Engineering Failure Analysis 73 (2017) 46–56

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Failure analysis of multiphase flow corrosion–erosion with

three-way injecting water pipe
Haozhe Jin, Xiaoping Chen, Zhijian Zheng, Guofu Ou ⁎, Wenwen Liu
The Flow Included Corrosion Institution, Zhejiang Sci-Tech University, Hangzhou 310018, China

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

Article history: Water injection is an essential component in a reaction effluent air cooler (REAC) system be-
Received 20 July 2016 cause its primary function is to dissolve the generated ammonium salt, which leads to deposi-
Received in revised form 6 November 2016 tion or blockage accidents. A damage incident in a three-way pipe made of carbon steel under
Accepted 12 December 2016
the multiphase flow field was investigated. The failure analysis was performed by means of
Available online 14 December 2016
scanning electron microscope (SEM) inspection and computational fluid dynamics (CFD) sim-
ulation. CFD results show that a large velocity gradient exists near the area of 5–12d at the bot-
Keywords: tom of the main pipe. This gradient results in a region of low flow velocity, high wall shear
Air cooler system
stress, and high turbulent kinetic energy. The flow state becomes very chaotic, and the non-
Water injection
uniformity coefficient of velocity is high. The corrosive medium (NH4Cl or H2S) dissolving in
Pipe failure
CFD water increases the causticity of fluid medium and aggravates the flow corrosion. The high
Flow corrosion–erosion risk area from the CFD simulation coincides with the breakage area of the three-way pipe on
the spot. This failure incident is attributed to the flow corrosion–erosion.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

The safety and reliability of the reaction effluent air cooler (REAC) system in hydroprocessing equipment are common con-
cerns because of ammonium salt (NH4Cl or NH4HS) corrosion. Such system also suffers from serious accidents because of the cor-
rosion–erosion of multiphase flow. The proportion of processing of inferior crude oils of high sulfur, high acid, and containing
chlorine has increased in recent years, thereby resulting in constant failure incidents in the REAC system and unplanned parking
accidents [1–2]. The mixture of NH4Cl, NH4HS, and FeS from the reaction in the system frequently causes deposition or blockage
in the pipes; this type of failure presents obvious characteristics of locality and risk [3–5]. Water is thus injected to avoid depo-
sition or blockage induced by ammonium sulfide and ammonium chloride salt; however, this approach also causes other serious
failure problems [6–7]. The main reason is that wet salt deposits are extremely corrosive to pipes made of carbon steel. The aque-
ous sulfide salt solution may also be corrosive depending on its concentration and other factors.
Experimental investigations and numerical simulations have been performed in recent years to examine the abovementioned
type of failure in pipes. Ou et al. [8–9] studied the root causes of the tube explosion in a refinery hydrocracking REAC and
established the ionic equilibrium model. They also analyzed the effect of structure on the location of ammonium salt deposition
and the corrosion behavior of wet ammonium salt in REAC tubes. Wang et al. [10] studied a bursting incident that occurred in a
three-way pipe in a natural gas field by means of inspection, experiments, and CFD simulations. Their results indicated that the

⁎ Corresponding author at: The Flow Included Corrosion Institution, School of Mechanical Engineering and Automation, No.928, Second Avenue, Xiasha Higher,
Education Zone, Zhejiang Sci-Tech University, 310018 Hangzhou, Zhejiang Province, China.
E-mail address: ougf@163.com (G. Ou).

http://dx.doi.org/10.1016/j.engfailanal.2016.12.005
1350-6307/© 2016 Elsevier Ltd. All rights reserved.
 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56 47

