Experimental and numerical studies on the corrosion properties of AISI 316L stainless steel in two-phase upward slug flows

Abstract

The corrosion properties of AISI 316L stainless steel in upward slug flow were initially investigated using the electrochemical measurements and computational fluid dynamics (CFD) simulations. The results demonstrate that the slug flow may destroy the passivation film on the specimen surface due to its intermittency and volatility, accelerating the pitting corrosion process on the electrode surface. It results in a higher corrosion rate in the two-phase slug flow than in the single-phase flow. The variations of superficial velocity lead to changes in the slug flow configuration. The variation has a greater impact on the Taylor bubble than the liquid slug. The changes in slug flow configuration result in differences in shear stress and mass transfer coefficient, leading to corresponding differences in electrochemical properties of AISI 316L stainless steel.

KEYWORDS:

• AISI 316L stainless steel
• slug flow accelerated corrosion
• CFD simulation
• electrochemical measurement
• corrosion morphology analysis

1. Introduction

AISI 316L stainless steel (316L SS) is extensively applied in various pipelines because of its high corrosion resistance (Nosei et al., 2008). In an oxidizing environment, it will react with oxygen and form a passivation film on the surface to prevent further corrosion (Turnbull et al., 2003). However, in the petroleum industry, the fluid medium usually contains chloride ions, which can destroy the product scale film (Khan et al., 2002; Lu et al., 2014). Moreover, the medium is usually a two-phase gas–liquid flow, which results in much more complicated flow characteristics than do single-phase flows. Two-phase flows can exhibit a variety of flow patterns including annular, laminar, and bubbly flows in different working conditions. Among these patterns, slug flow in vertical tubes is usually in the spotlight due to its greater prevalence in the fields of petroleum exploitation, transportation, and processing (Lu & Prosperetti, 2009; Zheng et al., 2015).

The most notable feature of upward slug flow is its intermittency (Mercier-Bonin et al., 2000), caused by a series of slug units. The slug unit includes a bullet-shaped Taylor bubble and a continuous liquid slug (Araújo et al., 2012). Taylor bubbles move faster than the liquid in vertical tubes with buoyancy effects (Jean & Bernardo, 2014). In this process, a thin film of liquid is formed around Taylor bubbles (Tahir et al., 2017). Flow field characteristics in slug flow are complicated due to the flow configuration, dramatically accelerating the corrosion process. In recent years, due to the frequent occurrence of failures caused by slug flow accelerated corrosion, slug flow’s characteristic effects on metal corrosion have become key issues that urgently need to be resolved. Thus far, there have been many studies on slug flow hydrodynamic properties. For example, Orell and Rembrand (1986) and Andreussi et al. (1993) studied vertical slug flow characteristics and proposed analytical models of Taylor bubble. Taha and Cui (2006), Hayashi et al. (2011) and Lizarraga-Garcia et al. (2016) analyzed the rising velocities of different parts of the slug flow using dimensional analysis and found their dependence on the internal diameter, density, viscosity, and gas–liquid tension. In addition, some research on corrosion characteristics of carbon steel has been conducted. Villarreal et al. (2006) experimentally studied the corrosion rate of carbon steel in gas (air)–liquid (NaCl solution with dissolved CO₂) slug flows and revealed the relationship between corrosion rate and slug frequency. Turbulence in liquid slugs was considered as a key factor affecting the metal corrosion rate. Zheng et al. (2008) and Li et al. (2016) investigated corrosion properties of carbon steel in slug flow using electrochemical techniques. The corrosion rate was suggested to be related to the mass transfer coefficient and shear stress of fluid. Besides common parameters, some other factors like thiosulphate-reducing bacteria (Machuca et al., 2017) affecting carbon steel corrosion were also investigated. However, the aperiodic characteristics of slug flow often lead to great difficulties in determining the hydromechanical effects on corrosion rate from a purely experimental approach (Nogueira et al., 2003). Numerical investigations can just about make up for the shortcomings of experiments and have been an effective complement to experiments (Akbarian et al., 2018; Ban & Pao, 2018). Owing to the advantages of convenience, celerity, and economy, the numerical method has been used for many engineering simulations, for example, the estimation and optimization of flutter speed of a typical wing in the diesel engine and the simulation of a snubber in a CNG station (Farzaneh-Gord et al., 2019; Ghalandari et al., 2019).

In spite of the broad application of 316L SS in the petrochemical industry, few definitive conclusions about the corrosion characteristics of 316L SS have been obtained. The present research examines the corrosion characteristics of 316L SS in gas–liquid upward slug flows using experiments and numerical simulations. The hydrodynamics of slug flow under different conditions are investigated with a circulating loop system and FLUENT software. Corrosion properties of 316L SS in different slug flows are investigated using electrochemical impedance spectrum, potentiodynamic polarization curve, scanning electron microscope (SEM), and linear polarization resistance techniques.

