Engineering Failure Analysis 18 (2011) 22012210

Contents lists available at SciVerse ScienceDirect

Engineering Failure Analysis

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

Fast identication of O2 corrosion in economiser tubes

HockChye Qua, ChwinChieh Khaw , ChingSeong Tan, Xin Wang, JongBoon Ooi
Faculty of Engineering and Science, Universiti Tunku Abdul Rahman, Jalan Genting Kelang, 53300 Setapak, Kuala Lumpur, Malaysia

a r t i c l e

i n f o

a b s t r a c t
Failure investigation was conducted on three tube samples via visual/metallographic examination and analysis of operational history. The investigation employed a fast-identication approach, which allowed a shorter down time and faster production recovery after tube failure. Visual and metallographic evidence showed that the tube samples had suffered from abnormally high wastage at the external and internal surfaces; the latter surface, in addition had suffered from deep pitting. Analysis of operational history conrmed the presence of dissolved O2 in the feedwater. It was concluded that leakage was primarily caused by tube thinning from internal wastage/pitting, aggravated by external wastage. 2011 Elsevier Ltd. All rights reserved.

Article history: Received 9 June 2011 Accepted 20 July 2011 Available online 27 July 2011 Keywords: Boiler failure Economiser tube Metallography Pitting corrosion

1. Introduction Boilers are used to heat water to generate steam for a variety of purposes. Boiler components such as economisers, waterwall tubes and superheaters are today mainly made of steels and high temperature alloys. An economiser is a heat exchanger located in the lower gas temperature region (450600 C), designed to recover some of the heat from the discharged ue gas. It consists of a series of tubes through which feed water ows to the drum or to the inlet headers of furnace walls. Flue gases ow over the outside of the tubes. Economisers can be constructed with cast iron or steel tubes; the former is usually used in low-pressure industrial boilers (pressure 6 2.5 MPa) and the latter, in high pressure boilers [1]. Failure of boiler tubes is a very common phenomenon, therefore it is important to investigate their root causes to prevent future occurrences [2,3]. Lpez-Lpez et al. [4] investigated the association between the carburisation of austenitic stainless steels and high reside corrosion rates. The methodology encompassed microhardness proling, electron probe microanalysis and metallographic studies of the external surfaces of superheater and reheater tubes. Chaudhuri [5] described case studies related to failures of a 1.25Cr0.5Mo reheater tube, a carbon steel tube and some 2.25Cr1Mo nal superheater tubes. Lee et al. [6] presented failure analysis cases on nal superheater tubes in an ultra-supercritical coal power plant. Ahmad et al. [7] investigated the failure of a SA210-A1, rear water-wall tube by visual site inspection, tube wall thickness measurements and metallographic examination. Saha et al. [8] examined the probable causes of failure of a welded joint in a 210 MW thermal power plant. Dhua [9] investigated failed water-wall tubes from a thermal power station via visual examination, metallographic examination, and electron probe microanalysis. The reasons for boiler tube failures are varied, and common mechanisms include pitting, erosion, thermal fatigue, creep and stress corrosion cracking [2,3,10,11]; other more complex ones include stress assisted corrosion, caustic gouging and acid phosphate corrosion. Some of the failures, for example pitting, are caused by dissolved oxygen in the feed water. In boilers, water coming out of deaerators usually has some residual oxygen, which is removed chemically. Sodium sulphite and hydrazine, or some other chemicals, are used for removal of residual oxygen; sodium sulphite is used for low pressure boilers.
Corresponding author. Tel.: +60 341079802; fax: +60 341079803.
E-mail address: khawcc@utar.edu.my (C.C. Khaw). 1350-6307/$ - see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.engfailanal.2011.07.014