failure can attributed to the synergistic effect of galvanic corrosion and flow erosion. El-Sayed [11] explored a pipe spool from the
subsea water injection piping network for oil operations and its internal corrosion through visual inspection, chemical analysis,
metallographic examination, SEM/EDX analysis. This author proposed that the mechanism of such failure is the flow-enhanced
corrosion. Jenabali et al. [12] examined the increased corrosion failures in petrochemical plants, investigated the corrosion damage
of air cooler tubes, and proposed that the presence of daily varied corrosive agents (insoluble salt, chlorides, and sulfidic com-
pounds) in the crude oil corrodes and chokes the tubes. Zhu et al. [13] investigated the crystal and deposition behavior of ammo-
nium chloride salt (NH4Cl) and predicted the deposition of ammonium salts by Aspen software and CFD technology.
In the present study, a water injection point situation was modeled to study the corrosion–erosion behavior. The test specimen
(10 mm × 10 mm × 1 mm) in the corroded area was extracted, and the microscopic corrosion morphology of specimen was ob-
served with scanning electron microscope (SEM). Water injection was conventionally modeled with CFD simulation. The mathe-
matical model of multiphase flow was established to analyze the technology correlation process, reveal the flow corrosion–
erosion mechanism, and simulate its inner flow field using the mixture model. The results of this study provide significant refer-
ence for the in-service inspection and risk assessment of inferior oil processing in the piping system of hydrogenation REAC
system.

2. Formation processes of corrosion–erosion

2.1. Relevant technological process

The process flow diagram of a hydrogenation REAC system is shown in Fig. 1. As shown in the figure, the main pipe is located
between the heat exchanger (E-02) and the air coolers. The distances between the pipe and the upstream and downstream equip-
ment are 15.0 and 3.5 m. The crude oil is filtrated and dehydrated and then flows into the buffer tank with the increased pressure
by the high-pressure pump. The raw material and hydrogen (including circulating hydrogen and new hydrogen) are mixed, and
the heat is transferred using the reaction effluent. After pre-heating in the furnace, the mixture enters into the hydrocracking re-
actor directly. The reaction effluent drains from the bottom of the hydrocracking reactor and flows through the heat exchanger
(E-01), the high pressure and high temperature separator, the heat exchanger (E-02), and the air coolers. Thereafter, the effluent
is separated into circulating hydrogen (reused in the loop), oil, and sour water through the high pressure and low temperature
separator and the low pressure and low temperature separator. Two water injection points A and B are installed before and
after the air coolers (E-02). The temperatures in the inlet and outlet are 140 and 52 °C, respectively. The operating pressure is
15.8 MPa.
The injecting water chemistry in Table 1 satisfies the API 932-B standard [14]. The water is continuously injected through the
multiple injection points to dissolve the corrosive medium (NH4Cl and H2S) adequately. In one of the injection points, the
injecting velocity of wash water ranges from 1.0–1.4 m/s; the average velocity is used in the simulation. The existence of oxygen
increases the causticity of polluted water containing sulfur, and chloride leads to pitting and risk of stress corrosion cracking in the
equipment material.

Fig. 1. Process flow diagram of hydrogenation REAC system.

48 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56

Table 1
The injection water chemistry (ppmw).

Medium composition Maximum value Expected value

Oxygen 50 15
pH 9.5 7.0– 9.0
Water hardness 2 b1
Iron content 1 0.1
Chloride 100 5
H2S – b45
NH3 – b100
Cyanide – 0

2.2. Flow corrosion–erosion mechanism in REAC system

The multiple injection water technology is set between the heat exchanger and the air coolers to avoid the crystallization, de-
position, and blockage of ammonium salt. Accordingly, the NH4Cl or NH4HS crystal condensing and separating out from the inlet
media can be removed. Ammonium salt particles easily absorb moisture and become deliquescent in the presence of water, there-
by leading to serious under-deposit corrosion for carbon steel pipes in air cooler. This process is shown as follows:

þ − 2þ −
Fe þ 2NH4 þ 2Cl →Fe þ 2NH3 ðlÞ þ H2 ↑ þ 2Cl ; ð1Þ

þ −
xFe þ yNH4 þ yHS →Fex Sy þ yNH3 þ yH2 ↑: ð2Þ

In reaction Eqs. (1) and (2), ammonium salt solution reacts with carbon steel substrate and form the corrosion product film
FexSy, which prevents further corrosion to carbon steel. The large velocity gradient near the mixing area during this multiphase
flow results in the local damage and loss in the corrosion product film when the wall shear stress increases to a certain degree.
The carbon steel substrate exposes to the corrosion media, continues to be corroded, and finally forms an autocatalytic accelerated
corrosion–erosion system.