2. Experimental

2.1. Experimental apparatus

A two-phase gas–liquid flow system, as shown in Figure 1, was used to investigate the flow and corrosion processes of slug flow. It consisted of a reservoir (1), a nitrogen cylinder (2), controlling valves (3 and 4), a centrifugal pump (5), flowmeters (6 and 7), an electrochemical test cell (8, 9, and 10), and pipes (11). The liquid and gas media were provided from the reservoir and the nitrogen cylinder, respectively. By adjusting the pump rotational speed, the liquid solution flow rate was controlled. The gas medium flow rate was controlled by the valve at the entrance. By controlling the flow rates of gas and liquid, slug flows of different flow characteristics were presented in the upstream vertical pipeline and observed with a high-speed camera. Plexiglass tubes with inner diameter of 20.0 mm were used in the flow system. After pre-treatment, a three-electrode cell was placed in the test section with surfaces exposed according to the internal surface of the pipelines.

Figure 1. Schematic diagram of the flow system used in slug flow accelerated corrosion test: (1) reservoir, (2) nitrogen cylinder, (3 and 4) controlling valves, (5) pump (6 and 7), flowmeters, (8) three-electrode system, (9) Princeton Versa Stat Mc, (10) high-speed camera, (11) pipelines.

2.2. Electrochemical measurements

To study the corrosion characteristics of 316L SS in upward slug flow, a three-electrode electrochemical cell was used to conduct electrochemical tests. A cylindrical working electrode (WE) with an exposed area of 3.14 10⁻² cm² was made up of 316L SS. Saturated calomel reference electrode (RE) and carbon rod counter electrode (CE) were employed for electrochemical measurements. Table 1 shows the chemical composition of 316L SS. Before the test, working surfaces of WE and CE were ground with 800# silicon carbide paper and washed with a mixture of alcohol and deionized water. Each working electrode was employed for one operational condition.

Table 1. Chemical composition of 316L SS (wt %).

C	Mn	P	S	Si	Cr	Ni	Mo	Fe
0.03	1.99	0.035	0.029	0.98	16	10.3	2.03	Balance

Tests were conducted in 3.0 wt% NaCl aqueous solution prepared using analytical grade reagents and deionized water. Solution pH was between 7.0 and 7.5. Gas medium for test was nitrogen of 99.95% purity. The experiments were performed at room temperature (23.4 ± 0.1°C) and atmospheric pressure (1.01 × 10⁵ Pa). Before experiments, nitrogen was passed through the tubes to remove air from the system for half an hour.

Electrochemical impedance spectroscopy (EIS), potentiodynamic polarization curve, and linear polarization resistance (LPR) measurements were made to determine the corrosion behaviors of 316L SS at different operating conditions. Polarization curves were obtained at a constant sweep rate of 1 mV/s in the range of −1.0 to 1.6 V with respect to open circuit potential (OCP). LPR measurements were conducted by polarizing WE from -10 mV_SCE to +10 mV_SCE  with respect to E_corr at the scanning rate of 1 mV/s every 20 min. EIS measurements were conducted at the E_corr value with a scanning frequency from 10⁵ to 10⁻¹ Hz.

The corrosion current density $I_0$ of 316L SS was determined with the Stern–Geary formula, as shown in Equation (1): (1) $I_0=\frac{b_a{b}_c}{2.303(b_a+b_c)}\frac{1}{R_P}$(1) The average reduction of specimen thickness which is given by Equation (2) was applied to indicate the 316L SS corrosion rate: (2) $v_d=\frac{8.76v_0}{\rho }$(2) In Equation (2), $\rho$ denotes the density of 316L SS; $v_0$ is the weight reduction caused by slug flow accelerated corrosion and can be obtained from Equation (3): (3) $v_0=\frac{I_0{m}_a}{nF}$(3)

