HV) 1st

Pure Aluminum

Measured surface hardness (HV)

Substrate

Experiment

(b) SEM micrograph (enlargement of Fig. 5(a))of pure aluminum coating and hard anodizing on AL7075-T6

Fig. 6 SEM micrograph of hard anodizing on AL7075-T6 at voltage of 25 V, temperature of 0 °C, solution concentrate of 13 %, and time of 80 min

parameter which is surface hardness of hard anodized AL7075-T6 were referred to construct the rules. Fuzzy linguistic variables and fuzzy expression for input and output parameters are shown in Table 7. For each variable, four membership functions were used which are low, medium, high, and very high for inputs. The output variable (hardness) also used four membership function, ranging from bad, average, good, and excellent. 1.5 Membership functions for input and output fuzzy variables In choosing the membership functions for fuzzification, the event and type of membership functions are mainly dependent upon the relevant event. In this model, each input and output parameter has four membership functions. Gauss shape of membership function was employed to describe the fuzzy sets for input variables. In output variables fuzzy set, triangular shape of membership functions are used.


459

1.6 Structure of fuzzy rules A set of 16 rules have been constructed based on the actual experimental surface hardness of hard anodize coating on AL7075-T6. Experimental results were simulated in MatLab software on the basis of Mamdani fuzzy logic which was as follows:

(a) S/N curve of plain fatigue for uncoated specimens

(b) S/N curve of fretting fatigue for uncoated and hard anodized specimens Fig. 7 Comparison of S/N curve of fretting fatigue for uncoated and hard anodized specimen

Triangular membership function is generally used and has gradually increasing and decreasing characteristics with only one definite value [8]. The input variables have been partitioned according to the experiment parameter ranges. Membership functions for fuzzy set input voltage, temperature, solution concentration, and time variable are shown in Fig. 10a–d, respectively. Membership functions for surface hardness fuzzy set is shown in Fig. 11.

Fig. 8 Fracture surface and cross-section view of uncoated specimens under fretting fatigue

1. IF (A is L) and (B is L) and (C is L) and (D is L) then (hardness is average) 2. IF (A is L) and (B is M) and (C is M) and (D is M) then (hardness is average) 3. IF (A is L) and (B is H) and (C is H) and (D is H) then (hardness is average) 4. IF (A is L) and (B is VH) and (C is VH) and (D is VH) then (hardness is bad) 5. IF (A is M) and (B is L) and (C is M) and (D is H) then (hardness is good) 6. IF (A is M) and (B is M) and (C is H) and (D is M) then (hardness good) 7. IF (A is M) and (B is L) and (C is H) and (D is H) then (hardness is excellent) 8. IF (A is M) and (B is VH) and (C is M) and (D is M) then (hardness is bad) 9. IF (A is H) and (B is L) and (C is H) and (D is VH) then (hardness is excellent) 10. IF (A is H) and (B is M) and (C is VH) and (D is H) then (hardness is bad) 11. IF (A is H) and (B is H) and (C is L) and (D is M) then (hardness is average) 12. IF (A is H) and (B is VH) and (C is M) and (D is L) then (hardness is bad) 13. IF (A is VH) and (B is H) and (C is VH) and (D is M) then (hardness is bad) 14. IF (A is VH) and (B is VH) and (C is H) and (D is L) then (hardness is bad) 15. IF (A is VH) and (B is L) and (C is M) and (D is VH) then (hardness is average)

Location of friction pads

Tensile zone

Fretting zone

(a) Fracture in uncoated AL7075-T6 specimen

(b) Cross-section view of uncoated specimen

after 2,E+06 cycles at 200MPa stress

under fretting fatigue

460


Fig. 9 Fracture surface and cross-section view of hard anodized specimens under fretting fatigue

Location of friction pads

Tensile Zone Fretting zone

(a) Fracture in hard anodized AL7075-T6 specimen after 8,E+06 cycles at 200MPa stress

