You are on page 1of 7

Corrosion Science 50 (2008) 2094–2100

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Electrochemical characterization and CFD simulation of flow-assisted corrosion


of aluminum alloy in ethylene glycol–water solution
L.Y. Xu, Y.F. Cheng *
Department of Mechanical and Manufacturing Engineering, University of Calgary, Calgary, AB, Canada T2N 1N4

a r t i c l e i n f o a b s t r a c t

Article history: An impingement jet system was used to study flow-assisted corrosion (FAC) of 3003 aluminum (Al) alloy
Received 13 January 2008 in ethylene glycol–water solutions that simulates the automotive coolant by corrosion potential and elec-
Accepted 6 May 2008 trochemical impedance spectroscopy (EIS) measurements as well as computational fluid dynamics (CFD)
Available online 13 May 2008
simulation. The effects of solution pH and fluid impact angle on Al FAC were determined. An increase of
solution pH enhances the activity of Al due to dissolution of Al oxide film in alkaline environment. More-
Keywords: over, Al activity decreases with the increasing fluid impact angle to the specimen. A CFD simulation
A. Aluminum
shows that, with the increase of impact angle, the electrode area under high-velocity flow field decreases
B. EIS
C. Flow-assisted corrosion
and that under low-velocity flow field increases. Consequently, the shear stress generated on electrode
surface and the resultant mechanical effect on electrode activity decreases. Therefore, the electrode is
more stable than that impacted at a smaller impact angle. There is an essential role of fluid hydrodynam-
ics in corrosion of Al electrode, which is confirmed by corrosion potential and EIS measurements as well
as CFD analysis.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction shear stress on metal corrosion is very limited. In particular, the role
of sand impact in E–C cannot be quantified. For example, it was ob-
Aluminum (Al) alloys, due to their good thermal conductivity served [8] that most sand particles still deposit in the bottom of cell
and favorable strength-to-weight ratio, have been commonly used when RDE was rotated as high as 8000 rpm.
in a variety of industry fields, including the automotive industry. In Impingement jet system has been used extensively to investi-
particular, aluminum alloys have been broadly integrated into gate the effects of sands on erosion–corrosion of metals in the var-
vehicular heat exchange systems, replacing more traditional mate- ious systems [9–14]. However, to date, there have been few works
rials like steels and copper alloys. However, aluminum alloys are to study the fundamentals of Al FAC and E–C in the automotive
prone to flow-assisted corrosion (FAC) and erosion–corrosion (E– cooling system.
C) in automotive cooling system [1,2]. In this work, an impingement jet system was used to study FAC
In the previous work [3], the synergistic effects of fluid flow and of 3003 Al alloy in ethylene glycol–water solution that simulated
sand particles on corrosion of 3003 Al alloy in ethylene glycol– the automotive coolant by measurements of corrosion potential
water solution were studied through a rotating disk electrode and electrochemical impedance spectroscopy (EIS). The effects of
(RDE). It was found that, in the absence of sand particles, an en- solution pH and fluid impact angle on Al FAC were investigated.
hanced electrode oxidation is achieved due to accelerated oxygen Furthermore, computational fluid dynamics (CFD) software was
diffusion and reduction upon electrode rotation. With the addition used to simulate the hydrodynamic conditions established on
of sand particles, the de-stabilization of electrode is predominant specimen, determining the role of fluid hydrodynamics in FAC of
and the anodic current density of Al increases with the electrode Al alloy. It is anticipated that this work provides an essential in-
rotation speed, which is attributed to the enhanced impact damage sight into the correlation of Al FAC with the fluid hydrodynamic
to electrode by sand particles. conditions in the automotive cooling system.
It has been known [4–7] that rotating electrode techniques, includ-
ing RDE and rotating cylinder electrode, are capable of simulating the
mass-transfer process by controlling the electrode rotating rate, and 2. Experimental
investigating the corrosion of metals in a fluid flow system. However,
its ability to study the effects of fluid impingement and the resultant 2.1. Electrode and solutions