2.3. Failure situation analysis

A damage incident in a three-way pipe is shown in Fig. 2(a). The original wall thickness of the main pipe is 25 mm and the
average remaining wall thickness is approximately 10 mm at the failed zone. Given the three-year service life of the injecting
water three-way pipe, the calculated thinning rate is approximately 5 mm/a. This flow corrosion situation is serious. Visual in-
spections were performed on the provided three-way pipe. A hole with a diameter of 8 cm is found in the downstream area
and nearly midline direction at the bottom of the main pipe [Fig. 2(b)]. The obvious wall thickness thinning area with a feature
of hyperbolic shape is observed in the downstream area of 0°–180° at the bottom of the main pipe. However, slight wall thickness

(a) 180° 270°

Φ 4.25d
(b)
180°
90° 0° Φd

Water injection pipe

Flow direction
90°
Failure area

Main pipe 0°

Fig. 2. Macro-photographs of a damaged three-way pipe: (a) overall structure and flow direction; (b) cross section of thinning area with a feature of hyperbolic
shape in the downstream of main pipe.
 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56 49

thinning is found in the top area (180°–360°) of the main pipe and the water injection pipe. The junction between the water in-
jection pipe and the main pipe is surrounded by fillet weld reinforcement, and slight wall thickness thinning is observed.

2.4. Composition and microscopic morphology of specimen

The chemical composition of carbon steel is listed in Table 2. The carbon steel used in the present study satisfies the national
standard of China [15] and is related to the ASTM 1020 material. A hyperbolic thinning area and a damaged hole can be visibly
observed (Fig. 2). The corroded area in the carbon steel pipe was cut into a uniform test specimen with dimensions of 10 mm
× 10 mm × 1 mm. The microscopic corrosion morphology of the test specimen was observed by SEM, and the obtained morphol-
ogies are shown in Fig. 3(a)–(d). Rugged and crack areas are observed because of the presence of chloride ion in the materials.
Specifically, chloride ion loosens the surface structure of the corrosion products. The pitting area is attributed to the effects of
droplets. The result of energy dispersive spectrometer (EDS) analysis of the deposits is shown in Fig. 3(e). The dominant elements
are found to be iron and sulfur. Therefore, the failure is closely related to the corrosive compounds in the stream.

3. Governing equations

3.1. Continuity and momentum equations

The control equations are established as follows:

∂
ρ þ ∇•ρu ¼ 0 ð3Þ
∂t

∂ ∂
 ∂p ∂τij ∂ D 0 0
E
ðρui Þ þ ρui u j ¼ − þ þ −ρui u j þ ρg i þ F i ; ð4Þ
∂t ∂x j ∂xi ∂x j ∂x j

where u is the average velocity; μ is the dynamic viscosity coefficient; ρ is the density of mixture; i and j are the normal and tan-
gent directions, respectively; τ is the laminar shear stress tensor; h−ρui 0 u j 0 i is the Reynolds stress tensor.

3.2. Turbulence equations

The realizable k − ε turbulent model was adopted to solve the internal flow field parameters [16].
Turbulent kinetic energy k transport equation:

"  #
∂ ∂
 ∂ μ t ∂k
ðρkÞ þ ρku j ¼ μþ þ Gk þ Gb −ρε−Y M þ Sk ; ð5Þ
∂t ∂xi ∂x j σ k ∂x j

Turbulent dissipation rate ε transport equation:

"  #
∂
 2
∂ ∂ μ t ∂ε ε ε
ðρεÞ þ ρεu j ¼ μþ þ ρC 1 Sε−ρC 2 pffiffiffiffiffiffi þ C 1ε C 3ε Gb þ Sε ; ð6Þ
∂t ∂x j ∂x j σ ε ∂x j k þ uε k

where turbulent viscosity μt is expressed as the function of turbulent kinetic energy k and turbulence dissipation rate ε; in addi-
tion, μt = ρCμk2/ε, C1 = max {0.43,η/(η + 5)}, and η = Sk/ε. The model constant values C1ε , C2 , σk ,σε are set as 1.44, 1.9, 1.0, and 1.2,
respectively. S, Sk, and Sε are the source items.