16. IF (A is VH) and (B is VH) and (C is M) and (D is H) then (hardness is bad)

(b) Cross-section view of hard anodized specimen under fretting fatigue

2 Discussion 2.1 The effect of hard anodizing parameters on surface hardness

1.7 Defuzzification Defuzzification is the conversion of a fuzzy quantity to a precise value, just as fuzzification is the conversion of a precise value to a fuzzy quantity. Seven methods are available in the literature to be used by researchers for defuzzifying methods: centroid, weight average, mean of max, center of sum, center of largest area, and first (or last) of maximum method. The selection of the method is important and it greatly influences the speed and accuracy of the model. In this model, the centroid of an area defuzzification method is used due to wide acceptance and capability in giving more accurate result compared to the others [14, 15]. In this method, the resultant membership functions are developed by considering the union of the output of each rule, which means that the overlapping area of fuzzy output set is counted as one, providing more result [16]. Figure 12a–c are examples to demonstrate the appropriate assent between parameter change and hard anodize coating surface hardness values predicted by fuzzy based model.

Table 7 Fuzzy linguistic and abbreviation of variables for each parameter Inputs

Range

Parameters

Linguistic variables

A B C D

Low (L), medium (M), high (H), very high (VH)

10–40 0–25 5–20 30–120

Bad, average, good, excellent

180–360

Voltage (V) Temperature (C) Solution concentration (%) Time (min) Output Hardness (HV)

The selection of the hard anodizing conditions is essential for fabricating composite thin films. The most important parameters affecting the deposition rate and surface hardness are the voltage (in volts), temperature (in degree Celsius), solution concentration (in percent), and time (in minutes). From the experimental and fuzzy model prediction results, as it can be seen in Fig. 12a, the surface hardness of hard anodized specimens is low when voltage is at an amplitude of 10–25 V. Voltage below this range produces soft, porous, and thin films. With increasing of voltage from 25 to 35 V, the surface hardness of specimens is increased when temperature is kept constant at 0 °C; while with increasing of voltage up to 40 V, the surface hardness is decreased again. It is attributed to that at low voltage, the movement of ions is slow and less oxygen’s ions separate from cathode, so less aluminum oxide can be constructed on the surface of AL7075-T6 aluminum coated in both direction (inside and outside). As the voltage is increased, the film forms more quickly with relatively less dissolution by the electrolyte, consequently the film is harder and less porous. At very high voltage, there is a tendency for “burning”; this is the development of excessively high current flow rate at local areas with overheating areas. On the other hand, with increasing of temperature, the surface hardness is decreased; it means that the best hardness is achieved at 0 °C. The effect of an increase of electrolyte temperature is directly proportional with the increasing of the rate of dissolution of the anodic film resulting in thinner, more porous, and softer films. Low temperatures are used to produce hard coatings normally in combination with high current density and vigorous agitation. If temperature is increased further, the maximum

Int J Adv Manuf Technol (2013) 68:453–464 Membership function plots

Membership function plots Degree of membership

Degree of membership

Fig. 10 Membership function for input parameters

461

(a) Input variable “A” (Voltage (V))

(b) Input variable “B” (Temperature (ºC)) Membership function plots Degree of membership


Membership function plots

(c) Input variable “C” (Solution concentration (%))

thickness is reduced to lower values due to the higher dissolving power of the electrolyte [17]. The solution concentration also plays important roles in hard anodize coating to get high value of hardness. With mixing the oxygen and aluminum, the surface becomes ceramic and the surface hardness of hard anodized samples are increased with increasing the solution concentration from 5 to 13 %; while with more increase in solution concentration up to 20 %, surface hardness is decreased as it can be seen in Fig. 12b. This may be because the surface was more porous. The effect of increasing solution concentration on the coating characteristic is similar to temperature increase; however, the effect of temperature is more important than that of concentration. The increase in concentration limits the maximum film thickness due to the higher dissolving power of the concentrate solutions. In addition, with increasing the time from 30 to 85 min, when the solution concentration is around 10–14, the surface hardness is increased as it can be seen in Fig. 12c; this is attributed to the construction of aluminum oxide hard film, which can make the coating’s film thicker. But with


