Experimental study on corrosion of steels for flue gas reheaters in a coal-fired power plant

Accepted Manuscript Research Paper Experimental study on corrosion of steels for flue gas reheaters in a coal-fired power plant PeiYuan Pan, Heng Chen, Zhiyuan Liang, QinXin Zhao PII: DOI: Reference:

S1359-4311(16)34151-5 http://dx.doi.org/10.1016/j.applthermaleng.2016.12.066 ATE 9686

To appear in:

Applied Thermal Engineering

Received Date: Accepted Date:

16 July 2016 15 December 2016

Please cite this article as: P. Pan, H. Chen, Z. Liang, Q. Zhao, Experimental study on corrosion of steels for flue gas reheaters in a coal-fired power plant, Applied Thermal Engineering (2016), doi: http://dx.doi.org/10.1016/ j.applthermaleng.2016.12.066

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Experimental study on corrosion of steels for flue gas reheaters in a coal-fired power plant PeiYuan Pan, Heng Chen, Zhiyuan Liang, QinXin Zhao* Key Laboratory of Thermal Fluid Science and Engineering of MOE, School of Energy and Power Engineering, Xian Jiaotong University, Xian Shaanxi 710049, China

Abstract The corrosion failures occurring in the back-end of coal-fired power plants are of great importance, and there are many decades of research experience of the corrosion issues before or in the flue gas desulfurization (FGD) unit, but the situation for flue gas reheaters (FGR) which are placed after FGD is rather different. In this paper, in-plant corrosion tests were performed and the corrosion mechanism was studied. The corrosion products were mainly iron oxides and the droplets entrained by the wet flue gas were the major cause. The wall temperature of the tubes in FGR should be kept at a high level to mitigate the corrosion. Keywords: Corrosion; Flue gas; Power plant; Heat exchanger; Stainless steel Highlights： In-plant corrosion tests of five kinds of steels may be used in FGR were conducted. The results were analyzed by XRF, XRD, SEM and EDS. The corrosion mechanism was studied, and the testing steels were evaluated. The corrosive droplets entrained by the wet flue gas were the major cause of the corrosion in FGR. The corrosion rate was negatively correlated with the surface temperature of the steel.

1. Introduction The corrosion issues caused by low-temperature flue gas in the back-end of coal-fired power plants have long been a severe problem and gave rise to a lot of component failures. As the flue gas is cooled, condensation of H2SO4, HCl and water vapor in the flue gas will commence successively. In the flue gas desulfurization (FGD) unit, flue gas is usually cooled from 90°C to about 50°C, lower than the water dew point. The flue gas from outlet of FGD is saturated with water vapor and entrains a large number of droplets. Thus it could be called wet flue gas. The droplets entrained by this wet flue gas are mainly composed of condensate from the flue gas and absorber slurry which is used for desulfurization in FGD. These droplets are highly corrosive and make the wet flue gas extremely aggressive, threatening the safety and integrity of downstream Corresponding author *E-mail address: [email protected]. Tel.: +86-029-82665710.

flues, ducts and stacks. Furthermore, due to the low temperature, the wet flue gas is not buoyant enough for the diffusion of emissions and may lead to falling of gypsum around the plant. Usually tube-type heat exchangers called flue gas reheater (FGR) are applied to reheat the wet flue gas to 75~90°C thereby evaporating the corrosive droplets and minimizing acid condensation downstream. The plume rise height also increases with the rise in temperature of the wet flue gas. Rotary gas-gas heaters were once widely used in some districts to reheat the flue gas but have been gradually replaced by FGR due to the high risk of corrosion and blockage. In the past several decades, a number of studies have been conducted on the low-temperature corrosion in heat exchangers in coal-fired power plants. But most of the studies focused on the sulfuric acid dew point corrosion before FGD [1-11]. Dew point corrosion occurs in condensing flue gas, with condensation of acid vapors. According to these studies, the dew point of the H2SO4 in flue gas is usually in the range of 120 to 150°C, depending on the H2SO4 and water vapor concentrations. Taking into account the ash in the flue gas before the electrostatic precipitator (ESP), the practical dew point of H2SO4 is around or slightly below 90°C. In these studies, only the corrosion caused by sulfuric acid was considered. Because the acknowledged dew point of HCl in flue gas is usually below 60°C, and little chlorine is involved in the corrosion before FGD. However, the condition in FGR, which is placed after FGD, is quite different. Firstly, there is indeed some HCl, H2SO4 and other kinds of acid vapors in the wet flue gas from outlet of FGD. But technically, the corrosion occurring in FGR cannot be classified into dew point corrosion. Since the wet flue gas is heated in FGR, little new acid vapor will condense. It is evaporation of the existing droplets instead of acid condensation that occurs in FGR. The corrosive droplets entrained by the wet flue gas are the major cause of the corrosion. Secondly, because the temperature of the wet flue gas is around 50°C, most of HCl and H2SO4 in the flue gas have condensed before leaving FGD. The concentrations of chlorine ions and sulphate ions in the absorber slurry are also very high. Consequently, chlorine ions and sulphate ions in the droplets entrained by the wet flue gas should be both taken into account. Thirdly, with the help of ESP and FGD, the ash in the wet flue gas in FGR is too little to impede corrosion. There are also some studies about the corrosion occurring in FGD or in rotatory gas-gas heaters, where the influence of chlorine ions was emphasized [12-15]. But due to the great differences in structure and function, the situation for FGR, a tube-type heat exchanger, is completely different. In this paper, the corrosion caused by the corrosive wet flue gas in FGR was the major focus. In-plant corrosion tests were conducted. Then mechanism of the corrosion was studied and five kinds of steels which may be used to manufacture the heat transfer components in FGR were evaluated.

