Failure Analysis of a Waste Heat Boiler of a Sulfur-Burning Sulfuric Acid Plant

The waste heat boiler is a widely employed waste heat recovery device in the petrochemical industry, operating in challenging and harsh environmental conditions. The waste heat boiler is prone to failures due to corrosion. In 2014, Sui et al. [1] simulated the expanded connection between the tube and the tube sheet by ANSYS. It was found that the leakage of the tube-sheet joint was due to stress corrosion. The working medium contained chloride that accumulated in the gap between the tube and the tube sheet, and caused stress corrosion. In 2016, Ding et al. [2,3] conducted related research about the failure of waste heat boilers and analyzed the composition of sedimentary substances on the failed tubes. The failures were due to sulfur in methyl methacrylate that caused dew point corrosion of sulfuric acid. In 2018, Hu et al. [4] analyzed the tube burst accident of a coal-to-methanol waste heat boiler. Stress corrosion cracking was the primary cause of the accident. The low quality of the expanded connection between the tube sheet and the heat exchange tubes promoted the accident. In 2019, Xin et al. [5] studied the leakage of the heat exchange tube of an ethylene waste heat boiler. The leakage was caused by alkali corrosion. In 2020, Zhao [6] conducted research on the boiler water quality and found that the wrong method of drainage at the bottom of boiler also caused stress corrosion cracking. In 2020, Lv et al. [7] analyzed the failure of tubes of a waste heat boiler and found that sulfur trioxide combined with water formed sulfuric acid and caused dew point corrosion. The water-cooled wall and the tube of the liquid drum were perforated and leaked. Sui et al. [8,9] studied the failure of a waste heat boiler with leakage occurring at the tube-to-tube sheet joint. The results showed that the leakage was caused by alkali stress corrosion cracking. Xu et al. [10–12] analyzed the cracking failure of a new heat exchanger during the first startup operation. The main cause of failure was stress corrosion cracking.

While conducting the trial run of a chemical plant's 1.2 × 10⁶ tons/year sulfur-burning sulfuric acid facility, the waste heat boiler experienced five consecutive failures. The first four failures happened in the #2 waste heat boiler, and the last failure happened in the #1 waste heat boiler. Relevant analyses were conducted on the #2 waste heat boiler which suffered the first four failures. Based on inspections and analyses of the tube-sheet joints and the tubes of the waste heat boiler, such as the macroscopic morphology, the microstructure, the composition, the mechanical properties, and the fracture morphology, combined with the operating characteristics of the waste heat boiler, the causes of tube cracking of the waste heat boiler were analyzed.

Figure 1 shows the appearance and the front tube sheet of the #2 horizontal waste heat boiler. The waste heat boiler was a fixed tube-sheet heat exchanger, and its design parameters are shown in Table 1.

The site conditions after the first failure occurred are shown in Figs. 2–5.

The site conditions after the second failure occurred are shown in Figs. 6 and 7.

The site conditions after the fourth failure occurred are shown in Figs. 8 and 9.

The site conditions after the fifth failure occurred are shown in Figs. 10 and 11.

According to the site conditions of these failures, the authors found that the failures focused on the outermost circle of tubes at the bottom of the front tube sheet, and the leakage of the tube-sheet joint and the tube bursts were at the front tube sheet. After the first failure, many solid particles were formed by the refractory material falling off, and there were also some solid particles formed by the refractory material falling off after the second failure.

Cutting and sampling were done from the front tube sheet and tubes of the #2 boiler. The first failure was the tubes of 7-1, 7-2, 9-1, 9-2, and 9-3 of the front tube sheet. The second failure was the tubes of 16-1, 17-1, 17-2, and 18-1 of the front tube sheet. The fourth failure was the tube 19-1 of the front tube sheet. These failures were generally in the outermost circle of tubes at the bottom of the front tube sheet. For comparative study, specimens were cut from tubes 17-3, which had no leakage at the bottom of the front tube sheet, tube 21-19, which had no leakage in the middle of the front tube sheet, and tubes 21-28 that had no leakage at the top of the front tube sheet, and tubes of 161, 17-1, 17-2, and 18-1 of the back tube sheet. The code name of tube 7-1 represented the first tube in the seventh column. The tube distributions at the front tube sheet and the back tube sheet are shown in Fig. 12.

Macro-inspections revealed that there was a tube burst area in the tubes of 7-1 and 7-2, as shown in Fig. 13.

There was obvious thinning at the bottom of the inner wall of tubes 7-1 and 7-2. The top of the inner wall was normal, and the burst area was located in the thinned area. The inner morphology of the tube 7-2 is shown in Fig. 14.

One tube burst area was found by macroscopic inspection in tube 17-1, as shown in Fig. 15.

