CHAPTER 1 INTRODUCTION Understanding corrosion processes as they
relate to fatigue is becoming increasingly important as mechanical
systems are consistently moving toward lighter builds with higher
power densities in mechanical components. As the mechanisms of
corrosion fatigue become more defined, engineered metals can be
formulated specifically to increase resistance to corrosion
accelerated fatigue. Standard materials data handbooks offer the
design engineer information on fatigue through many different
testing standards. Testing standards have not been designed to
consider the use of Open Circuit Potential testing during cyclic
fatigue. An idealized test rig configuration will pull together
standards for Open Circuit Potential measurements and fatigue
testing without compromising the control area under study.
1
CHAPTER 2 LITERATURE REVIEW 2.1 Aluminum Corrosion: When looking
at Aluminum from a chemistry standpoint, it is very reactive.
Aluminum is the fourth most reactive metal in the Electromotive
Force Series. Table 2a. Electromotive Force Series for selected
metals [6]. Element Magnesium Aluminum Chromium Copper Copper
Platinum Electrode Reaction Mg Mg++ + 2e Al Al+++ + 3e Cr Cr+++ +
3e Cu Cu++ + 2e Cu Cu+ + e Pt Pt++ + 2e Standard Hydrogen Electrode
Potential -2.34 V -1.67 V -0.71 V +0.345 V +0.522 V +1.2 V
The reason Aluminum is resistant to corrosion is because it
builds protective layers or passivates. Aluminum forms passive
layers at a rapid rate in many environments. The rate or magnitude
of passivation is dependent on the alloy system chosen and the
environment it is subjected to. When passivation occurs, the
aluminum forms an inner barrier layer made of Alumina or Al2O3 and
a hydrated outer layer. This hydrated outer layer has different
compositions depending on the temperature and environment. At low
temperatures Aluminum forms an outer layer of Bayerite, or
aluminum-trihydroxide Al(OH)3 [7].
2
Figure 2a. Passive layer formed on Aluminum [7].
A Pourbaix Diagram can be constructed to determine conditions of
passivation or corrosion in metals. Standard Hydrogen Electrode
Potentials (SHEP) is needed to construct a Pourbaix Diagram. Often
times they are difficult to find and can be calculated from known
affinities of the compounds or extracted from electrochemical test
data. Equation 2a relates Gibbs free energy to SHEP.
Equation 2a. G = nFE
G is the change in Gibbs free energy in Joules for n moles of
electrons transferred. F is Faradays constant = 98485 C/mol e- and
E is the Standard Hydrogen Electrode Potential in Volts. Once the
SHEP is calculated for each possible reaction, the
3
Nernst Equation 2b is then used to relate SHEP to pH. This
relation is then plotted for each reaction to form a Pourbaix
Diagram. [9]
Equation 2b. Ecell = E - (R*T*ln(Q))/(n*F)
Ecell is the Potential for any given point, E is the SHEP for
the reaction, R is the gas constant = 8.314 Joule*K-1*mole-1, T is
absolute temperature, n is the number of moles of electrons in the
reaction, Q is the reaction quotient and F is Faradays constant =
9.65 x 104 Coulomb*mole-1. [9]
4
Figure 2b. Pourbaix Diagram for Al-H2O at Room Temperature
[8].
Information as to the nature of a corrosive environment can be
extrapolated from a Pourbaix Diagram. Below the bottom dashed line,
hydrogen evolution occurs. Above the top dashed line, oxygen
evolution occurs. A pH value less than 5 indicates acidic corrosion
while pH values greater than 9.5 indicate alkali corrosion regions.
Passivation occurs between hydrogen and oxygen evolution and from 5
to 9.5 pH. When inside of the passive region, corrosion is still
possible. Localized pitting corrosion can form.
5
The Mg2Si system in AA 6061 has superior Stress-Crack-Corrosion
(SCC) resistance. No known cases of service related SCC has been
reported for 6XXX series alloys. The Mg2Si is anodic to aluminum
and reactive in acidic environments. [11]
2.2 Aluminum Fatigue: Constant amplitude cyclic loading states
are specified by the maximum and minimum stresses that are defined
at cycle boundaries. Other standard variables used to define cyclic
loading include Alternating Stress (Sa), Mean Stress (Sm) and
stress ratios (R, A). These variables are defined as follows
[1].
Sa = (Smax Smin) / 2
Sm = (Smax + Smin) / 2
R = Smin / Smax
A = Sa / Sm
6
Figure 2c. Constant amplitude cyclic loading [23]
Constant Life Diagrams are commonly used to predict fatigue
failures at conditions other than fully reversed. Construction of a
Constant Life Diagrams requires basic knowledge of material
properties and data from one fatigue failure point. This data is
typically calculated at the fully reversed condition and can be
found in fatigue data books. A Constant Life Diagram with a
Modified Goodman line can be constructed with a materials Ultimate
Tensile Strength and Yield Strength as design criteria [1]. For AA
6061-T6, Su = 45 ksi and Sy = 40 ksi [10]. Fatigue failure for
15,000 cycles in an aqueous environment is estimated at 23 ksi.
This estimate was approximated from data found in a fatigue data
book [22]. Figure 2c is a Constant Life Diagram constructed for
this study. The Modified Goodman Line is indicated by 1 and the
Yield Line is indicated by 2.
7
50 Sa - Alternating Stress (ksi) 40 30 2 20 10 0 -40 -30 -20 -10
0 10 20 30 40 Sm - Mean Stress (ksi) 1
Figure 2d. Constant life Diagram for AA6061-T6 Fatigue failure
without yielding can be achieved above the Goodman line and below
the yield line. Below the Modified Goodman line, fatigue failure
for the desired number of cycles will not likely occur. The
Modified Goodman Equation is as follows [1].
Sa/Sf + Sm/Su = 1
Where Sa is the Alternating Stress, Sm is the Mean Stress, is
the materials Ultimate Tensile Strength and Sf is the failure
stress at the desired number of cycles in the fully reversed
condition when the stress ratio (R) is -1.