Table 2
Chemical composition (wt.%) of carbon steel.

C Si Mn Cr Ni S P Mo

0.18 0.27 0.65 0.10 0.20 0.02 0.03 0.35

50 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56

(a) (b)

Pitting

Crack area

Rugged area

(c) (d)

Crack area

Rugged area

(e)

keV

Fig. 3. Corrosion morphologies of the test specimen at thinning area (a) SEM image of rugged and pitting areas, (b) and (d) SEM image of crack area, (c) SEM
image of rugged area, (e) EDS pattern of the deposits.

3.3. Non-uniformity coefficient

The non-uniformity coefficient φ was employed to quantitatively calculate the extent of distribution, and this coefficient is de-
fined as [17]

σ
φ¼ ; ð7Þ
φ

where σ is the standard deviation of data in the section and can be obtained by calculating all grid nodes as follows:

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u N
u1 X 2
σ ¼t ðφi −φÞ ; ð8Þ
N i¼1

where N is the number of grid nodes and φi is the data of each grid node. φ is the average value in the section and is calculated as
 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56 51

Water injection
50d

α
Y
Flow direction

4.25d
50d
Z X

Fig. 4. 3D structure drawing of three-way pipe used for CFD simulation (α = 90°).

follows:

1X N
φ¼ φ: ð9Þ
N i¼1 i

4. Numerical modeling

4.1. Description of three-way pipe

In investigating the flow field inside the three-way pipe, a 3D structure with an inclined angle α of 90° was constructed (Fig.
4). The injecting water flows along the negative Y direction and fluid moves through the negative X direction inside the main
pipe. The water injection pipe with a diameter of 40 mm (d = 40 mm) and a vertical height of 50d is positioned from the
inlet of the main pipe. The effects of five inclined angles of 30°, 45°, 60°, 75°, and 90° on the inner flow field in the three-way
pipe were analyzed.

4.2. Mesh structure and division

The simulation domain was established on the basis of the structure feature mentioned above. As shown in Fig. 5, the three-
way pipe employs a structured mesh arrangement with hexahedral elements. The grid independency study shows that large

Velocity inlet 1

A Section A-A

Pressure out Velocity inlet 2

Fig. 5. The longitudinal section view of three-way pipe about hexahedral grids (α = 90°).
52 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56

Table 3
Inlet fluid physical parameters of main pipe with standard operation condition.

Phase state

Parameters Vapor phase Oil phase Water phase

Density/(kg·m−3) 26.08 676.30 929.30

Volume fraction/% 98.92 0.82 0.26
Mixed flow velocity/(m/s) 2.08
Viscosity/(kg·m−1·s−1) 1.32 × 10−5 2.62 × 10−4 1.88 × 10−4

Water phase fraction

Fig. 6. Water phase fraction contours in the longitudinal section with merging angles of 90°.

number of grids does not influence the accuracy when the grid quantity is divided into 0.74 million, 0.81 million, 0.93 million, and
1.15 million. Furthermore, the relative flow parameters, such as flow velocity or pressure drop, do not change obviously. The grid
number used in the present study was approximately 0.93 million considering calculation precision and time.

4.3. Boundary condition and simulation procedure

The realizable k − ε turbulent model was adopted to solve the internal flow field parameters. The computational domain is set
as velocity inlet and pressure outlet. The inlet multiphase flow physical parameters of main pipe with standard operation condi-
tion are shown in Table 3, and the mass flow rate of water injection pipe is 1.51 kg/s.