The working electrode for electrochemical tests was made of


* Corresponding author. Tel.: +1 403 220 3693; fax: +1 403 282 8406.
E-mail address: fcheng@ucalgary.ca (Y.F. Cheng). 3003 Al alloy with the chemical composition (wt.%): Cu 0.20, Fe,

0010-938X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2008.05.004
L.Y. Xu, Y.F. Cheng / Corrosion Science 50 (2008) 2094–2100 2095

0.70, Si 0.60, Mn 1.50, Mg 0.05, Cr 0.05, Zn 0.10, Ti 0.05, and Al bal- 2.3. Electrochemical measurements
ance. The specimens were embedded in epoxy resin manufactured
by LECO Epoxy Cartridge System, leaving a circular working area of Electrochemical measurements were performed through a
0.68 cm2. The working surface was subsequently polished with Gamry Reference 600 system. Corrosion potential as a function of
800, 1000 and 1200 grit emery papers, cleaned by distilled water time was recorded. During EIS measurements, a AC disturbance
and methanol. signal of 10 mV was applied at open-circuit potential, with the
The basic test solution was a mixture of 50% (v/v) ethylene gly- measurement frequency ranging from 100,000 Hz to 0.01 Hz.
col + 50% distilled water + 100 ppm Cl solution, simulating the
automotive coolant, with an average pH of 6.8. To investigate the 2.4. CFD simulation
effect of solution pH on Al corrosion, a controlled amount of
0.01 M NaOH solution was added to the solution to obtain solution CFD is a numerical analysis technique based on solving the
pH of 7.5, 8.0 and 8.5, which was measured by OAKTON Acorn pH well-known Navier–Stokes equations, which are formulations of
Meter Kit with an accuracy of ±0.01. mass, momentum and energy conservation laws for fluid flows
[15–17]. The CFD simulations were conducted using a commercial
2.2. Impingement jet system CFD software package COSMOSfloworks 2007 Service Pack, where
the Favre-averaged Navier–Stokes equations were solved. The ini-
The impingement jet system, as shown in Fig. 1, was developed tial and boundary conditions as well as input parameters for CFD
to conduct FAC tests. The system, based on a design by Burstein simulation included:
et al. [9], consisted of a large plastic tank used as a reservoir, a high
pressure pump, flow velocity controller, sand concentration con- (1) Inlet: A flow of ethylene glycol–water (1:1) solution at the
troller, stirrer and valves. When the fluid entered the ejector at velocity of 3 m/s, with a dynamic viscosity of about
high speed, it produced a partial vacuum due to the venturing ef- 2.9  103 Pa s and a density of 1.05  103 kg/m3. The test
fect. The sands underneath the valve could be mixed with the flow- temperature was 27 °C. Turbulence intensity was set at
ing fluid by means of suction. A speed-adjustable mechanical 10%. In particular, j–e turbulence model was used, where
stirrer was used to ensure the homogeneity of the slurry. The ejec- the turbulent kinetic energy, j, was set as 1 J/kg and the tur-
tor, made of stainless steel with 7.6 mm in diameter, was 10 mm bulent dissipation, e, was 1 W/kg for simulation.
away from the test specimen. The detailed installation of specimen (2) Outlet: Environment pressure was 101,325 Pa.
in the test chamber is shown in Fig. 2. (3) Solid wall: The wall could be approximately considered as
An electrochemical cell was incorporated into the test rig to en- an adiabatic wall with roughness of 10 lm, and the surface
able in situ electrochemical measurements. The Al specimen was of Al electrode was also set at 10 lm.
used as working electrode (WE), a saturated calomel electrode (4) Flow type: Turbulent flow as calculated, which was shown
(SCE), which was held inside the test chamber, as reference elec- later.
trode, and a platinum ring as counter electrode. The impact angle
of the flowing fluid to WE, defined as the angle between the surface Fig. 3 shows the element points selected of finite element calcu-
of the specimen and the flow direction of fluid, was adjusted by lation of the average shear stress developed on the electrode. The
rotating the Al electrode. grid resolution was 0.769 mm. The calculation was finished once
The basic FAC tests were performed at a fluid flow velocity of one of the following conditions was satisfied: (a) minimum refine-
3 m/s with an impact angle of 90° for 10 h. ment number of 0; and (b) maximum travels of 4.