8
2.3 Aluminum Corrosion-Fatigue: Corrosion fatigue is based on
the crack growth model where notches are created by intergranular
corrosion or pitting. The notches create localized stress risers
and forms crack tips. The crack growth rates are dependent on
geometry and size of pits for pitting corrosion or grain size and
distribution for inter-granular corrosion. [27]
Aluminum alloy 6061-T6 is susceptible to inter-granular
corrosion in regular service. Susceptibility to inter-granular
attack can be determined by following MIL-H-6088 G [7]. The cause
of inter-granular attack in this alloy system is that the Mg2Si in
the grain boundaries is anodic to aluminum [11]. Corrosion fatigue
is a definite concern with this particular alloy system.
2.4 Electrochemistry: Open Circuit Potential (OCP) or
equilibrium potential is a measure of a metals voltage compared to
a known standard while no current is moving in the cell. The metal
of interest is the working electrode, while the know standard is
the reference electrode. To determine the metals placement in the
electromotive force series, simply add the OCP reading to the
reference electrodes known value within the series. A metals OCP
value can be used to establish ranges to run Potentiodynamic tests.
Changes in OCP of an aluminum alloy under fatigue conditions can
help to identify crack formation and growth
9
as the OCP for a passive layer differs from that of the base
alloy. Changes in OCP curves during fatigue testing can be
correlated to crack growth rates in the specimen.
Potentiodynamic polarization tests are used to extract important
electrochemical data from metals. This type of test utilizes a
working and reference electrode identical to the OCP test. The cell
voltage is regulated through a range of values while the current is
monitored. Tests can cover anodic, cathodic or both regions of a
metal. Tests performed using aerated electrolyte give useful
information as to passive and trans-passive behavior in oxide
forming metals while tests performed in de-aerated electrolyte
typically provide corrosion rates.
2.5 Welding of Aluminum Gas Metal Arc Welding (GMAW) is a fusion
welding process in typical use for joining aluminum alloys. A
consumable bare wire is used for the electrode and an inert gas is
used to shield the weld from the atmosphere. The arc creates a
molten pool of filler and base metal that is 100 to 200 degrees
Celsius above the base metals liquidus temperature [12]. The heat
generated during the welding process dissipates through the
surrounding metal and adjoining surfaces.
10
Figure 2e. Gas Metal Arc Welding [13] Several thermally affected
zones form depending on the temperature profile of the area of
interest. Grain growth, recrystallization and precipitation
evolution can be identified through in-depth microstructural
evaluations. The areas of interest in GMAW welding are the unmixed
zone (UMZ), the partially melted zone (PMZ) and the heat affected
zone (HAZ).
11
2.6 Aluminum Alloy 6061-T6: AA 6061 was developed by the
Aluminum Company of America in 1935. It is a solution heat treated,
artificially aged, marine grade alloy. The use of AA 6061 is of
wide spectrum. It is readily available in a variety of forms, easy
to machine, easy to weld and has very good corrosion resistance
properties [5].
Copper has been added for extra strength and Chromium for extra
strength, grain refinement and corrosion resistance. Mg2Si is the
main precipitate that forms when hardening while CuAl2 contributes
slightly. An excess of Si present in the alloy increases corrosion
resistance.
Table 2b. Nominal Chemical Composition of AA 6061 in percent
[4]. Element Magnesium Copper Chromium Silicon Aluminum Composition
1.0 % 0.27 % 0.25 % 0.6 % Balance
AA 6061-T6 has very good SCC corrosion resistance properties.
ASTM has rated AA 6061-T6 as an A class alloy for resistance to
Stress-Corrosion Cracking using 3.5% NaCl alternate immersion
testing in accordance with ASTM practice G44-99 [3]. Despite
12
its excellent SCC resistance, AA 6061-T6 is susceptible to
inter-granular attacks [7]. Corrosion fatigue can be a cause of
concern with this alloy.
Annealing AA 6061 requires a temperature of 775 degrees F for
2-3 Hours. A solution treatment at 970 degrees F followed by a
coldwater quench is needed to bring the alloy to T4 condition. The
final stage of heat treatment for the T6 condition requires a
precipitation heat treatment. A temperature of 320 degrees F for 16
to 20 hours or a temperature of 350 degrees F for 6 to 10 hours
will accomplish this.[10]
Welded AA 6061 heats up to 577 degrees Celsius when Gas Metal
Arc Welded (GMAW) [12] and from 300 to 475 degrees Celsius when
Friction Stir Welded (FSW) [15]. The process temperatures for both
GMAW and FSW generate well defined microstructural differences
within and adjacent to the weld regions. GMAW welding of AA 6061
requires 4043 filler wire [14]. This is a bare wire consumable
electrode.
2.7 Three-Point Bending: Three point bend tests are standard
tests that can be used to extrapolate various mechanical property
data for a particular material. A specimen is simply supported at
two 13
points while adding a load to the middle. The maximum stress
occurs on the opposite side of the specimen from loading, normal to
the load direction and is tensile. The location of this maximum
stress is commonly referred to as the critical fiber. Reversed
three-point bending was the second choice for these investigations
at UND. The center is simply supported, while the two outside
points apply equal forces. The critical fiber is now located at the
top of the sample, allowing placement of a fluid reservoir for
aqueous tests.
Figure 2f. Reversed Three-point Bending
Using a sample that has a rectangular cross-section allows for
simple calculations of stress at the critical fiber. This holds
true for any stress that is below the materials yield point.
Calculations of stress at the critical fiber for a specimen that is
below the yield point are as follows. max = M x c / I M=L/2xF I =
1/12 x b x h3 Where max is the maximum stress at the critical
fiber, M is the bending moment, c is the critical fibers distance
from the neutral axis, I is the area moment of Inertia, L is the
14
distance between the two end supports, F is the total force
applied to the specimen, b is the length of the specimens base and
h is the specimens height.