3. Numerical simulation

3.1. Governing equations

The commercial software FLUENT was employed to simulate the upward slug flow in vertical tubes. Numerical simulations were carried out using the method of volume of fluid (VOF) introduced by Hirt and Nichols (1981). In this approach, continuity and momentum equations are described as follows: (4) $\begin{array}{rl}& \frac{\mathrm{\partial }\rho }{\mathrm{\partial }t}+\mathrm{\nabla }\cdot (\rho \mathbf{\text{v}})=0\end{array}$(4) (5) $\begin{array}{rl}& \frac{\mathrm{\partial }}{\mathrm{\partial }t}(\rho \mathbf{\text{v}})+\mathrm{\nabla }\cdot (\rho \mathbf{\text{vv}})\\ & =-\mathrm{\nabla }P+\mathrm{\nabla }\cdot {\mu }_{eff}(\mathrm{\nabla }\mathbf{\text{v}}+\mathrm{\nabla }{\mathbf{\text{v}}}^T)+\rho g\end{array}$(5) In order to identify the interface position in the upward slug flow, a function for the volume fraction is employed, described by Drew (1983): (6) $\frac{\mathrm{\partial }({\alpha }_q{\rho }_q)}{\mathrm{\partial }t}+\mathrm{\nabla }\cdot ({\alpha }_q{\rho }_q\mathbf{\text{v}})=0$(6) In the fully developed slug flow, strong turbulence is formed in the liquid slug region. To take both accuracy and processing time into consideration, a standard k–ε turbulence model was applied to calculate the two-phase upward slug flows in vertical tubes. For this model, the turbulent viscosity ${\mu }^T$ is calculated with the following equation: (7) ${\mu }^T=C_{\mu }\frac{k^2}{\epsilon }$(7) The turbulent kinetic energy k and its dissipation rate ε are given as follows: (8) $\begin{array}{rl}& \frac{\mathrm{\partial }\rho k}{\mathrm{\partial }t}+\mathrm{\nabla }\cdot \left[\rho \mathbf{\text{v}}k-\left(\mu +\frac{{\mu }^T}{{\sigma }_k}\right)\mathrm{\nabla }k\right]=p^k-\rho \epsilon \end{array}$(8) (9) $\begin{array}{rl}& \frac{\mathrm{\partial }\rho \epsilon }{\mathrm{\partial }t}+\mathrm{\nabla }\cdot \left[\rho \mathbf{\text{v}}\epsilon -\left(\mu +\frac{{\mu }^T}{{\sigma }_{\epsilon }}\right)\mathrm{\nabla }\epsilon \right]\\ & =C_1\rho S\epsilon -C_2\rho \frac{{\epsilon }^2}{k\sqrt{\mathbf{\text{v}}\epsilon }}\end{array}$(9) The constants used for the solution were $C_{\mu }$ = 0.09, ${\sigma }_{\epsilon }$ = 1.3, ${\sigma }_k$ = 1.0, $C_1$ = 1.44, and $C_2$ = 1.92. $p^k$ stands for turbulence production, defined by: (10) $p^k={\mu }^T(\mathrm{\nabla }\mathbf{\text{v}}+{(\mathrm{\nabla }\mathbf{\text{v}})}^T):\mathrm{\nabla }\mathbf{\text{v}}$(10) In the upward slug flows, wall shear stress is described by (11) $\tau \text{ = }\mu \left(\frac{\mathrm{\partial }{u}_i}{\mathrm{\partial }{x}_j}+\frac{\mathrm{\partial }{u}_j}{\mathrm{\partial }{x}_i}\right)$(11) The mass transfer coefficient of corrosive particles (H⁺, Fe²⁺, Cl⁻, etc.) is described by (12) $\begin{array}{rl}K& =0.023CR{e}^{0.8}S{c}^{0.33}\left(\frac{D_{\text{d}}}{L}\right)\\ & =0.023\times 0.27{\left(1+\frac{Frs}{\beta }\right)}^{0.8}{\left(\frac{\mu }{\rho D_{\text{d}}}\right)}^{0.33}\left(\frac{D_{\text{d}}}{L}\right)\end{array}$(12)

3.3. Grid settings and boundary condition

A two-dimensional CFD model was built in accordance with the experimental pipe to simulate the upward slug flow. The model was composed of three parts: a horizontal straight section (length 500 mm) for feeding gas and liquid, an upstream section (length 2000 mm) for observing flow configurations of slug flow, and a 90° elbow with a curvature radius of 30 mm to connect the two straight sections. The inner diameter of the pipe model was 20 mm, the same as for the experimental tubes.

The tube model was developed and meshed by pre-processing software ANSYS-ICEM, which provides a module to build simple models. Given the simplicity and symmetry of this tube model, quadrilateral structured grids were established. A grid independence study was conducted by verifying 38200, 51240, and 64050 elements. It was found that there was little difference in the flow configurations obtained by the three grid numbers. 38200 elements could meet the requirements of grid independence and calculation accuracy. Table 2 shows the characteristics of nitrogen and water employed in simulations.

Table 2. Properties of nitrogen and water in the simulation.