2. Experimental 2.1

Materials

Five kinds of frequently-used steels in corrosive flue gases were tested in this paper, including low alloy steel 09CrCuSb, austenitic stainless steels S30403 and S31603, duplex stainless steel S32205 and super duplex stainless steel S32750. Their chemical compositions are listed in Table 1, and pitting resistance equivalent numbers (PREN) of the four stainless steels are also calculated by the following equation:

2.2

Apparatus

Surface temperature of the steel is the dominant factor affecting low-temperature corrosion in coal-fired power plants. However, simple coupons without temperature control were used in most field tests. Hence a new type of double-tube corrosion probe was designed to study the corrosion resistance of different kinds of steels in flue gas at different surface temperatures. Which was specialized for in-plant tests. The schematic diagram of the corrosion probe is shown in Figure.1. The corrosion probe consisted of a flange and two coaxial tubes with different diameters, which were fixed together by welding. Circulating water could flow in through the inside tube and flow out through the annular gap between the inside tube and outside tube. All the testing steels were in the form of short circular tubes with the same diameter and thickness. And five such short tubes of different testing steels were welded together to make up an integrated outside tube. The connecting sequence of the steels was S31603, S32205, S32750, S30403 and 09CrCuSb. The inside tube and the flange were made of 09CrCuSb. The tubes could be inserted into the flue and maintained perpendicular to the direction of flue gas. The flange could be fixed on the flue wall thus the corrosion probe could be kept in place. A circulating pump with heating function was used to supply circulating water at a constant temperature. The circulating water kept flowing in and out. The outer wall temperature of the outside tube was approximately equal to the temperature of the circulating water. Consequently, the surface temperature of the testing steels could be controlled.

2.3

Condition

In-plant tests were done in a 600MW coal-fired power plant in China. The ultimate analysis of the coal used during the tests is shown in Table 2. Both the sulphur and chlorine contents were quite low. Typical limestone-gypsum wet flue gas desulfurization was applied in this plant. Lime slurry was used as desulfurizer in FGD to neutralize sulfur dioxide and produce gypsum. The characteristic parameters of the wet flue gas at outlet of FGD during the tests were measured online and the average values are listed in Table 3. The velocity of the flue gas was around 12m/s.

2.4

Procedure

The testing position was located after FGD and before the stack (see Figure.2). FGR had not been installed in this power plants yet. The double-tube corrosion probes were used for the tests. Under most conditions, wall temperature of the tubes in FGR ranges from 50°C to 100°C, mainly depending on temperature of the water in the tube. Five tests were done with different surface temperatures of the testing steels, which were controlled by the circulating water. The temperatures were 50, 60, 70, 80 and 90°C. The period of each test was 72h. After the tests, the probes were taken out of the flue, and the outside tubes were cut apart for observation and analyses.

3. Results 3.1

Visual inspection

Appearances of the corroded and unused outside tubes are shown in Figure.3. All the photos were taken from the windward side of the tubes, because the corrosion on the leeward side was relatively slight. The surfaces of the unused steels were smooth, and the color was silver grey with metallic luster. At the wall temperature of 50°C, the austenitic stainless steels S30403 and S31603 both exhibited obviously general attack since their colors had turned rusty. The windward surface of low alloy steel 09CrCuSb was covered by a tan layer which could be easily scraped off by a blade and the color of the substance under this layer was dark black. Thus it can be inferred that a higher level of general attack might also took place. As for duplex stainless steel S32205 and super duplex stainless steel S32570, rusty stains appeared, indicating some degree of localized attack. The situations at 60°C and 70°C were similar to that at 50°C. However, there was little deposit on 09CrCuSb and the color of its surface was reddish black. At 80°C, the corrosion on S31603, S32205 and S32570 was quiet slight. While S31403 exhibited some localized attack and 09CrCuSb still exhibited some general attack. When the wall temperature went up to 90°C, hardly any attack was observed on all the four stainless steels. And 09CrCuSb had just become dim and dark.

3.2

Analysis of droplets entrained by the wet flue gas

As illustrated in Introduction, the droplets entrained by the wet flue gas were the major cause of the corrosion occurring in FGR. Numerous droplets were collected from the wet flue gas for analysis. The obtained liquid was turbid and whitish. The characteristic parameters of the liquid were listed in Table 3. According to the data in Table 3, the droplets in the wet flue gas were slightly acidic. Chlorine ion was the dominating anion, the concentration of which was up to 269.2mmol/L, equal to nearly 10000mg/L. The sulfate ion concentration was less than one-tenth of chlorine ion. And other anions were very little. As for metal ions, the calcium ion showed the highest concentration. Which mainly came from the absorber slurry because lime was used as the desulfurizer. In conclusion, the droplets were highly corrosive, especially considering the strong aggressivity of chlorine ions.

3.3

Analysis of the scraped sample

The sample scraped off from the windward surface of low alloy steel 09CrCuSb at the wall temperature of 50°C was analyzed by X-ray fluorescence (XRF) and X-ray diffraction (XRD). The results are shown in Table 4 and Figure.4 respectively. The XRF and XRD results suggested that the corrosion products were mainly iron oxides, in the forms of goethite and hematite. Sulphur was mainly combined as gypsum, which was actually not produced by corrosion but the product of limestone-gypsum desulfurization process and the main component of the solids in absorber slurry. Despite the demisters in FGD, some absorber slurry was inevitably carried into the flue after FGD and left gypsum on the tubes.