Similar to that of tubes 7-1 and 7-2, the bottom of the inner wall of tubes 16-1, 17-1, 17-2, 18-1 had obvious thinning, the upper part of the inner wall was normal, and the burst area of tube 17-1 was located in the thinned area.

By macroscopic inspection, the authors found that there was a burst area in tube 191 and rust-red flocculent material and yellow powdery material stuck to the inner wall, as shown in Figs. 16 and 17.

The tubes 17-3, 21-19, and 21-28 of the front tube sheet had no obvious thinning, and the cut lengths of the tubes were limited. The tubes 16-1, 17-1, 17-2, and 18-1 of the back tube sheet showed no significant thinning, and the cut lengths were also limited, as shown in Fig. 18.

Tests mainly involved tubes 7-1 and 7-2 in the first failure, tube 17-1 in the second failure, tube 19-1 in the fourth failure, tube 17-3 without leakage at the bottom of the front tube sheet, tube 21-19 without leakage in the middle part of the front tube sheet, and tube 21-28 without leakage in the upper part of the front tube sheet. The specific sampling and testing plans were as show in Table 2.

The specimens of tube burst areas of tubes 7-1, 17-1, and 19-1 are shown in Figs. 19–21. There was a yellow viscous floc near the tube burst area on the inner wall of tube 19-1.

The tube sheet material was P355GH, which was similar to the domestic Q345R, and the standard was GB713-2014 “Steel sheet for boilers and pressure vessels.” The heat exchange tube material was 20 G, and the standard was GB9948-2013 “Seamless Steel Tubes for Petroleum Cracking.” The compositions of the heat exchange tube and the tube sheet were analyzed by the fluorescent spectrum analyzer, and the results are shown in Table 3. It can be seen that the composition of the tube and the sheet met the requirements of relevant standards [13,14].

Three specimens of tube 16-1 and three specimens of the sheet between tube 17-1 and tube 17-2 were machined, and the mechanical properties of the tube and the sheet were tested. The sizes of the specimens are shown in Fig. 22. Due to the size limitation of the tube sheet, the mechanical property of the sheet was tested on the small specimen, and only the tensile strength was obtained. The test results are shown in Table 4. The mechanical properties of all specimens were within the range required by the relevant standard.

The Vickers hardness tester was used to test the hardness of the tube-sheet joint, as shown in Table 5. Since the tube-sheet joint of the waste heat boiler was a welded structure, the hardness of the heat exchange tube, the welding seam, the tube sheet, and the different positions at different distances from the welding seam were different. In general, the hardness of the heat exchange tube and the tube sheet was relatively small, and the welding seam hardness was significantly higher than that of the heat exchange tube and the tube sheet, and the closer the cracks were to the welding seam, the higher the hardness. The reason for this result was related to the faster cooling rate during the welding process. At the same time, the higher hardness of the welding seam near the crack indicated that the formation and propagation of the crack were more sensitive to the hardness.

Microstructure analyses were conducted in the different areas of the tube-sheet joints of tubes 7-1 and 21-19. The microstructures of the heat exchange tube, the tube sheet, and the welding seam in the tube-sheet joint were analyzed. Figure 23 shows the microstructure of heat exchange tube 20 G. It was a typical ferrite and pearlite structure. The microstructure was uniform and the morphology was normal. The band microstructure was rated as level 3. There was a small amount of granular tertiary cementite at the boundary of the ferrite.

The base material of the tube sheet was P355GH, which is similar to the domestic hot-rolled low-alloy steel Q345R. Its normal microstructure should be ferrite and pearlite. Figure 24 shows the microstructure of the tube sheet. The microstructure was ferrite and pearlite. There was a little granular tertiary cementite at the ferrite grain boundary. The banding microstructures of the tube sheets were very obvious near tubes 7-1 and tubes 21-19, and the direction of the banding microstructure was parallel to the surface of the tube sheet. The existence of the banding microstructure was related to the rolling process of the steel sheet. According to the standard of GB/T 13299-1991 “Method for Evaluating the Microstructure of Steel,” the banded microstructure was rated as level 3 [15]. The banding microstructure caused differences in the tube sheet's mechanical properties in the thickness and diameter directions. Obvious banding microstructure is harmful to material properties.

Figure 25 shows the microstructure in the central area of the tube-sheet joint welding seam. Thick needles and massive ferrites were distributed along the columnar crystals. There were a large number of granular carbides and little inclusions at the grain boundaries and in the crystals.

Cracks were found in the tube-sheet joints of tubes 17-3, 21-19, and 21-28 which did not leak and were located in the lower part, middle part, and upper part of the front tube sheet, respectively. The morphologies of the above three cracks were observed by a scanning electron microscope (SEM), and energy spectrum analyses were conducted on the cracks. The morphology of the crack in the tube-sheet joint of tube 17-3 is shown in Fig. 26, the morphology of the crack in the tube-sheet joint of tube 21-19 is shown in Fig. 27, and the morphology of the crack in the tube-sheet joint of tube 21-28 is shown in Fig. 28. The analysis of the composition of above three cracks are shown in Table 6.