Properties	Nitrogen	Water
Density (kg/m^3)	1.25	998.2
Viscosity (kg/(m·s))	1.789 × 10^−5	1.003 × 10^−3
Surface tension (N/m)	0.072	–

Mass flow rates and pressure outlet boundary conditions were imposed. The inner wall of the tube was no-slip. The finite volume method was used because of its unique advantages in discretizing governing equations. The pressure and velocity fields of the upward slug flow were coupled with the PISO algorithm. A second-order upwind scheme was applied for momentum. Simulations were conducted with a time step of 10⁻⁵ and the maximal iteration per time step was set as 150.

4. Results and discussion

4.1. Flow characteristics of upward slug flows

4.1.1. Flow configurations at different superficial velocities

Flow rate is the crucial parameter governing slug flow characteristics (Gueta et al., 2006). For the convenience of analysis, the slug unit is divided into several regions, the Taylor bubble, which consists of nose and main body, wake, liquid film around Taylor bubble, and liquid slug (Thaker & Banerjee, 2015), shown in Figure 2. ‘Taylor bubble nose length’ refers to the distance between the bubble tip and the position where the bubble diameter does not change visibly. ‘Diameter of the bubble nose’ refers to the diameter of the bubble corresponding to the central position. ‘Length’ and ‘diameter of Taylor bubble’ refer to the dimensions of the Taylor bubble’s main body.

Figure 2. Sketch diagram of the upward slug unit.

Figure 3 shows experimental results obtained for slug flows at superficial gas velocities (U_G) of 0, 0.0265, 0.053, 0.0796, 0.106, 0.159, 0.212, and 0.265 m/s while superficial liquid velocity (U_L) was kept at 0.637 m/s. It can be found that U_G has a significant influence on slug flow configurations, especially for the Taylor bubble and the liquid slug. Taylor bubble and liquid slug lengths are enhanced by increasing U_G. Dimensions of different regions of the slug flow at different U_G are also measured and shown in Table 3.

Figure 3. Experimental results of slug flow at different U_G while U_L was 0.637 m/s. U_G (m/s): (a) 0; (b) 0.0265; (c) 0.053; (d) 0.0796; (e) 0.106; (f) 0.1325; (g) 0.159; (h) 0.1855; (k) 0.212; (l) 0.265.

Table 3. Flow characteristics of slug flow at U_L = 0.637 m/s.

Characteristics UG (m/s)	0.053	0.106	0.159	0.212	0.265
Taylor bubble nose length (cm)	0.8	0.8	1.0	1.3	1.0
Taylor bubble nose diameter (cm)	1.2	1.2	1.3	1.5	1.3
Taylor bubble length (cm)	2.3	5.8	9.0	12.0	12.5
Taylor bubble diameter (cm)	1.8	1.7	1.8	1.8	1.7
Liquid film thickness (cm)	0.1	0.15	0.1	0.1	0.15
Wake region length (cm)	1.0	2.0	2.0	2.0	3.0
Liquid slug length (cm)	12.2	14.0	15.5	16.0	16.0
Slug unit length (cm)	15.5	21.8	26.5	30.0	31.5

It is observed in Table 3 that Taylor bubble nose length ranges from 0.8 to 1.3 cm with a 1.2–1.5 cm diameter as U_G is increased. The length of Taylor bubble is increased from 2.3 (U_G = 0.053 m/s) to 12.5 cm (U_G = 0.265 m/s) with a 1.7–1.8 cm diameter. In proportion, the thickness of liquid film is 0.15–0.1 cm. In this case, the length of bubble wake is slightly increased from 1.0 to 3.0 cm, and the length of liquid slug is increased from 12.2 to 16.0 cm. Accordingly, the total length of slug unit is increased from 15.5 to 31.5 cm with U_G.

Table 4 shows experimental results of slug flow characteristics at U_L = 0.734 m/s. It is observed that slug flow configurations at this operating condition share a similar variation tendency with those shown in Table 3. The nose length is increased from 1.0 to 1.5 cm with a diameter of about 1.0 cm. The length of Taylor bubble is increased from 2.0 (U_G= 0.053 m/s) to 10.0 cm (U_G = 0.265 m/s) with a 1.2–1.7 cm diameter and the corresponding liquid film thickness is 0.4–0.15 cm. The wake length is increased from 0.5 cm to 3.5 cm. In addition, the length of liquid slug is increased from 5.0 to 19.0 cm. As a result, the total length of the slug unit is increased from 7.0 to 32.5 cm as U_G is increased from 0.053 to 0.265 m/s.

Table 4. Flow characteristics of slug flow at U_L = 0.743 m/s.