Once yield has been exceeded, a plastic hinge will form and the
model can not be assumed linear elastic. Removal of a force after
plastic hinge will result in residual stresses in the specimen. It
is important to avoid plastic deformation during fatigue testing to
ensure the actual stress remains a normal tensile one. Severe
deformations from plastic pileup can lead to complex stress tensors
that can no longer be related to both S-N and -N curves.
2.8 Four Point Bending Four point bending gives an advantage
over three point for the studied application because the loads that
are applied are not in line with the critical fiber. This allows
fully reversed cyclic fatigue testing and will give more access of
maximum stress areas to instrumentation. This is critical if a
reservoir is to be mounted on the specimen. Maximum deflection at
the critical fiber can be found be the equation: [28] x =
P(L-a)/(6LEI)[(L/(L-a))(x-a)^3-x^3 + (L^2-(L-a)^2)x] +
Pa/6LEI[L/a(x-(L-a))^3-x^3 +(L^2-a^2)x] Where P is Loading, L is
the Length of the beam, a is the distance between loading and
support points, x is the measurement to the fiber of interest x=L/2
for center fiber, I is
15
the Area moment of Inertia and E is the modulus of elasticity.
Stress between Loading Points can be calculated as: = Pa/c Where c
is the distance between the neutral axis and the critical
fiber.
Figure 2g. Four-Point Bending [29]
As one could see from the equations, the 4-point bending
presents an opportunity to exploit the uniform nature of stress
between the loading points. This allows the study of an area under
condition rather than a finite section in 3-point bending. To
further restrict 16
the area of study for failure, one could machine the studied
section down to reduce the distance to the critical fiber and thus
increasing the tensile stress. Care should be taken in doing this
as sharp geometrical changes in the specimen could cause notch
effects and become the source of failure. Smooth, rounded
transitions are recommended. This writer strongly recommends
4-point bending as an ideal load configuration for fatigue testing
in aqueous conditions.
2.9 Flexural Bending Flexural bending involves the usage of
rectangular specimens bent in a typical 3-point or 4-point fashion.
Initial investigations and testing relating to this paper at UND
were performed using a Shimadzu Autograph AG-IS machine. Load
overshoot and undershoot can occur with machines that have low
force control response. Other flexural bending methods involve
displacement control. While displacement control will work to
obtain fatigue data, it produces -N data curves instead of the
desired and most commonly reported S-N data curves. It is this
writers professional opinion that flexural bending methods are
inferior to rotating bending methods for this type of
investigation.
2.10 Rotating Bending Rotating bending fatigue operates similar
to that of 3 point or 4 point flexural bending fatigue in that the
critical fiber is located in the middle of the specimen and loads
are 17
applied the same as well. Inherent advantages that rotating
bending has over flexural bending is that specimen geometry is
cylindrical, specimens are easy to manufacture and the applied load
is constant and easy to control. The Aluminum Association has used
rotating bending endurance limits to report fatigue resistance for
over half a century. High cycle rotating bending fatigue tests are
extremely useful to detect the effect of small variations of
material properties on fatigue. [24]
2.11 Surface Roughness Characterization In making a surface
measurement, two parameters are usually considered. The first
parameter, amplitude; is a measure of vertical characteristics that
surface deviations exhibit. The second parameter, Spacing; measures
horizontal characteristics of the surface deviations. Hybrid
parameters can be measured as well, but they are combinations of
amplitude and spacing parameters that are used to find customized
results. Both two and three dimensional measurements can be made,
depending on the type of measurement device.
Several geometric factors may be involved with each type of
measurement. These factors will vary from surface to surface, and a
clear understanding of them is needed to interpret the results of
surface measurements. Taking measurement of roughness without
regard to waviness and lay would be disastrous to the end result.
Refer to figure 1 for a picture of a 18
basic geometry of roughness, waviness and lay. This type of
surface geometry is common with grinding operations.
Figure 2h. Common geometric factors that are used in surface
measurements [17].
When calculating roughness, the most common approach is to
reference the amplitude parameter to an averaged value for the
entire measured surface [18].
Ym = ABS(yi ybar)
19
Where Ym is the measured value, yi is the current amplitude and
ybar is the average amplitude of the entire measured surface. When
considering all measurement points, roughness parameters can be
obtained.
Roughness parameter Ra is an averaged roughness over all points
of measure. This is the most widely used parameter in the world
[18].
Roughness parameter Rz is the 10 point height of the entire
measured span [18].
Stylus-type profilometers are the most commonly used in
industry. It utilizes a mechanical tip to physically drag across
the surface to obtain the measurement. An example of this type of
machine can be seen in figure 2f. The advantages of stylus-type
profilometers are economy, speed, and they can be used on a wide
range of roughness values. One disadvantage to using a stylus is
that only two-dimensional measurements 20
are possible. The physical dimensions of the tip may also
distort the recorded profile of the measurement. Sharp deviations
may be depicted as transitional or missed altogether if the tip is
larger than the pit that is measured. This makes stylus-style
measurements poor for smaller roughness values. Stylus-type
instruments are found widely throughout industry.
Figure 2i. Stylus-type Profilometer [19].
Laser Scanning Confocal Microscopes can offer high resolution,
three dimensional renderings of surface topography. It does this by
sectioning the surface into multiple focal planes. These focal
planes are then rendered together to form a virtual three
dimensional topographical surface. Some offer software packages
capable of three dimensional profilometry readings and can even
allow the user to make stylus type profilometry readings on these
renderings. This feature proves to be a useful tool to get accurate
21
readings perpendicular to lay. UND currently has this capability
at the Advanced Engineered Materials Center. It would be highly
recommended to use such technology to report accurate surface
roughness data of fatigue specimens.
Figure 2j. Laser Scanning Confocal Microscopy [25]
22
Figure 2k. Virtual Profilometry [26]
23
2.12 Design of Experiments Design of Experiments or DOE can be
utilized to maximize the usefulness of information in testing. A
Central Composite Design is typically used for process or operating
point optimization where numerical modeling leaves linearity. The
following is a template of a central composite design.