Fig. 1. Schematic diagram of impingement jet loop system (RE – reference electrode; WE – working electrode; CE – counter electrode).
2096 L.Y. Xu, Y.F. Cheng / Corrosion Science 50 (2008) 2094–2100

Holder
X=10 mm

d=7.6 mm
V=3 m/s θ
Angle of
Ejector Specimen
impact

Fig. 2. Detail of impacting test parameters in the specimen chamber.

Fig. 3. Element points selected for finite element calculation to measure the shear stress on electrode surface.

-0.6 -0.6

-0.7
Corrosion potential (V, SCE)

-0.7
Corrosion potential (V, SCE)

-0.8

-0.8
-0.9
pH=6.8
30° pH=7.5
-0.9 -1
45° pH=8.0
60° pH=8.5
90° -1.1
-1
-1.2
0 500 1000 1500 2000 2500
-1.1
t(s)
0 400 800 1200 1600 2000
t(s) Fig. 5. Corrosion potential of Al electrode impacted at 900 in the test solutions with
the different pH.
Fig. 4. Corrosion potential of Al electrode at different impact angles in ethylene
glycol–water solution. then performed. Fig. 4 shows the corrosion potential as a function
of time under different impact angles in the basic test solution (pH
6.8). It is seen that corrosion potential of Al electrode increased
3. Results rapidly first, and then reached a relatively steady value. Further-
more, with the increase of impact angle, corrosion potential in-
3.1. Corrosion potential measurements creased (shifted positively).
Fig. 5 shows the time dependence of corrosion potential of Al
To ensure a fresh Al electrode surface with an identical state for electrode impinged at 90° in the solutions with the various pH val-
all tests, a cathodic potential of 1.1 V (SCE) was applied on elec- ues. It is seen that corrosion potentials were shifted negatively
trode for 2000 s, and measurements of corrosion potential were with the increasing solution pH.
L.Y. Xu, Y.F. Cheng / Corrosion Science 50 (2008) 2094–2100 2097