Characteristics UG (m/s)	0.053	0.106	0.159	0.212	0.265
Taylor bubble nose length (cm)	1.0	1.0	1.2	1.5	1.0
Taylor bubble nose diameter (cm)	1.0	1.0	1.0	1.0	1.0
Taylor bubble length (cm)	2.0	3.0	7.0	8.0	10.0
Taylor bubble diameter (cm)	1.2	1.5	1.6	1.7	1.7
Liquid film thickness (cm)	0.4	0.25	0.2	0.15	0.15
Wake region length (cm)	0.5	0.5	2.0	3.0	3.5
Liquid slug length (cm)	5.0	8.0	12.0	15.0	19.0
Slug unit length (cm)	7.0	11.5	21.0	26.0	32.5

Table 5 shows experimental results of slug flow characteristics at U_L = 0.849 m/s. It shows a significant difference from the aforementioned results. At U_G = 0.053 m/s, many tiny bubbles but no large-length-scale bubbles can be found in the vertical pipe, indicating that the flow patternat this gas flow rate is bubbly flow (Abdulmouti & Abdulmouti, 2014; Morgado et al., 2016). As U_G is generally increased, the dispersed tiny bubbles merge into elongated bubbles, namely Taylor bubbles, forming the slug flow (Brauner & Ullmann, 2004). With increased U_G, the nose length is increased from 0.5 to 1.0 cm with a 1.0–1.4 cm diameter. At the same time, the length of Taylor bubble is increased from 1.5 to 7.5 cm with a diameter of about 1.6 cm. The length of liquid slug is increased from 5.0 to 18.0 cm, resulting in an increase of slug unit length from 8.0 to 30.0 cm.

Table 5. Flow characteristics of slug flow at U_L = 0.849 m/s.

Characteristics UG (m/s)	0.053	0.106	0.159	0.212	0.265
Taylor bubble nose length (cm)	Bubbly flow	0.5	1.0	1.0	1.0
Taylor bubble nose diameter (cm)		1.0	1.2	1.4	1.3
Taylor bubble length (cm)		1.5	4.0	5.0	7.5
Taylor bubble diameter (cm)		1.6	1.7	1.6	1.7
Liquid film diameter (cm)		0.2	0.15	0.2	0.15
Wake region length (cm)		0.5	1.0	3.0	3.0
Liquid slug length (cm)		5.0	8.0	8.5	18.0
Slug unit length (cm)		8.0	15.0	17.0	30.0

Figure 4 shows the relationship between slug flow characteristics and U_L. It can be seen that the lengths of liquid slug, Taylor bubble, and bubble wake are decreased with U_L. The length of Taylor bubble is decreased from 5.8 (U_L = 0.637 m/s) to 1.5 cm (U_L = 0.849 m/s) with a diameter of about 1.6 cm. At the same time, the length of liquid slug is decreased from 14.0 to 5.0 cm. However, it should be noted that the bubble nose and liquid film are not sensitive to liquid flow rate variation at constant U_G.

Figure 4. Flow characteristics of slug flow at different U_L wile U_G was 0.106 m/s.

In summary, conclusions can be drawn that with an increase of U_G both liquid slug and Taylor bubble in vertical tubes get enlarged, and the Taylor bubble gets a greater amplification with U_G. When U_L = 0.637 m/s, the length of Taylor bubble is increased by 5.4 times, while the liquid slug is increased by about 1.3 times. When U_L = 0.734 m/s, the length of Taylor bubble is increased by five times, and the liquid slug is increased by about 3.8 times. Similarly, when U_L = 0.849 m/s, the length of Taylor bubble is increased by five times, and the liquid slug is increased by about 3.6 times. The increase of Taylor bubble length decreased the thickness of liquid film. Conversely, as U_L increases, both Taylor bubble and liquid slug become shorter, and liquid film thickness increases slightly.

4.1.2. Distributions of mass transfer coefficient and shear stress

Slug flow accelerated corrosion is the coupling effect of hydrodynamic action and electrochemical reaction. The process of electrochemical reaction includes four steps (Randazzo et al., 2011): (1) mass transfer of the reactants, (2) electric double layer charging, (3) electrode redox reaction, and (4) stripping of the reaction product. For most operating conditions, the corrosion process of 316L SS is controlled by the mass transfer and electrochemical processes of the reactants. The former is mainly determined by mass transfer coefficient and fluid shear stress (Leishear et al., 2010); the latter is mainly determined by the nature of the electrode and operating conditions such as temperature (Carlson et al., 2018).