Table 2c. Central Composite Design [20]X1 -1 1 -1 1 -1.41 1.41 0
0 0 X2 -1 -1 1 1 0 0 -1.41 1.41 0 X1X2 1 -1 -1 1 0 0 0 0 0 X1^2 1 1
1 1 1.99 1.99 0 0 0 X2^2 1 1 1 1 0 0 1.99 1.99 0
For a successful DOE implementation, it is important to apply
statistics and randomize experimental run orders. Replication is
necessary to acquire statistical significance. Each repeated run
adds a single Degree of Freedom to the system. It is also important
to balance the DOE with equal runs at each factorial point
[20].
Each run on the DOE results in a yield or numerical result.
Multiple runs at the same point results in the ability to calculate
statistical significance. The first step to calculating statistical
significance is to calculate the variance of each point. 24
Variance = ((Y-Ybar)2) / DOF
Where the summation encompasses each measured value of Y. Ybar
is the average value at the design point and DOF is the Degrees of
freedom of the point or the number of replicates. Once the
variances are established for each point, the Pooled Standard
Deviation or Sp can be calculated [20].
Sp = ((Variance x DOF) / (DOF))1/2
The summations encompass each experimental point. Once the
Pooled Standard Deviation is calculated, the Standard Deviation of
Effect or SE can be determined [20].
SE = 2 x Sp / Nf(1/2)
Where Nf is the number of factorial points ran, including
replicate runs. Effects and Interactions can now be formulated as
shown below [20].
Ei = (Coded value of i x Ybari) / Nc
25
Where Nc is the number of comparisons made or two times the
number of variables under study. The summation encompasses all
experimental runs. With effects calculated, a Signal-Noise-Ratio
can be obtained [20].
tE = Ei / SE
The variable tE is the Signal-to-Noise Ratio. This value is
compared to the critical value of the Students t distribution to
determine statistical significance. Significant Effects and
Interactions can be put into a mathematical model that encompasses
the entire range of the variable field. This allows contour plots
to be generated to give indication of relative maxima and minima
criterion for process optimization points. The Coefficient of any
variable in a DOE is the Effect or Interaction divided by two.
Through the Law of Inheritance, any variable that is within a
statistically significant Interaction becomes statistically
significant itself [20].
26
CHAPTER 3 EXPERIMENTAL METHODS
3.1 Specimen Preparation: Metallurgical specimens should be
prepared to evaluate microstructural characteristics of the base
alloy system, Friction Stir Weld samples and Metal Inert Gas weld
samples. The base alloy is cut into several samples for screening
different etchants. The samples are then prepared as hot-mounted as
per table 3a and machine polished to mirror finish as per table
3b.
Table 3a. Sample Hot-mount preparation instructions. 1 2 3 4 5 6
7 8 9 Turn on Buehler Simplimet 1000 automatic mounting press
Figure 3a. Verify machine settings are right for the resin system
used 1 min heat time, 2.5 min cool time, 4200 psi pressure, 300
Deg. F Temperature Open mounting cylinder and raise until specimen
can be placed on the head Apply release agent to cylinder head and
place specimen face down Lower cylinder head and put 20 g of
Buehler Phenocure resin in chamber Close the chamber and press
start cycle After cycle is complete, open chamber and raise
cylinder head Remove sample and move on to surface finishing
procedures Table 3b. Metallurgical specimen preparation Machine
polishing 1 2 3 4 5 6 Place 320 grit sandpaper in the Struers
LaboPol-21 machine Figure 3b Adjust sample tensioning rods to apply
moderate force to specimens Insert hot or cold mount specimens into
the slots on polishing head. Lower polishing head until it is
parallel with sanding surface Ensure the polishing head locks into
place and will not move upward Turn on for 5 to 10 minutes. Adjust
water flow as needed to reduce friction and 27
7 8 9 10 11
carry away polishing debris. Inspect specimen and verify it is
uniformly polished Repeat steps 1-7 for 1200 grit and 6 micron
diamond suspension. Rinse with distilled water 3 times Rinse with
isopropyl alcohol 3 times Measure and record surface roughness
using profilometer
Micro-structural characteristics of the samples can be obtained
using etching techniques. For the general grain structure, Kellers
etchant is used with a 20 to 60 second surface swab until
micro-structural differences are visually apparent [16].
Table 3c. Kellers etchant 1 2 3 4 5 2.5 mL HNO3 1.5 mL HCl 1.0
mL HF 95 mL water For color tint etchant, mix 1 part of the above
steps 1-4 with 4 parts water
To analyze Aerated and De-aerated Potentiodynamic Polarization
Scans, electrochemical specimens must be prepared. Samples of the
base alloy, FSW and GMAW should be examined. The samples first need
a thin copper wire attached to the surface using 28
Locktite 3888 2-Part conductive adhesive (Part Number 29840).
After curing, the sample needs to be coated with Ameron Americoat
90HS Pearl Grey Resin system. The system requires a 4 part resin to
one part curing agent ratio. This is performed volumetrically. The
Pearl Grey Americoat resin system prevents crevice corrosion of the
samples. Once this resin cures, the samples can be cold mounted as
per table 3c and hand polished as per table 3d.
Table 3d. Cold Mounting Procedure 1 2 3 4 5 6 7 8 9 Apply
Release Agent to glass sheet. Buehler PN 626-500551 Drill a hole in
the side of phenolic cold mount ring to allow glass tube inside
Insert glass tube into the hole on the phenolic ring Place the
sample inside the phenolic ring on the glass plate Route copper
wire from sample through the glass tube Center the sample to the
middle of the phenolic ring Mix Epoxicure resin and hardener.
Buehler PN 20-8130-032 and 20-8132-08 respectively. Use a 5:1 resin
to hardener weight ratio. Mix thoroughly. Fill the phenolic ring
with prepared resin and allow time to cure. De-tool the glass plate
and sand off excess resin flush with the ring.
29
Table 3e. Hand polishing specimens 1 2 3 Put the sample on flat
table and apply a moderate pressure. Start with 320 grit Move the
sample away from you in a level forward motion while applying force
After 5 min of repetitive strokes in the same direction, check the
sample for uniform abrasion in the direction of motion. If uniform,
continue to step 4. If not 4 5 6 7 8 9 uniform repeat from step 1.