Chlorine was also detected by XRF, but not much enough to produce a peak on the XRD pattern.

3.4

Micromorphological analysis

After the field tests, small samples of the corroded testing steels were cut off from the outside tubes, embedded in epoxy resin and polished to a smooth and lustrous shine. So that the cross sections of the tubes could be observed by a scanning electron microscope (SEM). According to the visual inspection, the corrosion on the windward side were more severe than that on the leeward side. Which was also verified by microscopic observation. Typical cross-sectional SEM micrographs of the corrosion layers of low alloy steel 09CrCuSb on the windward side, lateral side and leeward side (see Figure.5) at different wall temperatures are shown in Figure.6, Figure.7 and Figure.8 respectively. Take Figure.6 (a) for example, the dark grey part on the right was the corrosion layer, while the light grey part on the left was the steel substrate. According to Figure.6, Figure.7 and Figure.8, it was likely that 09CrCuSb had suffered uniform corrosion at all temperatures. Its corrosion layers were compact, even and coherent. Hence, thicknesses of the corrosion layers in these SEM micrographs could be measured with the scale. Dozens of such micrographs were taken at different positions on the cross sections of 09CrCuSb outside tubes, and thicknesses of the corrosion layers were measured one by one. Then the average thickness of each corrosion layer was calculated and shown in Figure.9. The corrosion rate of 09CrCuSb under the corresponding condition could also be reflected in Figure.9, since the thickness of a corrosion layer was positively correlated with the corrosion rate. Generally, in the testing range, the corrosion rate of 09CrCuSb became lower as the wall temperature went up. The thickness of the corrosion layer at 60°C was quite close to that at 50°C, and the corrosion layer on the windward side at 70°C was just about 29μm thinner than that at 50°C. However, there seemed to be a sudden drop in the level of corrosion around 70°C. Since the thicknesses of the corrosion layers at 80 and 90°C were much thinner. Moreover, at a certain temperature, the corrosion rate was the highest on the windward side, followed by the lateral side and lowest on the leeward side, consistent with previous visual inspection. And such difference was greater at a higher wall temperature. At 50 and 60°C, the corrosion on the leeward side was appreciable compared to the windward side. While when the wall temperature exceeded 70°C, the corrosion on the leeward side became insignificant. Although the corrosion behavior of 09CrCuSb was dominated by uniform corrosion, at the border of corrosion layer and steel substrate, pitting corrosion could also be observed (see Figure.6 (a)). Typical cross-sectional SEM micrographs of the corrosion layers of austenitic stainless steel S30403 on the windward side at different wall temperatures were shown in Figure.10. S30403 had suffered different degrees of uniform corrosion at 50, 60 and 70°C. But the corrosion layers were uneven, incoherent, porous and relatively thin. Furthermore, owing to low levels of uniform corrosion, obvious pitting corrosion was frequently observed (see Figure.10 (b) and (c)). At 80 and 90°C, there was only some localized corrosion. In addition, corrosion on the leeward side was also slighter. The situation for another austenitic stainless steel S31603 was quite similar to S30403. Typical cross-sectional SEM micrographs for S31603 were shown in Figure.11. At 50, 60 and 70°C, uniform corrosion dominated, while localized corrosion prevailed at 80 and 90°C. S31603 showed apparently better corrosion resistance than S30403, by thinner corrosion layers. But pitting

corrosion still existed. It is worth mentioning that the corrosion rates of S30403 and S31603 at 60°C were even a little higher than those at 50°C. Of all the testing steels, duplex stainless steel S32205 and super duplex stainless steel S32750 exhibited the best performance. At all temperatures, the uniform corrosion rates were negligible. In fact, there was hardly any corrosion at 70, 80 and 90°C. Some of cross-sectional SEM micrographs of the rare corrosion areas of S32205 and S32750 were shown in Figure.12. Average thicknesses of the corrosion layers of the testing steels on the windward side at different wall temperatures are listed in Table 5. The values for S30403 and S31603 at 80 and 90°C, S32205 and S32750 at all temperatures were not meaningful due to low uniform corrosion rates.

3.5

Elemental analysis of the corrosion layers

The elemental composition of the corrosion layers also needed to be learned to study the corrosion mechanism. Small areas of the corrosion layers were analyzed by energy dispersive spectrometer (EDS). Figure.13 to Figure.17 show some of such areas, which were enclosed by green rectangles on the micrographs. All the corrosion layers in these micrographs were on the windward side of the tubes. The EDS results are listed in Table 6. According to the data in Table 6, major elements of the corrosion layers were found to be iron and oxygen. Thus iron oxides made up the backbone of the corrosion products, consistent with the results by XRD. Generally, the oxygen content was relatively higher in the outer corrosion layer than in the inner corrosion layer, and also higher at a lower temperature than at a higher temperature. Sulphur and chlorine also existed, their contents varied between 0.5 and 5wt%. The chlorine content was higher in the inner corrosion layer than in the outer corrosion layer, whereas sulphur showed an opposite trend. Furthermore, compared with low alloy steel 09CrCuSb, the corrosion layers of austenitic stainless steels S30403 and S31603 showed lower oxygen contents but higher sulphur and chlorine contents. Which means that more iron was combined as chlorides and sulfates instead of oxides in the corrosion products of stainless steels. The corrosion in Figure.13 was formed at the wall temperature of 50°C. For low alloy steel 09CrCuSb, the contents of sulphur and chlorine in Area 2 and 3 were approximately 2wt% or below. However, Area 1 was different. It showed a pitting at the border of corrosion layer and steel substrate, where the chlorine content shot up to 4.66wt%, indicating a gathering of chlorine under the corrosion layer. It seemed that the pitting corrosion, which would grow into uniform corrosion later, was mainly induced by chlorine ions. The corrosion in Figure.14 was formed at 60°C. The chlorine content in the corrosion layer more than doubled from 50 to 60°C. While the sulphur content decreased. The corrosion in Figure.15 was formed at 70°C. There was a slightly decline in the chlorine content and a rise in the sulphur content from 60 to 70°C. The conditions at 80°C (see Figure.16) and 90°C (see Figure.17) were a little different. Though the sulphur content kept growing slowly to about 2wt%, there was hardly any chlorine in the corrosion layers. In addition, the corrosion of the stainless steels at 90°C (see Figure.17 (b) and (c)) was too slight to be analyzed accurately by EDS.