Figure 5 shows the simulation results of slug flow in the tube. The results were obtained at t = 52 s while the flow field had reached a stable state. In the figure, U_G from left to right is 0.053, 0.106, 0.159, 0.212, and 0.265 m/s, respectively, while U_L is 0.637 m/s. Simulation results show that the average value of Taylor bubble length at each condition is 2.4, 5.6, 8.5, 10.8, and 11.7 cm, respectively. Experimental results in Table 3 show that the values at the same conditions are 2.3, 5.8, 9.0, 12.0, and 12.5 cm. Liquid slug length in simulations is 4.5, 13.2, 14.4, 15.8, 16.7 cm, while it is 12.2, 14.0, 15.5, 16.0, 16.0 cm in experiments. It can be observed that Taylor bubble and liquid slug are enlarged with U_G, which agrees well with the experimental results. Moreover, other dimension values of bubble nose, wake region, and liquid film in simulations are almost same as from the experiments, demonstrating the validity of CFD simulations.

Figure 5. CFD results of flow configurations of the slug flow at U_L = 0.637 m/s. U_G (m/s): (a) 0.053; (b) 0.106; (c) 0.159; (d) 0.212; (e) 0.265.

Figure 6 presents the distributions of shear stress (τ) and mass transfer coefficient (K) in vertical slug flows, which were also obtained at t = 52 s. It is indicated that the shear stress in the liquid film is higher than that in other regions. Liquid film thickness around the Taylor bubble is not uniform, resulting in different shear stress in this region. Liquid films with lower thickness generate a greater shear stress. This can be explained by the fact that the thinner liquid film leads to a greater velocity gradient, which results in a larger shear stress. As far as the mass transfer coefficient is concerned, it is greater in the liquid film than in other regions. Furthermore, the thinner the liquid film, the greater the mass transfer coefficient.

Figure 6. CFD results of distributions of shear stress and mass transfer coefficient in the slug unit.

Figures 7 and 8 show maximums of shear stress and mass transfer coefficient in slug flows at various superficial velocities. Here, every point in the plot represent the maximum of shear stress coefficient or mass transfer under specific conditions. For example, when U_G is 0.053 m/s and U_L is 0.849 m/s, the shear stress in the entire computational domain ranges from 0 to 34 Pa. Therefore, the maximum shear stress under this condition is 34 Pa. It is suggested that the mass transfer coefficient and shear stress share a similar tendency with U_G and U_L. As U_G is increased from 0 to 0.265 m/s, the maximum of shear stress is increased from 16 to 30 Pa at U_L = 0.637 m/s, from 20 to 37 Pa at U_L = 0.743 m/s and from 30 to 40 Pa at U_L = 0.849 m/s. At the same time, the maximum mass transfer coefficient is increased from 3.1 × 10⁻⁵ to 4.3 × 10⁻⁵ m/s at U_L = 0.637 m/s, from 3.6 × 10⁻⁵ to 4.6 × 10⁻⁵ m/s at U_L = 0.743 m/s, and from 4.4 × 10⁻⁵ to 5.4 × 10⁻⁵ m/s at U_L = 0.849 m/s. It can be found that an increase of U_L increases both shear stress and mass transfer coefficient in slug flow.

Figure 7. Maximums of shear stress in slug flow under various conditions.

Figure 8. Maximums of mass transfer coefficient in slug flow under different conditions.

In the electrochemical reaction process, greater mass transfer coefficient means faster transfer of chloride ions from bulk solution to metal surface and of metal ions from surface to bulk solution (Wang et al., 2016). Moreover, the greater shear stress leads to worse damage to the passivation film. When U_L is increased, the slug flow turbulence is increased, which enhances the shear stress and mass transfer coefficient. Consequently, the accelerated corrosion rate of slug flow in the 316L SS specimen increases with U_G and U_L.

4.2. Electrochemical characteristics of 316L SS

4.2.1. Polarization curves

As mentioned above, slug flow accelerated corrosion is the coupling effect of hydrodynamic flow and electrochemical reaction. The electrochemical corrosion process is drastically impacted by the hydrodynamic characteristics (Wang et al., 2016). Figures 9–11 show the polarization curves of 316L SS in the upward slug flow at different U_G and U_L. A limiting current was observed in the cathodic branches of curves which is directly related to oxygen depletion near the surface. Right below the corrosion potential, oxygen molecule diffusion transport from bulk solution to metal surface controls the cathodic reaction process. When the potential is  more negative to the corrosion potential, the collective effect of diffusion and polarization dominates the cathodic reaction. The anodic branches of polarization curves show the same tendency at different superficial velocities. The current density increases with an increase of anodic polarization potential, indicating that iron atoms are unlimitedly supplied at the bulk solution and 316L SS specimen interface. With further increase of polarization potential, the growth of the corrosion current gradually slows down, meaning that a passivation film is forming on the specimen surface.

Figure 9. Polarization curves of 316L SS in upward slug flow at U_L = 0.637 m/s.

Figure 10. Polarization curves of 316L SS in upward slug flow at U_L = 0.743 m/s.

Figure 11. Polarization curves of 316L SS in upward slug flow at U_L = 0.849 m/s.