Rotate the sample 90 degrees so that the next step sands
perpendicular Repeat steps 1 through 4 for 240, 0, 00 and 000
sandpapers. Apply diamond paste or desired abrasive suspension to
polishing pad. Turn on polishing wheel and place sample with a
light force on the pad. Rotate the sample CW while polishing for 5
minutes. Rinse 3 times with Distilled water then 3 times with
isopropyl alcohol
Table 3f. Three-point-bending sample preparation 30
1 2 3 4 5 6
Cut Bar-stock material to 5 lengths. Chamfer cut edges. Cut .05
deep circular slot at the center of the sample using a 0.25
diameter ball mill. Sand specimen to appropriate roughness. Measure
sample using the profilometer and ensure the sample conforms to
statistical specifications in section 3.4 Designed Experiment Rinse
sample 3 times with Distilled water then 3 times with isopropyl
alcohol Coat a 0.375 inside diameter ring of Ameron Americoat 90HS
Pearl Grey Resin system on the sample face located at the critical
fiber. The system requires a 4
7
part resin to one part curing agent volumetric ratio. Allow
overnight cure. Glue 0.5 outside diameter, .375 inside diameter
fluid reservoir to cured resin ring. Use 100% silicon adhesive and
ensure glue does not contact the inside of
8
grey resin ring. Allow overnight cure. Attach a thin copper wire
one end of the sample using Locktite 3888 2-Part conductive
adhesive. Allow overnight cure.
3.2 Experimental Setup: In-situ Open Circuit Potential (OCP)
tests were performed using the Shimadzu Autograph AG-IS mechanical
properties test machine and a Gamry Series G-750 potentiostat
driven by Gamry Framework 5 software. The primary operation of
the
31
Shimadzu Autograph AG-IS machine is tensile testing. Exacting
control of crosshead motion on the machine does not allow axial
testing.
A reversed 3-Point Bending test was designed to perform fatigue
tests. The sample was placed exactly center on the fixture. The
Gamry Series G-750 potentiostat was utilized with connection to the
sample as the working electrode. The reference electrode was placed
in the electrolyte. The setup utilized a stand to fix the reference
electrode into place.
Figure 3a. Specimen Configuration
3.3 Fatigue Design Criteria: Using the Modified Goodman
Equation, various maximum and minimum forces were calculated for
the Shimadzu Autograph AG-IS mechanical properties test machine as
can be seen in table 3f. 32
Forces 1200 N and under were eliminated because they were below
the minimum force criteria for crosshead-to-sample mating
stability. Stresses that were above 40 ksi were eliminated for
exceeding the yield strength of the material. The ideal operating
condition was selected from the center of the remaining operating
points.
The forces programmed into the Shimadzu Autograph AG-IS are Fmin
= 1754 N and Fmax = 2959 N. The test machine moves alternately
between these two forces to create a mean stress of 30 ksi and an
alternating stress of 7.67 ksi. The corresponding stress ratio R is
0.59.
Table 3g. Potential operating conditions for fatigue testingSm
(ksi) 7 8 9 10 Sa (ksi) 19.42 18.91 18.40 17.89 Smin (ksi) -12.42
-10.91 -9.40 -7.89 Smax (ksi) 26.42 26.91 27.40 27.89 Fmin (N) -976
-857 -738 -620 Fmax (N) 2075 2114 2152 2191 R -0.47 -0.41 -0.34
-0.28
33
11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
32 33 34 35 36 37 38 39 40 41 42 43 44 45
17.38 16.87 16.36 15.84 15.33 14.82 14.31 13.80 13.29 12.78
12.27 11.76 11.24 10.73 10.22 9.71 9.20 8.69 8.18 7.67 7.16 6.64
6.13 5.62 5.11 4.60 4.09 3.58 3.07 2.56 2.04 1.53 1.02 0.51
0.00
-6.38 -4.87 -3.36 -1.84 -0.33 1.18 2.69 4.20 5.71 7.22 8.73
10.24 11.76 13.27 14.78 16.29 17.80 19.31 20.82 22.33 23.84 25.36
26.87 28.38 29.89 31.40 32.91 34.42 35.93 37.44 38.96 40.47 41.98
43.49 45.00
28.38 28.87 29.36 29.84 30.33 30.82 31.31 31.80 32.29 32.78
33.27 33.76 34.24 34.73 35.22 35.71 36.20 36.69 37.18 37.67 38.16
38.64 39.13 39.62 40.11 40.60 41.09 41.58 42.07 42.56 43.04 43.53
44.02 44.51 45.00
-501 -382 -264 -145 -26 93 211 330 449 567 686 805 923 1042 1161
1279 1398 1517 1635 1754 1873 1992 2110 2229 2348 2466 2585 2704
2822 2941 3060 3178 3297 3416 3535
2229 2267 2306 2344 2383 2421 2459 2498 2536 2575 2613 2651 2690
2728 2767 2805 2843 2882 2920 2959 2997 3035 3074 3112 3151 3189
3227 3266 3304 3343 3381 3419 3458 3496 3535
-0.22 -0.17 -0.11 -0.06 -0.01 0.04 0.09 0.13 0.18 0.22 0.26 0.30
0.34 0.38 0.42 0.46 0.49 0.53 0.56 0.59 0.62 0.66 0.69 0.72 0.75
0.77 0.80 0.83 0.85 0.88 0.91 0.93 0.95 0.98 1.00
3.4 Designed Experiment (DOE): Range of surface roughness
variation was conducted using a Surfcom-480A stylus type
profilometer. The low range of the experimental field was
determined from 6 measured Ra values of a sample that was abraded
on 320 grit paper with light pressure. The high range was
determined from 6 measurements of a sample that was abraded on 240
grit 34
paper with high pressure. The mean and standard deviations of
the measurements were calculated and are shown in table 3g. The
mean to standard deviation ratio of both data sets are less than 1%
different, indicating that linear interpolations of mean and
standard deviation target points within this range are acceptable.