4. Discussion In this section, based on the experimental results and analyses, the mechanism of the corrosion caused by the wet flue gas will be discussed step by step.

4.1

Motion of the droplets

As mentioned above, the real major cause of the corrosion in FGR was the corrosive droplets entrained by the wet flue gas, which were formed in FGD and carried into the flue after FGD. Based on the comprehensive study of the corrosion behaviors in this paper, a two-dimensional motion model of a single droplet in the wet flue gas was proposed (see Figure.18). At the beginning, the droplet was entrained by the wet flue gas. The droplet and the flue gas shared a same velocity (see Figure.18 (a)). When coming close to the high-temperature tube, the droplet was being heated. Evaporation occurred on the surface of the droplet and thus the droplet became smaller in size (see Figure.18 (b)). Due to the loss of water, the solution in the droplet was concentrated simultaneously. Then the flue gas would flow around the tube, but the droplet might strike on the tube because of inertia. After arriving at the windward side of the tube, the droplet would transform into a thin liquid film and flow forward along the wall of the tube. Since the film continued being heated during the flow, evaporation on the surface and concentration of the solution would not stop. Thus the film would become thinner and thinner, and finally disappear (see Figure.18 (c)). Except for being concentrated, the liquid in the film generally had the same composition as that in the droplet. Accordingly, the film on the tube was also very corrosive, resulting in corrosion of the steel. But if the wall temperature was high enough, or the droplet was small enough, the droplet would evaporate completely before reaching the tube. The motion of most droplets could be interpreted using this simplified motion model. It's important to note that the size of the droplet and the spread of the liquid film in the diagram were both significantly amplified to make the process clear. In practice, the droplets could not always strike on centerline of the windward side, and their paths were inevitably influenced by ambient flue gas. The area of each liquid film was also extremely small. However, as a whole, the droplets struck usually on the windward side, sometimes on the lateral side, and very occasionally on the leeward side. Therefore, at a moderate temperature, there were always many droplets, especially the larger ones, striking directly on the windward side and transforming into liquid films. The droplets striking on the lateral side were less. But the lateral side was likely to be influenced by the liquid films formed on the windward side. Little droplets would struck on the leeward side, and the liquid films on the leeward side mainly came from other parts of the tube. To sum up, the proposed motion model was reasonably representative.

4.2

Corrosion process

In the following discussion on the corrosion process, of all the metal elements in steels, only Fe was considered. Based on motion of the droplets, steels dissolved when reacting with the liquid films on the tubes. The corrosion process of the testing steels in the wet flue gas could be roughly divided into three stages.

(i) First stage: Initiation. The main chemical reaction in the first stage was: Fe+2H+→Fe2++H2↑ (steel corroding, hydrogen evolution) (1) + + The corrosion of the steel was initiated by H in the liquid films. H from the liquid films formed by the first few droplets, reacted with Fe from the steel, resulting in hydrogen evolution (reaction (1)). According to the analysis of the droplets, the H+ concentration in the films was very low, thus the corrosion rate in the first stage was very slow. But this stage didn’t last long, and the corrosion soon went forward to the next stage. (ii) Second stage: Acceleration. The main chemical reactions in the second stage were: 4Fe2++2H2O+O2→4Fe3++4OH- (oxygen reduction) (2) 3+ 2+ 2Fe +Fe→3Fe (steel corroding) (3) 3+ Fe +3OH →FeOOH↓+H2O (product generation) (4) 2FeOOH→Fe2O3↓+H2O (product generation) (5) 3+ Fe +3Cl →FeCl3 (product generation) (6) 3+ 2 2Fe +3SO4 -→Fe2(SO4)3 (product generation) (7) In this stage, the chemical reactions mainly took place on the surface of the steel, and oxygen reduction became the predominant cathodic reaction. The Fe2+ generated in the first stage was immediately oxidized to Fe3+ by oxygen (reaction (2)). Fe3+ reacted with Fe from the steel, thus the steel corroded and Fe3+ was reduced to Fe2+ again (reaction (3)). But Fe3+ would then be regenerated (reaction (2)). In consequence, Fe3+ acted as a catalyzer and significantly accelerated the corrosion. Reaction (4) to (7) show generation of corrosion products. With an abundance of Fe3+, insoluble iron oxides FeOOH and Fe2O3 were generated continuously (reaction (4) and (5)). They were the main corrosion products in the second stage and gradually formed a corrosion layer on the surface of the steel. Iron chloride FeCl3 and iron sulfate Fe2(SO4)3 were also generated (reaction (6) and (7)). FeCl2 and FeSO4 were little because Fe2+ was readily to be oxidized to Fe3+ by the oxygen in the flue gas. Unlike the iron oxides, FeCl3 and Fe2(SO4)3 were very soluble. They might tend to dissolve in the liquid films and be ionized into free ions again. Only some of the FeCl3 and Fe2(SO4)3 could be deposited in the corrosion layer. Therefore, the oxygen content in the corrosion layer was high, while the chlorine content and sulphur content were quite low. (iii) Third stage: Expansion. The main chemical reactions in the third stage were: Fe2++2Cl-→FeCl2 (product generation) (8) + FeCl2+2H2O→Fe(OH)2↓+2H +2Cl (product generation) (9) Fe(OH)2+ 2Fe(OH)3→Fe3O4↓+4H2O (product generation) (10) In this stage, the chemical reactions mainly took place at the border of the previous corrosion layer and steel substrate. With thickening of the corrosion layer, reaction (2) was inhibited due to the insufficiency of oxygen. But Fe in the steel continued losing electrons and there was a sharp increase in the amount of Fe2+. Thus a great deal of Cl- moved from the liquid film to the border to maintain electrical neutrality, resulting in a high level of chloride (reaction (8)). And pitting corrosion under the corrosion layer was also induced by Cl-. Little sulphate was generated due to