2202

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

Boiler tube failure investigations often involve the use of advanced diagnostic equipment such as SEM/EDX and XRD, which are costly and time consuming to perform [4,12]. However, with a good understanding of the operational features of the plant, and of boiler tube failure mechanisms, a fast and accurate analysis may be made with only minimal usage of expensive laboratory equipment. This paper presented a case where a fast, low cost investigation was successfully conducted on O2 corrosion of economiser tubes, many of which were reported to have leaked. Speed was of the main essence, as decisions had to be made regarding the extent of repairs necessary, to allow the boiler to come on-stream again in the shortest possible time. In the present work, three tube samples from leaking economiser tubes were submitted for investigation; they were designated as Tube A, Tube B and Tube C (Fig. 1). The tubes came from a biomass boiler producing steam for processing palm oil fruit and for power generation. 2. Materials and methods The original outer diameter (OD) was given to be 38.1 mm, and the thickness, 4.0 mm. The outer diameters of the supplied tubes were rst measured at the Locations 15 of Fig. 1. Longitudinal and transverse specimens were then cut out from these locations, mounted and prepared for metallographic examination. Wall thicknesses were measured from the mounted specimens. The boiler operating log for a period of about one year preceding the rst detected tube leakage was summarised and tabulated for easy comparison. The main parameters looked at were sulphite content, total Fe content, and appearance of the water. Where O2 content was analysed, the results were noted. Finally, SEM/EDX analysis was performed on the internal and external deposits of Tube A to conrm the accuracy of the fast investigation. 3. Results and discussion 3.1. Laboratory examination 3.1.1. Visual examination None of the supplied tubes contained any leaking areas or visible bulging; their external surfaces were rough, lightly pitted all over and contained light brownish oxides (Figs. 1 and 2). Their internal surfaces were even rougher and contained deeper pits with thicker brownish oxides (Figs. 37), except at areas where remnants of darkish oxides were present. Figs. 6 and 7 show one end of Tube A containing a butt-welded joint; the internal surface at the vicinity of the joint had deeper and more isolated pits than at other locations. A deep discontinuity was present over a part of the weld root, likely caused by lack of root penetration. 3.1.2. Tube wall measurements The average OD and maximum/minimum wall thicknesses are shown in Table 1; the wall thicknesses were measured from the sections shown in Fig. 8. These values must not be taken to be absolute; they are only values measured at the particular micro-sections and it is very likely that other locations may have more extreme values. However, results generally show that there had been wastage on the outer surfaces; for an original OD of 38.1 mm, the maximum external thickness loss detected would be 0.85 mm, at Location 1. The rest of the metal loss would have been due to wastage and pitting at the internal surface; this amounted to 1.15 mm at pitted areas, leaving a remaining thickness of 2.0 mm, which was only 50% of the original thickness.

Tube B

Tube A

Tube C

Fig. 1. The three tube samples supplied for examination, designated as Tube A, Tube B and Tube C; arrows 15 point to Locations from which Specimens 1 5 were cut out for metallographic examination.

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

2203

Fig. 2. Tube C, external surface, uniformly pitted all over; pits were however much lighter than at the internal surface.

Fig. 3. Tube C, internal surface, uniformly populated with shallow pits; the pits were covered with a light layer of brownish corrosion products. Non-pitted areas (arrowed) had a light layer of darkish oxides.

Fig. 4. Tube B, Location 3, with shallow corrosion pits at the internal surface, covered with brownish oxides. Remnants of darkish oxides were present on non-attacked areas.

3.2. Metallographic examination 3.2.1. General microstructure The microstructures typically consisted of about 10% of pearlite in a matrix of ferrite, corresponding to mild steel with a carbon content of about 0.08% (Fig. 9). The pearlite did not have any signs of thermal degradation, showing that the tubes had not been heated to beyond 450 C or so.

2204

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

Fig. 5. Tube B, Location 4; same comments as in Fig. 4.

Fig. 6. Tube A, cut-off end from Location 1, containing a welded joint; the internal surface was badly pitted at locations where the darkish oxide layer was not present. Non-attacked areas were still covered with a layer of darkish oxides.

Fig. 7. Same part as in Fig. 6, but viewed from another angle; a discontinuity is seen at the weld root, likely caused by lack of root penetration.

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210 Table 1 Outer diameter and thickness of tube samples supplied. Tube Tube A Tube B Tube C Location (see Fig. 1) 1 2 3 4 5 Average OD (mm) 36.4 36.6 36.9 37.0 Not measured Thickness range (mm) 2.13.4 2.02.9 2.53.5 2.03.3 2.03.4

2205

Weld joint S1

S5

S4 S2 S3

Fig. 8. Longitudinal and transverse micro-specimens, S1S5, from the Locations 1 to 5 respectively (see Fig. 1); all specimens have shallow pits at their internal surfaces. The longitudinal S1 specimen contains a weld joint with a deep discontinuity.