The coded Ra values can be seen in table 3h. Table 3h. Calculated
roughness experimental rangeMeas. # 1 2 3 4 5 6 Mean ==> Sd
=> LOW - 320L Ra in u" 17.68 15.51 16.82 15.9 18.13 14.01 16.34
1.52 HIGH - 240H Ra in u" 53.7 47.55 41.17 52.8 50.5 48.68 49.07
4.52
Table 3i. Coded Ra values for DOE in Coded ==> Mean ==> Sd
=> -1.41 16.34 1.52 -1 21.10 1.96 0 32.71 3.02 1 44.31 4.08 1.41
49.07 4.52
Every sample was checked on a statistical base for conformity to
the roughness with a comparative t test at a 95% Confidence Level.
A t distribution was used to determine acceptance of the sample
[21].
Ho: 1 = 2
35
t = (ybar1-ybar2) / ( Sp x (1/n1 + 1/n2)(1/2))
Where t is the test statistic, ybar1 and ybar2 are the
comparative means, Sp is the standard deviation, n1 and n2 are the
number of measurements taken. Three measurements were taken for
each sample. Samples with standard deviations 10% or greater than
the previously calculated standard deviation were automatically
rejected. The comparative t* value for 7 DOF and 95% confidence is
2.365. Samples that did not conform were reworked and measured
again.
The NaCl solution percent mass (%wt) experimental field ideally
contained 0 to 3.5+ percent concentrations. The experimental key
revolved the experiment around 3.5 percent. . Formulation of the
mixture assumed that one milliliter of H2O equals one gram of H20
and the density of NaCl is 2.16 grams per cubic centimeter. One
liter mixtures were made as shown in table 3j.
Table 3j. Experimental key for percentage composition of
NaClCoded ==> % Conc. => -1.41 0 -1 1.02 0 3.50 1 5.98 1.41
7
Table 3k. Solution mass and volumesNaCl (g) 10.257 35.671 NaCl
(cm^3) 4.75 16.51 H2O (g=ml) 995.25 983.49 Total (g) 1005.51
1019.16 wt% 1.02 3.50
36
61.785 72.735
28.60 33.67
971.40 966.33
1033.18 1039.06
5.98 7.00
A randomized run order was generated. This was performed to
prevent lurking variables that could negatively influence the
experiment unknowingly. The coded experimental settings were
decoded and reordered in Table 3k.
Table 3l. Experimental run orderrun # 1 2 3 4 5 6 7 8 9 Ra - u"
21.10 21.10 16.34 32.71 49.07 44.31 32.71 32.71 32.71 % Conc. 1.02
5.98 3.5 0 3.5 5.98 3.5 3.5 3.5
37
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
30
16.34 21.10 44.31 32.71 16.34 44.31 21.10 49.07 32.71 49.07
32.71 32.71 44.31 21.10 32.71 32.71 21.10 44.31 32.71 32.71
44.31
3.5 5.98 1.02 3.5 3.5 1.02 1.02 3.5 0 3.5 3.5 7 1.02 1.02 3.5 0
5.98 5.98 7 7 5.98
CHAPTER 4 RESULTS AND DISCUSSION 4.1 Three Point Bending: A
sample of AA 6061-T6 was prepared for stand alone fatigue testing.
The sample was loaded with Fmin = 1754 N and Fmax = 2959 N Shimadzu
Autograph AG-IS machine and tested to failure. This occurred
prematurely. Visual inspection clearly indicated severe 38
plastic deformation prior to failure. The failure mode was no
longer tensile but complex. Plastic pileup had occurred and the
test is invalid. Subsequent tests with variations of the loading
were proven fruitless as well. It is clear that AA6061-T651 is too
ductile to test in the original 3-point configuration for low cycle
fatigue.
4.2 In-Situ Open Circuit Potential of Reversed 3-Point Cyclic
Bending: The samples were tested in 0.5 M NaCl solution and
de-ionized water. After a stable OCP reading was achieved, the
Shimadzu Autograph AG-IS machine was started. Specimens lost fluid
from the cell reservoir prior to failure. The sample experience
plastic pileup and deformed. As this had occurred, the protective
paint cracked and the reservoir failed to hold the solution. The
OCP curves were as expected for the short duration the reservoirs
were intact.
4.3 Micro-structural Evaluations Scanning Electron Microscopy
(SEM) can be used to analyze failures and determine root cause.
Below is an image of a steel specimen that was broke via cyclic
torsion. Note the beach marks that can be seen on the surface that
are indicative of classical fatigue failure.
39
Figure 4a. SEM image of torsion fatigue in steel.
Grain structure, distribution, size, shape and even composition
can be determined from the use of various etchants and optical
microscopic techniques. Two samples of AA6061T651 were welded using
a TIG welding process and FSW process. The samples were polished
and treated with Kellers etchant solution as outlined in chapter
3.
Figure 4b. FSW and TIG welded samples treated with Kellers
etchant
40
As can be seen in the photographs, the TIG welded sample did not
fully penetrate the barstock. The grain microstructure is much
larger than in the FSW sample and the transition from HAZ to the
base alloy very abrupt where in the FSW sample, the HAZ
microstructure is not nearly as evolved with smaller grains and is
more transitional to the base alloy. Kellers etchant works well for
AA6061T651 as it attacks the alloy intergranularly. Grains that
evolve will precipitate more into the grain boundaries thus leaving
more distinct traces on the surface.
4.4 Potentiodynamic Testing Polarization curves were generated
for aerated and de-aerated AA 6061-T651 specimens in 0.5 M NaCl
solution. The de-aeration was performed by nitrogen percolation and
aeration with compressed air through the solution for 30 minutes
prior and throughout the test. The samples were prepared as
described in chapter 3.
Figure 4c. AA 6061-T651 Sample prepared for potentiodynamic
testing
41
Figure 4d. Polarization curve for AA 6061-T651 base alloy
The de-aerated curve is as expected for the range of the test.