Membership function plots

Output variable “Surface Hardness” (HV)) Fig. 11 Membership function for the output parameter

(d) Input variable “D” (Time (min))

more increase in the solution concentration, increasing time can decrease the surface hardness of specimens. Figure 6 shows the cross-section view of hard anodizingcoated specimen at the best parameters condition to obtain the highest hardness. The coating thickness was measured and found to be approximately 18 μm. Observations on surface of coatings indicated presence of a relatively little number of cracks in hard anodizing coating. 2.2 Fretting fatigue and S/N curves The corresponding plain fatigue and fretting fatigue S/N curves at contact pressure of 100 MPa are displayed in Fig. 7a and b, respectively. It is apparent that fretting has a deleterious effect on the fretting fatigue life of AL7075-T6 in substrate and coated conditions at all values of the applied bending stress. Figure 7a shows that fatigue (plain) strength reduces with increasing stress. Figure 7b shows a comparison between fretting fatigue life of uncoated and hard anodized specimens. The reduction of fatigue strength for hard anodizing coated specimens is less than untreated substrate. On the other hand, the trend of the effect of hard anodizing depends on the value of stress. Hard anodize coating has increasing effect on fatigue life of the specimens in low-stress region at approximately 220 MPa. It is obvious that the influence of hard anodizing coating is more performed at lower stress. The no effect of hard anodize coating at higher stress in fretting fatigue life may be a result from early initiation of crack of hard anodizing film due to high-local stress concentration resulting from bulk stress. The increase in fretting fatigue life in low-stress region for conditions considering (substrate hardness, pads material, coating thickness and hardness, and kind of loading) in this study may be attributed to low coefficient of friction that prevents metal to metal contact, which may result in higher fretting fatigue life because of retardation

462


Temperature (˚C) Voltage (V)

(a) Surface hardness in relation to change of Voltage (A) and Temperature (B)

Voltage (V) Solution concentrate%

(b) Surface hardness in relation to change of Voltage (A) and Solution concentrate (C)

It is suggested that a fretting fatigue crack forms at the region where the frictional shear stress on contact surface locally concentrates. Thus, the decrease in fatigue life by the fretting damage is considered to be due to the decrease in crack initiation life caused by the local stress concentration caused by fretting, and the acceleration of the initial crack propagation by fretting [14, 18–20]. As one of the main mechanisms of acceleration of initial crack by fretting, the wedge effect where the wear debris goes into the small initial fretting fatigue crack is considered [15, 21]. However, if the crack is fully filled with the wear debris, it is considered that the effect is decreased because the wear debris cannot go into the crack furthermore. The action of fretting causes considerable damage to the specimen surface. Figures 8a and 9a show the appearance of the fretting scars on substrate and hard anodized specimens. It can be observed that the extent of the fretting damage include hard anodize coating is less than that in the substrate. This effect may be due to increased hardness due to hard anodize coating of the surface. It is clear that during plain fatigue, cracks originate randomly at one or several points around the periphery of the specimen case while during fretting, cracks inevitably start from the same location at point adjacent the leading edge of the fretting areas where the bending stress and the induced shear stress highest. Crack propagation occurs from two sides resulting in the appearance of a final fracture area of the specimen as shown in Figs. 8b and 9b. 2.3 Investigate the fuzzy model accuracy and error After the fuzzy rules were constructed, other new five experimental tests from separated experiment were carried out to investigate the fuzzy model accuracy and error as shown in Table 8. The individual error percentage was obtained by dividing the absolute difference of the fuzzy predicted and measured values as shown in Eq. 2 where ei is individual error, Hm is measured value, and Hp is predicted value [16]. ei ¼

Solution concentrate%

Time (min)

(c) Surface hardness in relation to change of Solution concentration (C) and Time (D)

Fig. 12 The surface hardness obtained by fuzzy logic in relation to parameters change

of crack initiation resulting from lower stress concentration compared to the substrate.