From observations of the SEM, cracks were found in the tube-sheet joints of three nonleaked tubes. The selection of the above three nonleaked tubes was random, and it showed that the cracks in the front tube-sheet joints of the #2 waste heat boiler were ubiquitous. The results of the energy spectrum analysis in the cracks are shown in Table 6. The alkaline substance Na was found in all three cracks, and the Na content in some cracks was high, which confirmed that the cracks in the tube-sheet joints were caused by alkaline stress corrosion cracking. Among the above three tubes that did not leak, the crack length of tubes 17-3 was about 200 μm, the crack length of tubes 21-19 was about 700 μm, and the crack length of tube 21-28 was about 850 μm. Judging from the current observations, there was no obvious tendency that the cracking was more serious near tubes in the lower part of the front tube sheet. The mass fractions of Na in the cracks were 1.03% for lower tubes 17-3, 9.06% for middle tubes 21-19, and 2.87% for upper tube 21-28. Also, there was no obvious tendency of more alkaline substances in the crack of the lower tube. Of course, the initiation and propagation of the stress corrosion cracks were highly random and were affected by many factors such as stress, environment, and material microstructure. The crack lengths and the contents of alkaline substances were different for the difference of sampling locations of the same tube. However, the random detection of the cracks in the tube-sheet joints of the above three nonleaked tubes proved that the stress corrosion cracks were ubiquitous in the front tube sheet of the #2 waste heat boiler. Stress corrosion cracks are usually dendritic. In the three tubes that did not leak, the cracks were still in the initiation or early propagation stage, the crack lengths were short, and the cracks did not appear dendritic which was normal.

The inner and outer surface of burst areas of tubes 7-1, 7-2, 17-1, and 19-1 were observed by a stereo microscope and a scanning electron microscope, and energy spectrum analyses were conducted. The morphologies of the inner surface and the outer surface of burst areas of tubes 7-1 and 17-1 are shown in Figs. 29–32, and the analysis of the composition of the inner and outer surfaces of the burst areas of relevant tubes are shown in Table 7. Energy spectrum analyses were conducted on the yellow powder that was generally present on the inner wall of the tubes and the yellow flocs present on the inner wall of the burst area of tubes 19-1. The results are shown in Table 8.

According to the stereo microscope and the SEM observations of the inner and outer surface morphologies of the burst areas of tubes, as well as the previous macroscopic inspections, the tube wall near the burst area of tubes had been significantly thinned, and it was thinned from the inner wall to the outer wall, the inner wall had strong corrosion marks, and a large number of corrosive products were attached to the inner wall of some specimens. The results of the energy spectrum analysis of the burst area of the tube are shown in Table 7. The main components on the inner and outer walls of the burst tube were S, O, and Fe. The energy spectrum analysis results of the yellow powder and the yellow floc are shown in Table 8. The specific composition of the yellow powder was further determined by the X-ray Powder diffractometer as shown in Fig. 33. It can be seen from Table 8 and Fig. 33 that the ubiquitous yellow powder on the inner wall of the tube was Fe₂(SO₄)₃ produced after the tube was corroded by the sulfuric acid, and the yellow viscous floc on the inner wall of tube 19-1 was a mixture of Fe₂(SO₄)₃ and asbestos from the corundum casing. In each specimen, the S content of the inner wall was significantly higher than that of the outer wall, which further proved that the burst was caused by sulfuric acid corroding the tube wall from the inside. Under normal circumstances, the flue gas of tubes is sulfur dioxide, and a small amount of sulfur trioxide, oxygen, nitrogen, and moisture; there is no strong corrosive sulfuric acid. However, due to the previous confirmation that the stress corrosion cracks had been ubiquitous in the tube-sheet joints, the leakage of the tube-sheet joint of the front tube sheet had occurred many times in previous failures. Therefore, it can be confirmed that these failures were caused by the alkali stress corrosion cracking in the tube-sheet joint of the front tube sheet, which caused water to enter into the front smoke box. After the water and the flue gas were mixed, sulfuric acid was formed and moved into the tubes, causing strong corrosion of the tubes. In particular, when the temperature dropped as the failure happened and the device was shut down, the gaseous sulfuric acid condensed into a liquid state, formed the dew point corrosion, and tube wall thinning occurred, and the tube burst under the action of pressure. As for the detected elements such as Si, K, Ca, and especially K and Ca, which did not belong to the tube, they were left after the refractory material fell off and the solid particles eroded the tube.