3.2. Electrochemical impedance spectroscopy measurements be reversed back to affect the fluid flow field. Simultaneously,
the reversed fluid from the surrounding wall of the test chamber
Fig. 6 shows the EIS plots measured at the different impact an- can result in the same influence. When fluid impacts at other an-
gles in solutions with the various pH values. It is seen that there gles, the fluid flow is not symmetrical on the specimen surface.
was an identical feature for all EIS plots, i.e., a semicircle over the Thus, the rectangular pattern of shear stress is not obvious.
whole test frequency range. Furthermore, with the increase of Table 1 summarizes the results of average shear stress vs. im-
solution pH, the diameter of semicircle decreased apparently. At pact angle. It is seen that the average shear stress of the electrode
each pH, the size of the semicircle increased with the decrease of increased with the decreasing impact angle.
impact angle.
4. Discussion
3.3. CFD simulation
4.1. Corrosion electrochemistry of Al alloy in ethylene glycol–water
Fig. 7 shows the simulated fluid flow field developed on Al elec-
solution
trode at different impact angles. It is seen clearly that the fluid
fields established at the different impact angles were quite differ-
Generally, the anodic and cathodic reactions of Al alloy in an
ent. In particular, at the impact angle of 30°, most of the electrode aerated, near-neutral aqueous solution are characterized by the
surface was under a fluid flow field of 3 m/s. When the impact an-
Al oxidation and reduction of oxygen:
gle increased to 45° and 60°, the electrode area experiencing low-
velocity flow field of about 1.5 m/s increased. Moreover, the re- Anodic reaction : 2Al þ 6OH ! Al2 O3 þ 3H2 O þ 6e ð1Þ
verse flow field enlarged. At the impact angle of 90°, the whole Cathodic reaction : 3=2O2 þ 3H2 O þ 6e ! 6OH ð2Þ
electrode surface was under a very low flow field of about 1.2–
1.5 m/s, and the flow field was separated into two symmetrical, The dissolved oxygen diffuses towards the electrode surface for the
opposite directions. cathodic reaction which generates hydroxide ions used by the oxi-
Fig. 8 shows the 3D views of flow field developed on electrode dation of Al to form aluminum oxide film, leading to the positive
surface at the different impact angles. The sunken area represented shift of corrosion potential of the electrode, approaching a relatively
the low-velocity field like a basin. It is clear that, with the increas- steady-state value with time.
ing impact angle, the peak area representing the high-velocity flow Upon the increase of solution alkalinity, Al oxide film experi-
field became smaller, and the sunken area enlarged. ences dissolution by
Fig. 9 shows the shear stress distribution on the electrode sur- Al2 O3 þ 2OH ! 2AlO2 þ H2 O ð3Þ
face at different impact angles. At 30°, a uniform shear stress field
with the amplitude of 50–60 Pa was distributed over almost all the The increase of electrode activity with solution pH results in a more
electrode surface. When the impact angle increased to 45° and 60°, negative, steady-state corrosion potential in the solution with a
a zero-shear stress region developed at the edge of electrode sur- higher pH, as shown in Fig. 5.
face, and the area increased with the increasing impact angle. The enhanced electrode activity is also shown by the EIS results.
When the impact angle was 90°, the zero-shear stress region fur- In general, the existence of only one semicircle over the whole fre-
ther increased and concentrated on the center of electrode, with quency range in impedance plot indicates a capacitive behavior of
a low shear stress of about 40 Pa surrounding the central area. Fur- the metal electrode. The size of semicircle corresponds to the
thermore, a rectangular pattern of shear stress was generated at charge-transfer resistance, which is directly associated with the
impact angle of 90°. It is mainly due to the special geometries of interfacial electron-exchange reaction. For an impedance response
the specimen holder and test chamber. The cylindrical specimen with a larger real impedance, i.e., a bigger semicircle, the electrode
was installed in a rectangular platform with a flange above the will be more resistant to corrosion attack. Moreover, the imped-
specimen. When fluid impacts the specimen at 90°, the fluid will ance feature also indicates the rate-controlling mechanism for cor-
rosion process. The presence of one semicircle is correspondent to
the activation-controlled corrosion, while a diffusive Warburg
impedance existing in the low-frequency range indicates that cor-
100000 rosion process is controlled by mass-transfer step [18–20].
The present work shows that, with the increase of solution pH,
the diameter of the semicircle that represents the charge-transfer
pH=6.8
resistance decreases significantly, indicating the enhanced dissolu-
80000
tion of Al electrode at the alkaline solution. Furthermore, the
impedance feature of only one semicircle over the measured fre-
quency range indicates that the electrode process is controlled by
Zimag (Ω cm )
2

60000
activation step, rather than the oxygen diffusion. It could be attrib-
uted to the accelerated mass-transfer process in the flowing fluid
pH=7.5 system. The electrode process is thus controlled by activation step
40000 at all impact angles.