Hm Hp 100% Hm

ð2Þ

Meanwhile, accuracy was calculated to measure the closeness of the fuzzy predicted value to the measured value. The model accuracy was the average of individual accuracy as shown in Eq. 3 where A is the model accuracy and N is the total number of dataset tested. Hm Hp 1 XN A¼ 1 100% ð3Þ i¼1 N Hm The error for dataset result was calculated and the model accuracy for fuzzy logic was determined. The experimental


463

Table 8 Accuracy and error of the fuzzy logic model No of experiment

Parameters (inputs)

Surface hardness result (output) (HV) Measured surface hardness

Error (%)

Accuracy (%)

95.24 96.14

Fuzzy predicted surface hardness

A

B

C

D

1st

2nd

3rd

Average

1 2

15 25

2 3

7.5 12.5

40 50

268 281

195 256

228 317

230 285

241 296

4.76 3.86

3 4 5

27 35 37

15 18 22

16 17 18

70 80 100

193 188 170

231 205 172

207 208 198

210 200 180

217.5 212.5 191.5

3.57 96.43 6.25 93.75 6.40 93.60 Accuracy of model=95.032

condition, surface hardness results, and fuzzy model predicted value are shown in Table 8. The highest percentage of error for

fuzzy model prediction is 6.4 %. The low level of errors shows that the fuzzy surface hardness results were very close with actual experimental surface hardness values. Table 8 also shows that the fuzzy model accuracy is 95.032 %. The value of accuracy shows that the proposed model can predict the surface hardness of hard anodize coating on AL7075-T6 satisfactorily as it can be seen in Fig. 13a–c.

3 Conclusion

(a) Comparison of fuzzy logic model prediction with the experimental results

In this research work, first, hard anodize was coated on AL7075-T6 samples at different parameters condition and hardness of all samples were measured by microhardness machine. The parameters of this study include voltage, temperature, solution concentration, and time. Second, prediction of surface hardness of hard anodizes coating on AL7075-T6 alloy was investigated at same parameters condition using fuzzy logic technique. Third, plain fatigue and fretting fatigue test of two types of specimens, uncoated and hard anodized, were carried out for investigating the fatigue and fretting fatigue life of specimens. From the experimental and computational results, the following conclusions are obtained:

(b) The fuzzy model accuracy percentage

(c) The fuzzy model error percentage Fig. 13 Comparison of fuzzy logic model prediction with the experimental results for surface hardness of hard anodize coating on AL7075T6, the accuracy, and error percentage

1. Fretting decreases the fatigue life of AL7075-T6 alloy drastically. The deduction of the fatigue life is attributed to the introduction of shear stress on the surface though contact between the fretting pads and the substrate. 2. Hard anodize coating improved fretting fatigue life of AL7075-T6 alloy at low stress. However, toward higher stress levels, the extent of increase in fatigue life deduced and at applied bending stress of approximately 210 MPa; it was observed that hard anodize coating in fretting fatigue life nearly has no effect slightly. 3. Hard anodize coating can be used as one method to improve the fretting fatigue life of AL7075-T6 at low service loads.

464

4. Pure aluminum coating using magnetron sputtering technique on the surface of substrate is improved the ability of AL7075-T6 for acceptance of become hard anodized. 5. In the hard anodize coating on AL7075-T6 alloy, using voltage (25–30 V), temperature (0 °C), solution concentration (13 %), and time (90 min) are recommended to obtain the highest surface hardness for the specific test range 360 HV; while the hardness of uncoated samples was 170 HV. 6. The fuzzy model percentages of error and accuracy were found to be 6.4 and 95.032 %, respectively. It is indicated that the fuzzy logic prediction model could be used to predict the surface hardness of the coated thin film of hard anodize coating on AL7075-T6 alloy in a very accurate manner. Acknowledgments The authors acknowledge financial support under the University Malaya Research Grant (grant no.: UM.TNC2/RC/ AET/GERAN (UMRG) RG133/11AET) from the University of Malaya, Malaysia.