Water phase (a) (b) (c)

fraction

(d) (e) (f)

Fig. 7. Water phase fraction contours in different transversal sections with merging angles of 90° (a) x/d = −0.5, (b) x/d = −1.25, (c) x/d = −2.5, (d) x/d =
−3.75, (e) x/d = −5, (f) x/d = −25.
 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56 53

2.1 2.2

2.0
2.0

1.8

Non-uniformity coefficient ϕ
1.9

-1
1.6

Average velocity / m·s

Flow direction
1.8 1.4

1.7 1.2

Perpendicular to the 1.0

1.6 water injection outlet
0.8

1.5
0.6

1.4 0.4
-50 -40 -30 -20 -10 0 10 20 30 40 50
X/d

Fig. 8. Average velocity and its non-uniformity coefficient in different cross-sections with merging angle of 90°.

The commercial computational fluid dynamics ANSYS 13.0 software was employed in the numerical calculation. The SIMPLEC
algorithm [18] was utilized to solve the coupling of pressure and velocity field. The standard discretization schemes were used for
the pressure terms, and the second-order upwind was used for the convection and divergence terms. The convergent criteria for
all the steady simulations were set at 10−5 in each equation. The standard wall function was combined with the realizable k −ε
model mentioned above.

5. Results and discussion

5.1. Numerical simulation results

The distance between the inlet and outlet of the main pipe was set to be long in the process of CFD simulation to reflect the
actual flow condition and the full development flow state. The water phase fraction contours of the longitudinal and transversal
sections with merging angles of 90° are shown in Figs. 6 and 7. In the longitudinal section, an obvious distribution of water frac-
tion is observed. After mixing with the mainstream media, the injecting water flows to the bottom of the main pipe under the
action of perpendicular feature and gravitational effect. The inlet flow velocity of the mainstream media (2.08 m/s) is larger
than that of the injecting water (1.20 m/s); as a result, water is prone to deviate from its own direction and form an angle be-
tween the actual flow and perpendicular directions. The change regularity of water fraction in the transversal section is shown
through six cross sections in Fig. 7(a)–(f). The water is driven by mainstream media to the downstream, distributes unevenly,
and deposits at the bottom of the main pipe wall. The generated ammonium salt via chemical reaction in the mainstream

0.30 0.14

0.12
0.25
Turbulent kinetic energy / m·s

0.10
Flow direction
Wall Shear Stress / Pa

0.20
0.08

0.15
0.06

High risk area of failure

0.10 0.04
2
-2

0.02
0.05

0.00
0.00
-50 -40 -30 -20 -10 0 10 20 30 40 50
X/d

Fig. 9. Wall stress shear and turbulent kinetic energy at the bottom of main pipe with merging angle of 90°.
54 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56

Water phase fraction

Fig. 10. Water phase fraction contours in the longitudinal section with different merging angles (a) α = 75°, (b) α = 60°, (c) α = 45°, (d) α = 30°.

dissolves in the presence of water and then forms an acidic solution containing high salinity. Consequently, a serious corrosion–
erosion occurs at the bottom of the pipe wall.
The average velocity and its non-uniformity coefficient in various cross sections with merging angle of 90° are shown in Fig. 8.
Considering the converging effect of branch flowing, the flow direction of water changes suddenly from the vertical direction to
the horizontal direction. This phenomenon indicates that a large velocity gradient exists near the mixing area of 5–12d. The mean
velocity of injecting wash water (1.20 m/s) is smaller than that of the mainstream (2.08 m/s); consequently, the mixture flow
velocity decreases significantly in the mixing region. As shown in Fig. 8, the flow velocity decreases first and then recovers to a
steady level. Therefore, the velocity gradients in this region are larger than those in the other flowing regions. Meanwhile, the
velocity gradients near the bottom wall of the pipe are extremely large as a result of the effect of near-wall boundary layer.
The aqueous water is susceptible to deposit in the flowing process; thus, the fluid viscosity and wall shear stress increase. This
increase eventually leads to the high risk of flow corrosion–erosion. Such velocity gradient not only results in an area of low ve-
locity area but also leads to a region of high wall shear stress and high turbulent kinetic energy (Fig. 9); this region coincides with
the corroded area (Fig. 2). This region also becomes very chaotic, and the non-uniformity coefficient of velocity is high. The mix-
ture flows toward the downstream and stabilizes gradually after a distance of approximately 20d. Thereafter, the non-uniformity
coefficient of velocity declines gradually first and then smoothens and stabilizes.
The wall stress shear and turbulent kinetic energy at the bottom of the main pipe with merging angle of 90° is shown in Fig. 9.
As a result of the perpendicular feature and gravitational effect, most of the bottom space in the mixing area is occupied by the
mainstream media and injection water. Accordingly, wall shear stress and turbulent kinetic energy rapidly increase and the re-
moval of corrosion products or protective film is facilitated. Therefore, this change in flow state enhances mass transfer and fur-
ther aggravates the flow corrosion–erosion and eventually leads to rupture of pipe and enlargement of leakage area. Given the
deposition, the corrosive medium, such as NH4Cl or H2S, dissolves in the deposited water. As a result, the fluid causticity at the