In summary, the reason for the failure of the #2 waste heat boiler was that the boiling time was too long in the presence of a high concentration of alkali. That led to the common occurrence of alkali stress corrosion cracking in the tube-sheet joints. After cracking and leaking, the water and the high-temperature flue gas formed sulfuric acid, which backwashed into the tube and caused strong corrosion to the tube wall. As the tube wall thickness became thinner, the tube eventually burst. The erosion of the solid particles formed by the falling of refractory materials in the early stage promoted the thinning of the tube wall.

After the second failure, the results of nondestructive testing of the #1 and #2 boilers also confirmed the above conclusions. According to radiographic inspections, there were a large number of connected dense pores at the welding seam roots of all inspected tubes. In the above cases, the alkaline stress corrosion cracking generally occurred in the roots of tube-sheet joint welding seams. The morphology of the alkali stress corrosion cracking was similar to the morphology of dense pores on the radiographic image. In addition, ultrasonic testing was conducted on the wall thickness of the tubes around the burst tubes on the front tube sheet in the #2 waste heat boiler. The inspected tubes generally had some corrosion pits in the inner wall. It also confirmed that the conclusion of the above-mentioned sulfuric acid backwashed into the tube caused the tube wall corrosion.

The working environment of the waste heat boiler during the boil-out with the alkali period was harsh. The alkali concentration and the phosphate concentration of the boiler water were high, so it was more prone to condense. The lye passed through the annular gap between the tube and the sheet to contact the welding seam, which created environmental conditions for alkali corrosion. The numerical analysis results provided by the designer and the manufacturer showed that the outermost three circles of the tubes had larger stresses, and the outermost circle of the tubes had the largest stress. At the same time, due to the large number of tubes in the waste heat boiler, the outermost circle of tubes was on the transition arc of the tube sheet, and the stress concentration was more obvious. This also provided necessary conditions for the stress corrosion cracking. The temperature of the front tube sheet was as high as 1213 °C, which was much higher than the 375 °C of the back tube sheet, and the stress corrosion crack growth rate of the front tube sheet was also much higher than that of the back tube sheet. From this point of view, the outermost circle of tubes on the front tube sheet was in the most dangerous position of the alkali stress corrosion.

The schematic diagram of the structure of the waste heat boiler is shown in Fig. 34. The boiling liquid moved into the waste heat boiler through the injection tube from the bottom of the boiler, and a “dead zone” formed at the bottom of the boiler such as A and B areas. As time passed, the concentration of alkaline substances in the “dead zone” gradually increased. Also, the working temperature of the front tube sheet was much higher than that of the back tube sheet, and thus the heat load of the front tube sheet was greater. The combination of the above factors is the reason why the leakage always occurred at the bottom of the front tube sheet.

The installation position of #1 and #2 boilers affected the flowrate of the high-temperature flue gas into the boilers. As shown in Fig. 35, #1 and #2 boilers were installed in the vertical direction of the sulfur-burning furnace, and most of the flue gas entered into the #2 boiler at the end of the sulfur-burning furnace, and it was difficult for the high-temperature flue gas to enter into the #1 boiler as the flow direction was vertical to the axis of the #1 boiler. Therefore, the heat load of the #2 boiler was much greater than that of the #1 boiler, and the #2 boiler was more likely to fail when the #1 boiler and the #2 boiler simultaneously ran. Only when the #2 boiler was blocked, did the #1 boiler have a leakage accident.

Based on the above analyses, the following conclusions were obtained:

1. The alkali stress corrosion caused the cracking and the leakage of the tube-sheet joint of the front tube sheet of the waste heat boiler.

2. The stress corrosion cracking led to the leakage. The water and the high-temperature flue gas formed sulfuric acid which moved into the tubes, resulting in the strong corrosion on the tube wall, which was the main cause of the tube burst.

3. The heat load of the #2 boiler was much greater than that of the #1 boiler. Na accumulated at the bottom of the boiler. The temperature of the front tube sheet was much higher than the temperature of the back tube sheet. High stress appeared in the outermost circle of tubes. Therefore, the outermost circle of tubes at the bottom of the front tube sheet of the #2 boiler always failed.

Thanks to Dr. Edward C. Mignot, Shandong University, for your linguistic advice.

• National Natural Science Foundation of China (Grant Nos. 51705265 and 12075127; Funder ID: 10.13039/501100001809).

• Science Education-Industry Integration Pilot Project Plan (Basic Research) (Grant No. 2022PY059).

• Industry-University-Research Collaborative Innovation Fund Project (Grant No. 2022CXY-02).

• Weihai Science and Technology Development Plan Project (Grant No. 2022KC02; Funder ID: 10.13039/501100013352).

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The datasets generated and supporting the findings of this article are obtainable from the corresponding authors upon reasonable request.