the low mobility of SO42-. FeCl2 was prone to be hydrolyzed into insoluble Fe(OH)2 (reaction (9)) and Cl- was regenerated. Fe3O4 was also generated at the border (reaction (10)). Simultaneously, the environment at the border was acidified by the generation of H+ (reaction (9)). The corrosion at the border was exacerbated due to the low pH and high Cl - concentration. Then more Fe2+ was generated and more Cl- moved in. With the lack of oxidant and the protection of outer corrosion layer, the generated Fe(OH)2, Fe3O4 and FeCl2 could exist stably in the corrosion layer adjacent to the steel substrate. Expansion of the corrosion layer was based on the continuous generation of these corrosion products. Therefore, the inner corrosion layer which was newly formed showed a higher chlorine content and a lower oxygen content than the outer corrosion layer. It is important to note that although the diffusion of oxygen was hindered by the corrosion layer, the Fe(OH)2, Fe3O4 and FeCl2 generated in the third stage would finally be oxidized to form FeOOH, Fe2O3 and FeCl3. The oxidation rate was negatively correlated with the thickness of the corrosion layer above. The third stage was the last stage and the thickness of the corrosion layer would increase steadily from then on. In addition, a protective passive film is usually formed on the surface of a stainless steel. But in this paper, the passive film was vulnerable to Cl- in the liquid films (reaction (11)). During the corrosion process of the stainless steels, only the generation of iron oxides was slowed significantly. Therefore, compared to the low alloy steel, chorine and sulphur contents were higher in the corrosion layers of the stainless steels. Fe2O3+6Cl-+3H2O→2FeCl3+6OH(11) The schematic diagram of the steel substrate, corrosion layer and liquid film is shown in Figure.19. Representative components are also listed. The components in bold are the main corrosion products.

4.3

Influence of wall temperature

The influence of surface temperature of the steel may be the most significant on low-temperature corrosion. Which could be discussed from different aspects to explain the experimental phenomena. (i) Reaction rate. Based on the theory of molecular dynamics, the reaction rate increases with rising temperature. But under the conditions in this paper, the amount of the reactants in the liquid films was quite small. In practice, the reactant depletion was much faster than the reactant supply. Therefore, the reaction rate had little influence on corrosion rate. (ii) Deposition rate. The corrosion of the steel was limited by the reactant supply. Thus the deposition rate of the liquid films on the tube seemed to be the dominant factor in the corrosion in this paper. According to the proposed motion model of a single droplet in the wet flue gas, it could be inferred that droplets of the same size would transform into liquid films of different sizes at different wall temperatures of the tube (see Figure.20). When wall temperature of the tube was close to the temperature of the wet flue gas, evaporation on the surface of the droplet and the liquid film was mild. If the droplet was big enough, the formed film might be able to spread from the windward side to the leeward side