4.2.2. EIS measurements

EIS measurements have been performed to evaluate the electrochemical corrosion process of 316L SS in upward slug flow. Figure 12(a–c) shows the Nyquist plots obtained at different U_G and U_L. It can be observed from the figures that all measured plots show a common feature, i.e. semicircle arcs with centers depressed below the x-axis and only one time constant can be found from the plots, because of the coupled response of metal surface and passive film (Marcelin et al., 2013). An electrochemical equivalent circuit, as presented in Figure 13, is employed to obtain the EIS parameters, including the passive film resistance R_f, the constant phase element CPE_f, and the ohmic resistance of the solution R_s (Zhao et al., 2015).

Figure 12. Nyquist plots of 316L SS in slug flow under different conditions: (a) U_L = 0.637 m/s; (b) U_L = 0.743 m/s; (c) U_L = 0.849 m/s.

Figure 13. Equivalent circuit of the corrosion system.

In the slug flow accelerated corrosion process of 316L SS, the cathodic reaction is the oxygen-absorbing corrosion process, and the anode reaction is the metal dissolution process. Active surface area due to defects R_f, solution ohmic resistance R_s, and passive film resistance are key factors affecting the corrosion rate.

Figures 14–15 show the variations of R_f and R_s under different conditions. It is found that the variation trend of R_f with U_G is completely the opposite to that of R_s. As U_G is increased, R_f is drastically decreased, while R_S is increased. When U_G ranges from 0 to 0.265 m/s, R_f is decreased from 179.7 to 45.9 Ω at U_L= 0.637 m/s, from 148.6 to 32.1 Ω at U_L= 0.743 m/s, and from 110.8 to 30.8 Ω at U_L= 0.849 m/s, respectively. Nevertheless, R_s is increased from 1.9 to 5.1 Ω at U_L = 0.637 m/s, from 1.6 to 4.3 Ω at U_L = 0.743 m/s, and from 1.5 to 3.7 Ω at U_L = 0.849 m/s with the increased U_G. It is obvious that R_f has a much wider range of variation with U_G than R_s, indicating a greater influence on the corrosion process. It also can be concluded that both R_f and R_s are decreased with an increase of U_L. This is attributed to the fact that the liquid solution has a better conductivity than nitrogen. As U_L is increased, the liquid content in the slug unit gets higher, leading to smaller R_s. Moreover, the increased U_L leads to greater damage to the passivation film on the surface, which results in a reduction of R_f.

Figure 14. Variations of R_f under different conditions.

Figure 15. Variations of R_s under different conditions.

4.2.3. Corrosion rate of 316L SS

When U_G and U_L are changed, slug flow configurations change correspondingly, resulting in differences in the corrosion rate (Nemoto & Baliño, 2012). The slug flow accelerated corrosion rates of 316L SS samples are measured and counted according to Equation (1); the results are shown in Figure 16.

Figure 16. Corrosion rates of 316L SS under different conditions.

From the figure, it can be found that as U_G is enhanced from 0 to 0.265 m/s the corrosion rate of 316L SS is increased from 0.132 to 0.929 mm/a at U_L = 0.637 m/s, from 0.895 to 1.53 mm/a at U_L = 0.743 m/s, and from 1.1 to 1.98 mm/a at U_L = 0.849 m/s. Obviously, the 316L SS corrosion rate increases with an increase of U_G. Distribution regularities of the mass transfer coefficient and shear stress obtained from numerical simulations can explain this phenomenon well. As shown in Figure 8, with an increase of U_G, the Taylor bubble gets enlarged and the liquid film thickness decreases, resulting in an increase of reactant mass transfer coefficients and fluid shear stress in this region, which greatly promotes the redox reaction on the electrode surface. At the same time, the proportion of Taylor bubble in the slug unit increases, resulting in a larger proportion of the liquid film region, which also increases the corrosion rate. It is suggested that the 316L SS corrosion rate is also increased with the increased U_L. This is probably because reactant ions exist in the liquid medium. With an increase of U_L, the fluid turbulence is increased. It leads to an acceleration of the convective transfer process of the reactants, which promotes the corrosion process of 316L SS.

Moreover, it also can be found that U_G’s effect on the 316L SS corrosion rate is significantly greater than U_L’s. It may be due to the fact that the corrosion process of the 316L SS electrode is mainly related to the mass transfer of reactant ions and electric charges in the near-wall region that is located within the liquid film. As U_G is increased, the thickness of the liquid film decreases, resulting in enhanced mass transfer coefficient and fluid shear in the near-wall region, which can significantly accelerate the corrosion process on the electrode surface. When U_L is increased, although the liquid slug and Taylor bubble are decreased and the liquid film thickness increases, the fluid turbulence gets larger. This also leads to an increase in shear stress and mass transfer coefficient in the near-wall region. The results of the corrosion rate are consistent with polarization curves and EIS measurement results. Furthermore, from the distributions of mass transfer coefficient and shear stress shown in Figures 7 and 8, U_G’s effect on the mass transfer coefficient and shear stress is greater than U_L’s. Therefore, U_G has a relatively greater impact on the 316L SS corrosion rate than does U_L. The findings of this section are consistent with Yan and Che’s (2011) research.