30° 45°
4.2. Hydrodynamic analysis of FAC of Al alloy
20000
pH=8.0 60° 90°
Both corrosion potential and EIS measurements show that there
pH=8.5 are quite different electrode activity and FAC rate of Al at the dif-
0 ferent impact angles.
0 50000 100000 150000 200000 250000
2 Reynolds number, Re, is described as the ratio of inertial forces
Zreal (Ω cm )
to viscous forces and, consequently, it quantifies the relative
Fig. 6. EIS plots of Al electrode measured in ethylene glycol–water solutions with importance of these two types of forces for given flow conditions.
different pH values under the various impact angles. It is also used to identify and predict different flow regimes, such as
2098 L.Y. Xu, Y.F. Cheng / Corrosion Science 50 (2008) 2094–2100

Fig. 7. Fluid flow field developed on Al electrode under different impact angles.

Fig. 8. 3D plots of velocity flow field developed on electrode surface: (a) panoramic view and (b) cropped front view.

laminar or turbulent flow. Laminar flow occurs at low Reynolds by smooth, constant fluid motion, while turbulent flow occurs at
numbers, where viscous forces are dominant, and is characterized high Reynolds numbers and is dominated by inertial forces, which
L.Y. Xu, Y.F. Cheng / Corrosion Science 50 (2008) 2094–2100 2099

Fig. 9. Shear stress generated on electrode surface at different impact angles.

Table 1 even damage the electrode matrix, enhancing the electrode activ-
Average values of shear stress generated on electrode surface at different impact ity. For example, Hussain and Robinson [25] found that the thick-
angles
ness of oxide film on stainless steel is thinned at high turbulent
Impact angle 30° 45° 60° 90° flow range which is associated with high shear stress. Therefore,
Average shear stress (Pa) 50.38 43.68 38.84 33.36 the Al electrode impacted at 30° generates the highest shear stress
over the whole electrode surface (Table 1), resulting in the highest
corrosion activity, which is indicated by the most negative corro-
sion potential (Fig. 4) and the smallest charge-transfer resistance
tend to produce random eddies, vortices and other flow fluctua- (Fig. 6).
tions. Typically it is given as With the increase of impact angle, the electrode area under
qu0 L u0 L high-velocity flow field (3 m/s) decreases, and the low-velocity
Re ¼ ¼ ð4Þ flow field with the value of about 1.2–1.5 m/s dominates the elec-
l m
trode surface, which is clearly shown in Figs. 7 and 8. As a conse-
where u0 is the fluid velocity (m/s), L is the characteristic length quence, the electrode area with a small shear stress (0–10 Pa)
(m), l is the (absolute) dynamic fluid viscosity (Pa s), m is the kine- increases and that with a large shear stress (up to 60 Pa) decreases
matic fluid viscosity (m2/s), and q is the density of the fluid (kg/m3). with the increasing impact angle, as shown in Fig. 9. Finite element
Replace the parameters and variables in Eq. (4) with the present calculation of the average value of shear stress (Table 1) also indi-
conditions, Re is calculated to be about 2  104 for the fluid flow cates that the average shear stress generated on electrode surface
velocity of 3 m/s. The calculated Re value is much higher than the decreases with the increase of impact angle, resulting in a more
threshold value of 2300 for laminar flow [21]. It is thus apparent stable electrode state, as shown in corrosion potential and EIS
that the impingement jet system generates a turbulent flow. measurements.
The fluid impingement on electrode generates a mechanical When the impact angle increases to 90°, the electrode is under
shear stress, s, on the surface, which is the measurement of viscous low-velocity flow field with the value of about 1.5 m/s (Fig. 7). The
energy loss within the turbulent boundary layer and related to the resultant shear stress has the lowest value (Table 1). Therefore,
turbulent intensity acting on the surface [22–24]. The shear stress there is the smallest mechanical effect due to fluid impingement
is expressed as on the electrode state. Consequently, there is the most positive cor-
rosion potential and the largest charge-transfer resistance in
s  0:5fu2 q ð5Þ impedance measurement.
where f is the friction factor which is a function of Re and the rough-
ness of the surface. Apparently, an increasing fluid velocity results 5. Conclusions
in the elevation of shear stress, which is confirmed by the present
CFD simulation shown in Fig. 9. An increase of solution pH enhances the activity of Al in ethyl-
The CFD simulation results show clearly that there are different ene glycol–water solution due to dissolution of Al oxide film in
flow fields and flow velocity on the electrode surface at the differ- alkaline environment.
ent fluid impact angles. At a small impact angle such as 30°, the Fluid hydrodynamics play an important role in Al FAC. The Al
whole electrode surface is under a uniform flow field with a high electrode activity decreases with the increasing fluid impact angle
flow velocity of about 3 m/s, as shown in Fig. 6. The resultant shear to the specimen. With the increase of impact angle, the electrode
stress is thus high and distributes uniformly over the electrode sur- area under high-velocity flow field decreases and that under low-
face (Fig. 9). Shear stress, representing a mechanical effect, would velocity flow field increases. Consequently, the shear stress gener-
thin and remove oxide film and corrosion product deposit, and ated on the electrode surface and the resultant mechanical effect
2100 L.Y. Xu, Y.F. Cheng / Corrosion Science 50 (2008) 2094–2100