(see Figure.20 (a)). In addition, the film is thicker on the windward side than the leeward side. As the wall temperature of the tube increased, the size of the droplet arriving at the tube decreased. The formed liquid film also became thinner and the area covered by the liquid film decreased further. The film might not be able to reach the leeward side or even the lateral side (see Figure.20 (b)). If wall temperature of the tube was high enough, the droplets would evaporate completely before arriving at the tube (see Figure.20 (c)). The deposition rate on the tube could be deduced from the behavior of the single droplet. Overall, the deposition rate was higher at a lower temperature, resulting in a higher corrosion rate and a thicker corrosion layer. And the highest level of corrosion occurred on the windward side of the tube where the deposition rate was the highest. For 09CrCuSb, at 50 and 60°C, the evaporation was mild and the tube was wholly wetted. There was only a slight difference between the deposition rates on the windward side and leeward side. Thus the difference between the corrosion rates was also slight. At 70°C, the evaporation was accelerated due to a larger temperature difference and most of the liquid films formed by the droplets might not be able to reach the leeward side. So the corrosion rate on the leeward side was much lower than that on the windward side. At 80 and 90°C, the liquid film might disappear soon after their formation. Thus only the corrosion on the windward side was obvious. (iii) Ions concentrations in the liquid films. Concentrations of the ions in the droplets and liquid films increased with the evaporation rate on the surface. According to the analysis of droplets and the discussion about the corrosion process, the main ions affecting the corrosion were chlorine ions, sulphate ions and hydrions. Concentrations of the ions affected the reaction rate. But the corrosion rate was governed by the deposition rate of liquid films, not the reaction rate. Therefore, the effect of ions concentrations on the corrosion rate was not very significant. The concentrations of chlorine ions and sulphate ions seemed to have influence on the contents of chloride and sulphate in corrosion products. In general, according to the EDS results, the chlorine and sulphur contents in the corrosion layers were higher at higher wall temperatures, which could be explained by the higher concentrations of chlorine ions and sulphate ions in the liquid films. Actually, the influence of chlorine ions was far more complex than this. As discussed above, chlorine ions were very aggressive because of their high activity and mobility. They could induce pitting, break down passive films and exacerbate the corrosion. Even some stainless steels showed little corrosion resistance against chlorine ions. At the wall temperature of 60 and 70°C, the increase in chlorine ion concentration not only leaded to a higher chloride content in the corrosion layer, but also increased the corrosion rate. Especially for the stainless steels, the corrosion rate at 60°C might be even slightly higher than that at 50°C, because usually an increase in chlorine ion concentration could significantly increase the corrosion rate of the stainless steel. However, as the temperature went up, the chloride acid would start to escape from the droplets and liquid films in the form of hydrogen chloride. The SEM results of the corrosion layers of 09CrCuSb showed a sudden drop in corrosion rate at the wall temperature of about 70°C. In addition, according to the EDS results, the chlorine content in the corrosion layer seemed to reach a peak between 60 and 70°C and then fell to nearly zero at 80°C. The acknowledged dew point of HCl in the flue gas of coal-fired power plants was between 50 and 60°C. Accordingly, it could be assumed that the escape of chloride acid started at about 60°C, and most of the

chloride acid had escaped before reacting with the steel at 80°C.

4.4

Evaluation of testing steels

Five kinds of steels were tested in this paper, ranging from low alloy steel 09CrCuSb to super duplex stainless steel S32750. In general, the performance of the steels followed the PREN. 09CrCuSb was initially developed to resist sulfuric acid dew point corrosion. But it showed little resistance to chlorine ions, and uniform corrosion took place at all wall temperatures. Thus 09CrCuSb was not recommended to be used in FGR. The performances of austenitic stainless steels S30403 and S31603 were better. But their resistances to chlorine ions were also weak. The corrosion was acceptable only when the wall temperature was higher than 80°C, because few chorine ions would reach the surface of the steels then. Therefore, the precondition of using S30403 and S31603 in FGR was to keep the wall temperature high enough. To be prudent, the temperature should be over 90°C. The duplex stainless steel S32205 and super duplex stainless steel S32570 showed the best performance of all the testing steels. The high contents of chromium and molybdenum did work. Even the pitting corrosion was impeded. Thus these two steels were adequate to be used in FGR.

5. Conclusions The wet flue gas from outlet of FGD in the coal-fired power plant is remarkably corrosive, threatening the safe running of FGR. Thus the corrosion occurring in FGR needs to be studied. In this paper, in-plant corrosion tests of five steels at different surface temperatures had been conducted. Based on the results of the tests, the following conclusions may be drawn: (1) The corrosive droplets entrained by the wet flue gas was the major cause of the corrosion in FGR. Thus high-efficiency demisters would help to mitigate the corrosion. (2) The corrosion products in this paper were mainly iron oxides. (3) The corrosion rate was positively correlated with the deposition rate, thus the corrosion rate decreased as the wall temperature increased. (4) The sequence of the corrosion resistance of the five testing steels in this paper was: S32750>S32205>S31603>S30403>09CrCuSb, following the PREN. (5) The corrosion could be exacerbated significantly by the existence of chlorine ions in the liquid films below 80°C, even for the stainless steels. Thus in practice, it is best to keep the wall temperature of the tubes in FGR over 80°C.

Acknowledgements This work was supported by National Key Research and Development Program of China (No. 2016YFC0801904) and the Fundamental Research Funds for the Central Universities. And special thanks to Qingdao Daneng Environmental Protection Equipment Incorporated Company for providing corrosion probes used in this paper.

References [1]W.M.M. Huijbregts, R.G.I. Leferink, Latest advances in the understanding of acid dewpoint corrosion, Anti-Corrosion Methods and Materials, Volume 51 Number 3(2004) 173-188. [2]H. DR, Dewpoint corrosion, Institution of Corrosion Science and Technology, Birmingham,