0.35

0.30

30°
0.25
Water phase fraction

45°
60°
0.20 75°
90°
0.15

0.10

0.05

0.00

-50 -40 -30 -20 -10 0 10 20 30 40 50

X/d

Fig. 11. Water phase fraction at the bottom of main pipe with different merging angles.
 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56 55

0.8

30°
0.6 45°

Wall Shear Stress / Pa

60°
75°
90°
0.4

0.2

0.0
-50 -40 -30 -20 -10 0 10 20 30 40 50
X/d

Fig. 12. Wall shear stress at the bottom of main pipe with different merging angles.

bottom of the pipe increases. Meanwhile, the wall shear stress also increases owing to the increase in fluid viscosity. Therefore,
the erosion damage at this region is the most serious among all regions. In summary, the flow corrosion–erosion is the main
mechanism of failure. The high risk area ranges from 5–12d at the bottom of the main pipe and coincides with the corroded
area (Fig. 2).

5.2. Effect of merging angles on distribution of water fraction and wall shear stress

Using a three-way pipe changes the flow direction or adds other chemical agents to the contents of pipe. Currently, three-way
pipe is widely used for convergence or separation of two-pipe flowing and is an important part of piping systems [19]. The flow
state is sensitive to the structure of the three-way pipe. Thus, the effect of different merging angles on the distribution of water
phase fraction was discussed in the current work, and the result is presented in Fig. 10. The distribution rule in the longitudinal
section with various merging angles is similar, and water deviates from its own direction and forms an angle between the actual
flow and perpendicular directions. This trend differs depending on the distance of water influencing the bottom of the main pipe
wall. The range of this area is from 5–12d, thereby coinciding with the corroded area (Fig. 2).
The water phase fraction at the bottom of the main pipe with different merging angles is shown in Fig. 11, and the wall shear
stress is shown in Fig. 12. The distribution of water phase fraction and wall shear stress is both steady in the merging angle of 90°
compared with those in other angles. The merging angle of 90° shows satisfactory effect on the three-way pipe and can thus be
applied in practical engineering.

6. Conclusions

The three-way pipe is an important component in the flow industry, and accurate prediction of corrosion–erosion is essential
for controlling the size and cost associated with manufacturing and operation in hydrogenation REAC system. Based on the cor-
relation process of water injection technology, the flow corrosion–erosion mechanism in carbon steel pipe was revealed and the
inner flow field was simulated using the mixture model. The CFD results indicate that a large velocity gradient exists near the
mixing area of 5–12d at the bottom of the main pipe. This gradient results in a region of low flow velocity, high wall shear stress,
and high turbulent kinetic energy. The deposited corrosive medium (NH4Cl or H2S) in water and the increased wall shear stress
cause serious erosion damage. Therefore, the flow corrosion–erosion is the main mechanism of failure. The good agreement be-
tween the high risk area calculated from the CFD simulation and the corroded area of the three-way pipe on the spot verifies
the validity of the CFD simulation approach in engineering application.