The aerated sample does not include the needed range. If the sweep
was adjusted to -500 mV for the upper range on both aerated and
de-aerated samples, normal should be readily obtainable. [30]
CHAPTER 5 STANDARDS 5.1 Electrochemical Aspects One of the primary
concerns with electrochemical monitoring is that of solution
chemistry. As a sample is put into a corrosive environment, metal
dissolution into the solution can occur. The rate of dissolution
can be affected by concentration of metal ion in solution. It is
imperative to have an ample amount of solution for these
experiments. In 42
addition to metal ion contamination of the solution, a saturated
calomel electrode is used to monitor OCP readings. This electrode
must be set to have 3 micro-liters per hour leakage to ensure
proper function as per ASTM G5.
The previous configuration used solution reservoirs that held
1.25 ml of solution. This amount is too low. ASTM G5 proposes a
test cell configuration with 900 ml of solution. As a standard
practice, an electrochemical test cell should be designed to hold
1L of solution. Electrodes, samples and other objects will displace
some of this solution.
ASTM G5 calls out the usage of Type IV reagent water as per ASTM
D1193. This standard calls out for the use of DI water with the
following properties.
Table 5a. ASTM D1193 callout for Type IV grade reagent water
Conductivity < 5 uS/cm Resistivity < .2 M/cm pH of 5 to 8
Sodium < 50 ug/L Chloride < 50 ug/L
43
Solutions are to be prepared in a 1 L volumetric flask. To
prepare a proper solution, first add 58.44 x desired molarity in
grams of NaCl to the empty flask. Add ASTM D1193 type IV reagent
water to the solution in part. Heating and agitation may be needed
to dissolve the salt crystals. When the crystals are dissolved,
fill the flask to the point where the bottom of the meniscus is at
the fill line. Ensue the final volume is filled at room temperature
as the previous heating to encourage dissolution expands the water
in the flask. This could give inaccurate solution
concentrations.
The working electrode in this experiment differs from that of
ASTM G5s sample experimental set-up. Ensure the working electrode
is electrically connected to the sample under study outside of the
tested solution.
Prior to testing, the solution should be percolated with 150
cm^2/min of Nitrogen gas as per ASTM G5 for de-aeration or the same
flow rate of compressed air for aerated studies. This ensures a
uniform amount of dissolved gas in the solution for standardization
and reproducibility of the experimental data. This should be
performed for a minimum of 0.5 hrs as per ASTM G5.
The solutions temperate should be 30 +/- 1 degree C as per ASTM
G5. A convenient way to perform this is to mount a glass lined
immersion heater inside the test cell.
44
5.2 Fatigue Aspects When machining the test section, the radius
of curvature for reducing thickness must be equal to or greater
than 8 times the diameter of test section as per ASTM E466. This
gradual transition from one diameter to another is intended to
reduce the stress concentration factor in the necked areas that
would ultimately lead to failures outside the designed area.
Once specimen geometry has been chosen, it is important not to
change this geometry for the duration of the investigation. ASTM
E466 indicates that the final material removal of the specimen is
to be performed parallel to the long axis of the specimen. This can
pose a time consuming task for specimens that are cylindrical and
turned on a lathe. Extensive hand sanding may be needed to ensure
the standard is followed. Test specimens are to be visually
inspected prior to installation in the test machine as per ASTM
E466. Abnormalities such as cracks, scratches and gauges in the
area under study increase the risk of premature failure. This
failure will influence the validity of the experimental data.
5.3 Reporting Data Specimen shape, size and dimensions are to be
reported in accordance with ASTM E468. This information is to
include test machine gripping and specimen orientation. It is
important to record surface preparation technique, roughness
measurements and the time 45
between preparation and immersion into solution for these
experiments as aluminum alloys for passive layers that exhibit
different OCP readings immediately after preparation compared to
several hours after preparation. ASTM E468 stresses the importance
of recording the details of the last material removal operation in
particular.
Machine type, test type and details of the forces used and
waveform shape are reported as per ASTM E468. This is important to
duplicate results and claims. ASTM E468 suggests that S-N curves
and constant life diagrams include the R value of testing, test
frequency, environmental conditions and selected material
properties.
Cycle counting is outlined in ASTM E1049. If an R value of -1 is
chosen, peak reporting and counting is relatively easy. Complex
loadings require a large degree of attention to this standard.
Refer to ASTM E739 for details on statistical analysis. DOE
alone is not a substitution to standardized statistical analysis of
data, but serves as an excellent compliment to reports.
Solution temperature should be recorded at the start of the
experiment and periodically throughout it until the end. One half
hour increments should suffice. ASTM E648 mentions to record the
environmental conditions. Greater detail in environmental condition
reporting is needed as the environment is engineered to influence
the failure. 46
Reference electrode potentials other than saturated calomel and
hydrogen are reported with conversion boxes provided to the reader
as per ASTM G3. Where anodic and cathodic currents simultaneously
exist, ASTM G3 recommends assignment of cathodic current densities
a negative value for differentiation of data. Details for reporting
various polarization plots can be obtained in ASTM G3.
CHAPTER 6 CONCLUSION
This writer has looked at countless test types and possible
configurations to reiterate the original design into a better data
gathering rig. UND does not currently have proper test equipment
for the proposed rig. As this work is a literary review with
recommendations, this writer is not limiting the proposed rig to
resources at hand.
The proposed rig consists of a 4- point rotating-bending fatigue
style test machine. Specimens are now under area of failure rather
than a cross-section. Necking of the specimen will control the area
of failure as will the location of the Americoat paint. The 47
load amplitude control with rotating bending fatigue machines is
excellent. A few concerns arise with the use of this machine for
aqueous testing.
The first concern involves the ability of the specimen to
maintain electrical continuity to the working electrode of the
potentiostat. As the rotating-bending test specimen is turning, it
is difficult to physically maintain contact. A possible solution to
this problem is to press a conductive lubricant packed bearing on
the specimen. The case of the bearing can then be electrically
connected to the potentiostat without adverse effects.