1985. [3]Y. Wang, H. Ma, Z. Liang, H. Chen, Q. Zhao, X. Jin, Experimental study on dew point corrosion characteristics of the heating surface in a 65 t/h biomass-fired circulating fluidized bed boiler, Applied Thermal Engineering, 96(2016) 76-82. [4]Z.Y. Liang, Q.X. Zhao, Y.G. Wang, Y.X. Li, Z.C. Zhang, Coupling mechanism of dew point corrosion and viscous ash deposits, Materials and Corrosion, 65(2014) 797-802. [5]H. Han, Y.L. He, W.Q. Tao, A numerical study of the deposition characteristics of sulfuric acid vapor on heat exchanger surfaces, Chemical Engineering Science, 101(2013) 620-630. [6]Y. Wang, Q. Zhao, Z. Zhang, Z. Zhang, W. Tao, Mechanism research on coupling effect between dew point corrosion and ash deposition, Applied Thermal Engineering, 54(2013) 102-110. [7]Z. Li, F. Sun, Y. Shi, F. Li, L. Ma, Experimental study and mechanism analysis on low temperature corrosion of coal fired boiler heating surface, Applied Thermal Engineering, 80(2015) 355-361. [8]P.D. MOSKOVITS, Low-Temperature Boiler Corrosion and Deposits, Industrial and engineering chemistry, 51(1959) 1305-1312. [9]Y. Wang, Q. Zhao, Z. Zhang, Z. Zhang, W. Tao, Mechanism research on coupling effect between dew point corrosion and ash deposition, Applied Thermal Engineering, 54(2013) 102-110. [10]B. Xiang, M. Zhang, H. Yang, J. Lu, Prediction of Acid Dew Point in Flue Gas of Boilers Burning Fossil Fuels, Energy & Fuels, (2016). [11] LI Fei, SUN Fengzhong, SHI Yuetao, Hot-state experiment on double-pipe heat exchanger under coupled effect of fouling and acid, CIESC Journal,65 (2014) 2876-2881. (In Chinese) [12]R. Pettersson, E. Sophia, Corrosion of austenitic and duplex stainless steels in flue gas cleaning systems for waste combustion processes, Outokumpu corrosion management news, 1-2 (2013) 2-10. [13]M.T. Kim, S.Y. Chang, O.Y. Oh, J.B. Won, H.W. Park, Failure analysis of enamel-coated carbon steel heating elements of gas-gas heater for flue gas desulfurization system, Engineering Failure Analysis, 14(2007) 686-693. [14]CHEN Qichun, BAI Lin, Analysis of gas corrosion after WFGD, Engineering Journal of Wuhan University, 39 (2006) 245-249. (In Chinese) [15] MA Shuangchen, WENG Xiaoyu, WU Wei, ZHU Sijie, CHEN Rujia, Research on the influence of salt water on the corrosion and scaling of the absorption tower, CIESC Journal, 32 (2016) 1-5. (In Chinese)

List of Tables Table 1 Chemical compositions of the testing steels in wt%. Table 2 Ultimate analysis of the coal used during the tests in wt%. Table 3 Characteristic parameters of the wet flue gas. Table 4 X-ray fluorescence analysis of the sample from 09CrCuSb. Table 5 Average thicknesses of the corrosion layers on the windward side at different wall temperatures (μm). (-) denotes no value due to a low uniform corrosion rate. Table 6 EDS results of the areas shown in Figure.13 to Figure.17.

Table 1 Chemical compositions of the testing steels in wt%. 09CrCuSb S30403 S31603 S32205 S32750

Type

%Cr

Low alloy steel Austenitic stainless steel Austenitic stainless steel Duplex stainless steel Super duplex stainless steel

0.90 18.1 16.9 22.0 25.0

%Ni

%Mo

%N

Others

PREN

Cu, Sb 8.1 10.7 5.7 7.1

2.6 3.1 4.0

0.17 0.27

18.1 25.5 35.0 42.5

Table 2 Ultimate analysis of the coal used during the tests in wt%. Elements

Car

Har

Oar

Nar

Sar

Far

Clar

Contents

60.66

3.39

9.13

0.80

0.39

0.0123

0.021

Table 3 Characteristic parameters of the wet flue gas. Characteristic

Gauge

SO2

NOx

O2

Water vapor

Dust

pressure

concentration

concentration

concentration

concentration

concentration

Temperature parameters Values

48.29

-224.77

6.9

18.8

7.5

11.0

2.9

Units

°C

Pa

mg/m3

mg/m3

vol%

vol%

mg/m3

Table 4 X-ray fluorescence analysis of the sample from 09CrCuSb. Elements

Fe

O

S

Cl

Ca

Contents (wt%)

50.1

44.2

3.01

0.97

1.74

Table 5 Average thicknesses of the corrosion layers on the windward side at different wall temperatures (μm). (-) denotes no value due to a low uniform corrosion rate. 09CrCuSb S30403 S31603 S32205 S32750

50°C

60°C

70°C

80°C

90°C

116 29 20 -

112 32 22 -

86 23 15 -

48 -

31 -

Table 6 EDS results of the areas shown in Figure.13 to Figure.17. Areas

Steels

Temp.

O

Fe

Cr

Ni (Mass%)