Acknowledgments

The project is supported by the Coal Joint Fund from Natural Science Foundation of China and Shenhua Group Corporation
Limited (U1361107), Zhejiang Provincial Natural Science Foundation of China under Grant (LY17E060008 and LY17F030024)
and the Project Grants 521 Talents Cultivation of Zhejiang Sci-Tech University (11130132521505).
56 H. Jin et al. / Engineering Failure Analysis 73 (2017) 46–56

References

[1] A.B. Radwan, A.M. Abdullah, H.J. Roven, et al., Failure analysis of 316L air cooler stainless steel tube in a natural gas production field, Int. J. Electrochem. Sci. 10
(2015) 7606–7621.
[2] H.Z. Jin, G.F. Ou, Y.P. Wang, et al., Failure Analysis and Structure Optimization of Hydrogenation Air-cooler System Based on Imbalanced Degree, Paper 576, PVP,
ASME, 2010.
[3] K. Toba, T. Suzuki, K. Kawano, et al., Effect of relative humidity on ammonium chloride corrosion in refineries, Corrosion 67 (2011) 055005-1–055005-7.
[4] A. Popova, M. Christov, A. Vasilev, Mono- and dicationic benzothiazolic quaternary ammonium bromides as mild steel corrosion inhibitors. Part III: influence of
the temperature on the inhibition process, Corros. Sci. 94 (2015) 70–78.
[5] F.L. Fei, J. Hu, J.X. Wei, et al., Corrosion performance of steel reinforcement in simulated concrete pore solutions in the presence of imidazoline quaternary am-
monium salt corrosion inhibitor, Constr. Build. Mater. 70 (2014) 43–53.
[6] R.J. Horvath, M.S. Cayard, R.D. Kane, Prediction and Assessment of Ammonium Bisulfide Corrosion Under Refinery Sour Water Service Conditions, Paper 576, Cor-
rosion, 06, NACE, 2006.
[7] L. Sun, M. Zhu, G.F. Ou, et al., Corrosion investigation of the inlet section of REAC pipes in the refinery, Eng. Fail. Anal. 66 (2016) 468–478.
[8] G.F. Ou, K.X. Wang, J.L. Zhan, et al., Failure analysis of a reactor effluent air cooler, Eng. Fail. Anal. 31 (2013) 387–393.
[9] G.F. Ou, H.Z. Jin, H.P. Xie, et al., Prediction of ammonium salt deposition in hydroprocessing air cooler tubes, Eng. Fail. Anal. 18 (2011) 1458–1464.
[10] Z.Y. Wang, B. Liu, Y.G. Yang, et al., Experimental and numerical studies on corrosion failure of a three-way pipe in natural gas field, Eng. Fail. Anal. 62 (2016)
21–38.
[11] M.H. El-Sayed, Flow enhanced corrosion of water injection pipelines, Eng. Fail. Anal. 50 (2015) 1–6.
[12] S.A. Jenabali Jahromi, A. Janghorban, Assessment of corrosion in low carbon steel tubes of Shiraz refinery air coolers, Eng. Fail. Anal. 12 (2005) 569–577.
[13] M. Zhu, G.F. Ou, H.Z. Jin, et al., Top of the REAC tube corrosion induced by under deposit corrosion of ammonium chloride and erosion corrosion, Eng. Fail. Anal. 57
(2015) 483–489.
[14] API Publication 932-B (Second Edition), Design, materials, fabrication, operation, and inspection guideline for corrosion control in hydroprocessing reactor efflu-
ent air cooler (REAC) system, 2012.
[15] National Standard of China, Seamless Steel Tubes for Liquid Service, GB/T8163-2008, 2008.
[16] Fluent 13.0 User's Guide, Fluent Inc., Centerra Resource Park, New Hampshire, USA, December 2010.
[17] G.F. Ou, Z.J. Zheng, H.Z. Jin, Numerical analysis on the multiphase flow characteristics of static blade mixer, J. Chem. Eng. Chin. Univ. 30 (2016) 40–47.
[18] X.W. Cui, H.X. Lai, L. Zhao, A comparison of temporal discretization schemes for the SIMPLE method, J. Eng. Thermophys. 6 (2014) 476–479.
[19] H.J. Zhu, W.L. Zhang, G. Feng, et al., Fluid–structure interaction computational analysis of flow field, shear stress distribution and deformation of three-limb pipe,
Eng. Fail. Anal. 42 (2014) 252–262.