References 1. Sadeler R, Atasoy S, Arici A, Totic Y (2009) The fretting fatigue of commercial hard anodized aluminum alloys. J Mater Eng Perform 18(9):1280–1284 2. Majzoobi GH, Jaleh M (2007) Duplex surface treatment on AL7075T6 alloy against fretting fatigue behavior by application of titanium coating plus nitriding. Mater Sci Eng A 452–453:673–681 3. Ezuber HA, Houd E, Shawesh F (2008) A study on the corrosion behaviour ofaluminium alloys in sea water. Mater Des 29:801–805 4. Camargo AM, Voorwald HJ (2007) Influence of anodization on fatigue strength of 7050-T7451 aluminium alloy. Fatigue FractEng Mater Struct 30:993–1007 5. Kelly PJ, Abu-Zeid A, Arnell RD, Tong J (1996) The deposition of aluminum oxide coatings by reactive unbalanced magnetron sputtering. Surface Coat Tech 86–87(1):28–32

Int J Adv Manuf Technol (2013) 68:453–464 6. Chandrasekaran M, Muralidhar M, Murali Krishna C, Dixit US (2009) Application of soft computing techniques in machining performance prediction and optimization: a literature review. Int J Adv Manufac Tech 46(5–8):445–464 7. Shamshirband S, Kalantari S, Bakhshandeh Z (2010) Designing a smart multi-agent system based on fuzzy logic to improve the gas consumption pattern. Sci Res Essays 5(6):592–605, 18 March 8. ASM Jaya, SZM Hashim, and MN AbdRahman (2010) Fuzzy logic-based for predicting roughness performance of TiN coating. 10th International Conference on Intelligent Systems Design and Applications. 978-1-4244-8136-IEEE. 9. Leung RWK, Lau HCW, Kwong CK (2003) An expert system to support the optimization of ion plating process: an OLAP- based fuzzy-cum-GA approach. Expert Sys App 25:313–330 10. Hasmi K, Graham ID, Mills B (2006) Data selection for turning carbon steel using fuzzy logic. J Mater Proc Tech 135(4):44–58 11. ISO Standard (2010) Metallic materials—rotating bar bending fatigue testing. ISO International 12. Basquin LOH (1910) The exponential law of endurance tests, proc. ASTM 10(2):P625 13. Norton RL (2010) Machine design, an integrated approach. Worecester Polytechnic Institute, Worcester, MA 14. Mugadu A, Hills DA, Nowell D (2002) Modifications to a fretting-fatigue testing apparatus based upon an analysis of contact stresses at complete and nearly complete contacts. Wear 252:475–483 15. Fatigue Failures, Failure Analysis and Prevention, Vol 11 (2002) ASM Handbook, ASM International 16. Oktem H, Erzurumlu T, Erzinchanli F (2006) Prediction of minimum surface roughness in end milling mold part using neural network and genetic algorithms. Int J Materials Design 27:735–744 17. Shahzad M, Chaussumier M, Chieragatti R, Mabru C, Rezai-Aria F (2011) Surface characterization and influence of anodizing process on fatigue life of Al 7050 alloy. Int J Mater Des 32:3328– 3335 18. Araujo JA, Nowell D (2002) The effect of rapidly varying contact stress fields on fretting fatigue. Int J Fatigue 24:763–775 19. Jin O, Mall S (2004) Effect of slip on fretting behavior: experiments and analysis. Wear 256:671–684 20. Lin SK, Lee YL, Lu MW (2001) Evaluation of the staircase and the accelerated test methods for fatigue limit distributions. Int J Fatigue 23:75–85 21. Hirata H, Maejima M, Saruwatari K, Shigeno H, Takaya M (1996) Rotational bending fatigue of anodized coating of aluminum of aluminum. J Surf Finish Soc 47(4):376