Mo

S

Cl

45.39 47.33 49.75

49.08 49.81 47.14

-

-

-

0.87 1.44 2.09

4.66 1.42 1.02

1 2 3

09CrCuSb

4

S30403

45.41

38.24

7.08

3.98

-

2.87

2.42

5

S31603

44.12

37.61

8.19

4.14

1.17

2.32

2.45

6 7

09CrCuSb

45.19 47.83

50.20 48.14

-

-

-

0.68 0.94

3.93 3.09

8

S30403

41.52

38.63

9.78

4.18

-

1.37

4.52

9

S31603

39.06

39.50

8.73

4.40

1.87

2.20

4.24

10 11

09CrCuSb

42.88 45.31

53.36 51.59

-

-

-

0.84 1.07

2.92 2.03

12

S30403

42.66

34.80

10.45

5.67

-

2.46

3.96

13

S31603

38.92

41.38

9.34

2.22

1.30

2.55

4.29

14

09CrCuSb

42.68

55.33

-

-

-

1.99

-

15

S30403

42.36

39.93

10.16

4.06

-

2.67

0.82

16

S31603

38.74

39.25

12.02

5.70

1.02

2.76

0.51

17

09CrCuSb

39.50

58.56

-

-

-

1.94

-

50°C

60°C

70°C

80°C 90°C

List of Figures Figure.1. Schematic diagram of the double-tube corrosion probe. Figure.2. Location of the testing position. Figure.3. Appearances of the corroded and unused outside tubes. Figure.4. X-ray diffraction analysis of the sample from 09CrCuSb. Figure.5. Windward side, lateral side and leeward side of an outside tube. Figure.6. Typical cross-sectional SEM micrographs of the corrosion layers of 09CrCuSb on the windward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C. Figure.7. Typical cross-sectional SEM micrographs of the corrosion layers of 09CrCuSb on the lateral side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C. Figure.8. Typical cross-sectional SEM micrographs of the corrosion layers of 09CrCuSb on the leeward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C. Figure.9. Average thicknesses of the corrosion layers of 09CrCuSb. Figure.10. Typical cross-sectional SEM micrographs of the corrosion layers of S30403 on the windward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C. Figure.11. Typical cross-sectional SEM micrographs of the corrosion layers of S31603 on the windward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C. Figure.12. Some of cross-sectional SEM micrographs of the corrosion areas of S32205 and S32750： (a) S32205 at 50°C, (b) S32205 at 60°C, (c) S32205 at 50°C, (d) S32205 at 60°C. Figure.13. Some of the analyzed areas of the corrosion layers at 50°C: (a) 09CrCuSb, (b) S30403, (C) S31603. Figure.14. Some of the analyzed areas of the corrosion layers at 60°C: (a) 09CrCuSb, (b) S30403, (C) S31603. Figure.15. Some of the analyzed areas of the corrosion layers at 70°C: (a) 09CrCuSb, (b) S30403, (C) S31603. Figure.16. Some of the analyzed areas of the corrosion layers at 80°C: (a) 09CrCuSb, (b) S30403, (C) S31603. Figure.17. Some of the analyzed areas of the corrosion layers at 90°C: (a) 09CrCuSb, (b) S30403, (C) S31603. Figure.18. Motion model of a single droplet in the wet flue gas. Figure.19. Schematic diagram of the steel substrate, corrosion layer and liquid film. Figure.20. Behaviors of droplets of the same size at different tube temperatures: (a) a low temperature, (b) a medium temperature, (c) a high temperature.

Flue gas

Flange The outside tube

S31603 S32205 S32750 S30403

The inside tube

09CrCuSb

Water inlet Water outlet

Flue wall Figure.1. Schematic diagram of the double-tube corrosion probe.

Flue gas flow

Testing position Air heater

Boiler

Flue gas reheater

Flue gas cooler

Electrostatic precipitator

FGD

Figure.2. Location of the testing position.

Stack

Figure.3. Appearances of the corroded and unused outside tubes.

Figure.4. X-ray diffraction analysis of the sample from 09CrCuSb.

Flue gas

Windward side Lateral side

Leeward side

Figure.5. Windward side, lateral side and leeward side of an outside tube.

Figure.6. Typical cross-sectional SEM micrographs of the corrosion layers of 09CrCuSb on the windward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C.

Figure.7. Typical cross-sectional SEM micrographs of the corrosion layers of 09CrCuSb on the lateral side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C.

Figure.8. Typical cross-sectional SEM micrographs of the corrosion layers of 09CrCuSb on the leeward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C.

Figure.9. Average thicknesses of the corrosion layers of 09CrCuSb.

Figure.10. Typical cross-sectional SEM micrographs of the corrosion layers of S30403 on the windward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C.

Figure.11. Typical cross-sectional SEM micrographs of the corrosion layers of S31603 on the windward side at different wall temperatures： (a) 50°C, (b) 60°C, (c) 70°C, (d) 80°C, (e) 90°C.

Figure.12. Some of cross-sectional SEM micrographs of the corrosion areas of S32205 and S32750： (a) S32205 at 50°C, (b) S32205 at 60°C, (c) S32205 at 50°C, (d) S32205 at 60°C.

Figure.13. Some of the analyzed areas of the corrosion layers at 50°C: (a) 09CrCuSb, (b) S30403, (C) S31603.

Figure.14. Some of the analyzed areas of the corrosion layers at 60°C: (a) 09CrCuSb, (b) S30403, (C) S31603.

Figure.15. Some of the analyzed areas of the corrosion layers at 70°C: (a) 09CrCuSb, (b) S30403, (C) S31603.

Figure.16. Some of the analyzed areas of the corrosion layers at 80°C: (a) 09CrCuSb, (b) S30403, (C) S31603.

Figure.17. Some of the analyzed areas of the corrosion layers at 90°C: (a) 09CrCuSb, (b) S30403, (C) S31603.

Flue gas

Flue gas

Flue gas

Figure.18. Motion model of a single droplet in the wet flue gas.

Steel subtrate

Corrosion layer

Liquid film

Cl-

Fe alloy elements

Fe(OH)2

FeOOH

Cl-

Fe3O4

Fe2O3

SO42-

FeCl2

Flue gas

+

FeCl3

H

Fe2(SO4)3

O2

Fe3+

Formed in the Formed in the third stage. second stage.

Figure.19. Schematic diagram of the steel substrate, corrosion layer and liquid film.

Flue gas

(a)

Flue gas

(b)

Flue gas

(c)

Figure.20. Behaviors of droplets of the same size at different tube temperatures: (a) a low temperature, (b) a medium temperature, (c) a high temperature.