4.3. Corrosion morphology analysis

After the electrochemical measurement for 2 h, the samples were washed with deionized water and dried with cool air. SEM was used to observe the surface morphologies of 316L SS samples. Figure 17 shows 316L SS specimen surface morphologies in the upward slug flow at different U_G while U_L = 0.637 m/s. It is suggested that typical pitting corrosion occurs on the surfaces of 316L SS specimens. Furthermore, it is found that the passive film on the specimen surface is damaged more and more seriously with an increase of U_G. The SEM observations agree well with the electrochemical test and CFD results obtained in this research.

Figure 17. SEM images of 316L SS in the upward slug flow at U_L = 0.637 m/s. U_G (m/s): (a) 0.106; (b) 0.159; (c) 0.212; (d) 0.265.

6. Conclusions

Upward slug flows were formed in the circulating system from nitrogen and sodium chloride solution. Corrosion characteristics of AISI 316L SS in the slug flow were initially measured using electrochemical approaches. Also, the effects of shear stress and mass transfer coefficient on corrosion rate were revealed by numerical simulations. According to the results, an increase of U_G enhanced the lengths of the Taylor bubble and liquid slug and the latter has a greater amplification than the latter. At the same time, the liquid film thickness is decreased. As U_L increases, the lengths of liquid slug and Taylor bubble are decreased. However, the nose and wake of Taylor bubble are relatively insensitive to U_L. The 316L SS corrosion rate in two-phase slug flows is much greater than that in the single-phase liquids. The shear stress of liquid fluid and mass transfer coefficient of reactants get larger with increasing U_G and U_L, resulting in an increase in the 316L SS corrosion rate. In addition, U_G has a relatively greater impact on the 316L SS corrosion rate than does U_L. Upward slug flow may destroy the passivation film on the 316L SS specimen surface due to its intermittency and volatility, accelerating the pitting corrosion process. When U_G and U_L are changed drastically, the slug flow may transform into other flow patterns. However, there are still no strict boundaries for the transition of slug flow. Therefore, this study has only focused on 316L SS corrosion characteristics in typical slug flows. Future studies could focus on corrosion characteristic in the transition process.

Nomenclature

DB	=	Diameter of Taylor bubble (m)
DN	=	Diameter of bubble nose (m)
Dd	=	Diffusion coefficient (m2/s)
I0	=	Corrosion current density (A/cm2)
F	=	Faraday’s constant
Frs	=	Froude number
K	=	Mass transfer coefficient (m/s)
L	=	Characteristic length (m)
LB	=	Length of Taylor bubble (m)
LN	=	Length of bubble nose (m)
LW	=	Length of wake region (m)
LS	=	Length of liquid slug (m)
LU	=	Length of slug unit (m)
P	=	Average static pressure (Pa)
Rp	=	Polarization resistance (Ω)
Rf	=	Resistance of passive film (Ω)
Rs	=	Ohmic resistance of the solution (Ω)
TF	=	Thickness of liquid film (m)
UL	=	Superficial liquid velocity (m/s)
UG	=	Superficial gas velocity (m/s)
ba	=	Anode Tafel slope of the common logarithmic polarization curve
bc	=	Cathode Tafel slope of the common logarithmic polarization curve
k	=	Turbulent kinetic energy (J)
ma	=	Relative atomic mass of 316L SS
n	=	Moles of electrons transferred in electrochemical action (mol)
t	=	Time (s)
u	=	Characteristic velocity (m/s)
ui	=	Fluid velocity in the direction of axis i (m/s)
v0	=	Weight reduction caused by slug flow accelerated corrosion (g)
vd	=	Average reduction of specimen thickness (mm)
xi	=	ith spatial coordinate
v	=	Average velocity vector
αq	=	Volume fraction of the q phase
β	=	Taylor bubble length fraction
μ	=	Fluid dynamic viscosity (Pa·s)
μeff	=	Effective viscosity (Pa·s)
μT	=	Turbulent viscosity (Pa·s)
ε	=	Dissipation rate of turbulent kinetic energy
τ	=	Shear stress (Pa)

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by National Natural Science Foundation of China: Grant Number 51806198, 51876194.