on electrode activity decreases. Therefore, the electrode is more [7] V.G. Levich, Physicochemical Hydrodynamics, Prentice-Hall, Inc., Englewood
Cliffs, New Jersey, USA, 1962.
stable than that impacted at a smaller impact angle.
[8] B.R. Tian, Y.F. Cheng, Corros. Sci. 50 (2008) 773.
[9] J.B. Zu, I.M. Hutchings, G.T. Burstein, Wear 140 (1990) 331.
Acknowledgements [10] Y. Li, G.T. Burstein, I.M. Hutchings, Wear 186–187 (1995) 515.
[11] G.T. Burstein, K. Sasaki, Wear 240 (2000) 80.
[12] A. Neville, T. Hodgkiess, J.T. Dallas, Wear 186–187 (1995) 497.
This work was supported by Canada Research Chairs Program, [13] M.M. Stack, N. Corlett, S. Turgoose, Wear 255 (2003) 225.
Natural Science and Engineering Research Council of Canada [14] J.B. Zu, G.T. Burstein, I.M. Hutchings, Wear 149 (1991) 73.
(NSERC) and Dana Canada Corporation. [15] R. Peyret, Handbook of Computational Fluid Mechanics, Academic Press, 1996.
[16] M.J. Berger, P. Colella, J. Comp. Phys. 82 (1989) 64.
[17] C.A.J. Fletcher, Computational Techniques for Fluid Dynamics, vols. 1 and 2,
References Springer, 1997.
[18] F. Mansfeld, J. Appl. Electrochem. 25 (1995) 187.
[1] W.S. Miller, L. Zhuang, J. Bottema, A.J. Wittebrood, P. de Smet, A. Haszler, A. [19] F. Mansfeld, Corrosion 36 (1981) 301.
Vieregge, Mater. Sci. Eng. A 280 (2000) 37. [20] F. Mansfeld, M.W. Kendig, S. Tsai, Corrosion 38 (1982) 570.
[2] M.M. Stack, N. Pungwiwat, Tribol. Int. 35 (2002) 651. [21] C.C. Landreth, R.J. Adrian, Exp. Fluid 9 (1990) 74.
[3] L. Niu, Y.F. Cheng, Wear 265 (2008) 367. [22] B. Poulson, Corros. Sci. 35 (1993) 655.
[4] D.R. Gabe, J. Appl. Electrochem. 4 (1974) 91. [23] B. Poulson, Wear 233–235 (1999) 497.
[5] D.C. Silverman, Corrosion 40 (1984) 220. [24] F. Balbaud-Celerier, F. Barbier, J. Nucl. Mater. 289 (2001) 227.
[6] D.C. Silverman, Corrosion 44 (1988) 42. [25] E.A.M. Hussain, M.J. Robinson, Corros. Sci. 49 (2007) 1737.

You might also like