The second concern involves the problem of test machine
iteration with the solution reservoir. One proposed solution to
this problem involves the use of water tight bearings. These
bearings will be placed on the specimen in both locations where the
wall of the reservoir meets the specimen. These bearings must not
support any load. To minimize this risk, a large hole can be cut in
the sides of the reservoir and the hole is sealed with a flexible
rubber matt with the bearing mounted to it.
One other concern involves the speed of the rotating bending
machine. As the machine rotates, it acts as if the solution is a
flowing liquid. A low speed will make this effect minimized, but
experimentation is needed to see how much this actually affects the
testing condition. It could possibly exacerbate frequency related
fatigue effects.
48
Figure 6a. Proposed Test Rig Configuration
With a new proposed test configuration and detailed knowledge on
standards and statistical analysis complete with DOE, one could
successfully implement a very worthy research endeavor into OCP
during cyclic fatigue. Results from this style of testing can be
used to help formulate optimal alloy systems for corrosion
resistance.
49
REFERENCES
[1] R. I. Stephens, A. Fatemi, R. R. Stephens, H. Fuchs. Metal
Fatigue in Engineering, pp. 61,77-78. John Wiley & Sons Inc.
(2001)
[2] M. Fontana. Corrosion Engineering, McGraw-Hill Companies
(1986)
[3] Standard Classification of Resistance to Stress-Corrosion
Cracking of HeatTreatable Aluminum Alloys, ASTM standard G64-99,
Annual Book of ASTM Standards, Vol. 03.02 p. 249-250, ASTM, West
Conshohocken, PA, (2005)
[4] Aluminum Standards and Data, The Aluminum Association, New
York. (1965) 50
[5] Materials Data Handbook Aluminum Alloy 6061, NASA Contract
Number NAS8-26644. National Aeronautics and Space Administration,
Marshall Space Flight Center, Al, (1972).
[6] DOE Fundamentals Handbook - Chemistry, DOE-HDBK-1015/1-93
vol. 2, p. 4 . U. S. Department of Energy, Washington D. C.
(1993)
[7] J. Davis. Corrosion of Aluminum and Aluminum Alloys pp.
23-27,74. ASM International (1999)
[8] C. Bale. EpH Web Pourbaix Diagram Generator
http://www.crct.polymtl.ca/ephweb.php (2006)
[9] J. Talbot. Corrosion Science and Technology. pp. 103-144.
CRC Press (1998)
[10] ALOCA Structural Handbook A Design Manual for Aluminum
Aluminum Company of America. pp. 75, 85. ALOCA (1960)
[11] ASM Handbook Volume 13B Corrosion Materials ASM
International Handbook Committee. pp 100,109. ASM International
(2005) 51
[12] G. Totten, D. Mackenzie. Handbook of Aluminum Vol 1
Physical Metallurgy and Processes pp. 481-491. CRC Press (2003)
[13] D. Kopeliovich. Metal Inert Gas Welding Image of GMAW.
http://www.substech.com/dokuwiki/doku.php?id=metal_inert_gas_welding_mig_gmaw
[14] ALOCA Welding ALCOA Aluminum Aluminum Company of America. p
176, ALOCA (1958)
[15] O. Frigaard, O. Grong, O. Midling. Modeling of Heat Flow
Phenomena in Friction Stir Welding of Aluminum Alloys. P. 8.
Norwegian University of Science and Technology, Dept. of
Metallurgy. Trondheim, Norway
[16] Standard Practice for Microetching Metals and Alloys, ASTM
standard E407-99, Annual Book of ASTM Standards, Vol. 03.01 p. 502,
ASTM, West Conshohocken, PA, (2005)
[17] M.S. Lou, J.C. Chen and C.M. Li, Surface Roughness
Prediction Technique for CNC End-Milling, J. Industrial Technology,
vol. 15, pp. 1-6, Nov 1998 Jan 1999
52
[18] ISO 4287:1997, Geometrical product specifications (GPS)
Surface texture: Profile method Terms definitions and surface
texture parameters
[19] Experimental Study of Interface Behavior Between Composite
Piles and Two Sands United States Department of Transportation -
Federal Highway Administration
[20] J. Lawson, J. Erjavec, Modern Statistics for Engineering
and Quality Improvement pp. 111,163,208-221,788. Wadsworth Group
(2001)
[21] J. Devore, N. Farnum, Applied Statistics for Engineers and
Scientists pp. 366-367, Brookes/Cole (2005)
[22] S. Henry, K. Dragolich, N. DiMatteo, Fatigue Data Book:
Light Structural Alloys pp. 76-77. ASM International (1995) [23]
Engrasp Inc. Website image of constant amplitude stress.
www.engrasp.com/doc/etb/mod/fm1/stresslife/Figure5.gif. Engrasp
Inc. Arlington, TX.
[24] J. Kaufman, Properties of Aluminum Alloys: Fatigue Data and
the Effects of Temperature, Product Form and Processing pp. 1. ASM
International (2008)
[25] University of California, Berkley. Website image of
Confocal Miscroscopy. 53
http://microscopy.berkeley.edu/courses/tlm/clsm/index.html
[26] University of North Dakota, Advanced Engineered Materials
Center. Website image of virtual profilometry.
http://webapp.und.edu/dept/aem/laserscanningmicro.php
[27] R. Buchheit, J. Moran, B. Shaw, Corrosion and Corrosion
Prevention of Low Density Metals and Alloys pp. 220. The
Electrochemical Society, Inc. (2001)
[28] D. Roylance, Beam displacements pp. 10 Department of
Material Science and Engineering Massachusetts Institute of
Technology, Cambrige, MA (2000)
[29] Instru-Met Corporation, Website Image of four point bending
apparatus. http://www.instrumet.com/accessories.htm
[30] T. Monetta, M. Montuori, A. Squillace, F. Bellucci, D.
Mitton The Effect of Heat Treatment and Welding Parameters on the
Corrosion Behavior of a Friction Stir Welded 6056 Aluminum Alloy
Materials and Production Engineering Department University of
Napoli, Engineered Surfaces Center University of North Dakota
54