Fig. 9. Typical microstructure of the tube material, consisting of about 10% of pearlite in a matrix of ferrite; no thermal degradation was observed in the pearlite (Mag 1000).

Fig. 10. Specimen S1, transverse section, internal surface; view of one of the larger pits with sides having a fairly gentle slope (Mag 20).

2206

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

3.2.2. Internal and external pitting Figs. 10 and 11 show a transverse cross-section of one of the larger internal pits in Specimen S1 (Fig. 1). It was seen that the sides of the pit had a fairly gentle slope and that there was only a light layer of oxides on the surface of the pit. Many internal pits from the other specimens showed the same characteristics, though some others were seen to have fairly sharp, jagged sides, also lled with oxides (Fig. 12). 3.2.3. Weld joint Figs. 13 and 14 show a longitudinal section across the discontinuity at the weld joint of Specimen S1; the remaining ligament was measured to be only about 0.7 mm. The presence of different microstructures at opposite sides of the discontinuity shows that the discontinuity was not a crack, but was due to a lack of root penetration. 3.3. Analysis of operational history (Table 2) 3.3.1. Water appearance From logbook records, it was deduced that water appearance was blackish for a period of nearly 4 months, from 08-Nov2003 to 04-Mar-2004. The black colouration was attributed to welding slag dissolving but it was more likely due to the presence of ne dislodged magnetite (Fe3O4), due to corrosion processes. 3.3.2. Dissolved oxygen analysis Dissolved oxygen was detected to be 8 ppm with an O2 probe meter on 28-May-2004.This value is high and is an indication of some inadequacy in the oxygen removal processes at that time. An indirect indication of the presence of high oxygen over long periods was the requirement of high sulphite injection levels on 17-Apr-2004, 21-May-2004 and 28May-2004.

Fig. 11. Magnied view of the boxed area from Fig. 10, showing that only a light layer of oxides was present at the surface (arrowed) (Mag 50).

Oxides

Fig. 12. Specimen S1, showing the presence of other pits with sharp, jagged sides and lled with oxides (Mag 200).

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

2207

Fig. 13. Specimen S1, longitudinal section across the discontinuity at the weld joint, showing that there was only a thin ligament of about 0.7 mm present (Mag 25).

Fig. 14. Magnied view of the tip of the discontinuity, showing that the microstructures at opposite sides of the discontinuity were different; this shows that the discontinuity was not a crack (Mag 100).

3.3.3. Fe content in boiler water The Fe content had been signicantly above the control value of <5 ppm for a period of time, from 08-Nov-2003 to 11-Feb-2004. The high Fe value during this period is a strong indication that a fair amount of corrosion had been occurring at the ferrous components in contact with the boiler water.

3.3.4. Deaerator malfunction There was evidence that the deaerator was not able to operate as designed due to the incoming feedwater not being able to attain the minimum required temperature of 50 C from 13-Oct-2003 (when the boiler was rst commissioned) to 15-Oct2004. The reason for the temperature requirement is that the solubility of O2 in water increases with lower temperatures, and if the water contained too much O2, excessive amounts would still remain after deaeration. This would mean that the water entering the economiser would still contain excessive O2.

3.4. Analysis of evidence 3.4.1. Operational history From Section 3.3, it is clear that there had been corrosion in the ferrous components of the system, and that this corrosion had been due to high dissolved oxygen. The high oxygen was mainly a result of the inability of the deaerator to function as designed, due to insufcient temperature of the inlet water. Such oxygen would rst affect the economiser tubes, which are located directly after the deaerator; unfortunately for the operator, most, if not all of these tubes would have been affected.

2208

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

Table 2 Summary of important operational information from 08-Nov-03 to 15-Jan-05. Date 08-Nov-03 Sulphite (ppm SO3) Total iron (ppm Fe) 35 17 Control limits (ppm) 2040 <5 Comments Sulphite was within limits. Total Fe was high; water appearance was reported to be blackish; attributed to welding slag dissolving. Dissolved Fe was high. Sulphite was within limits. Total Fe was high. Sulphite was within limits. Total Fe was high. Sulphite was within limits. Total Fe was high. Sulphite was within limits. Total Fe was high; attributed to welding slag. Sulphite was low; attributed to pressure relief valve (PRV) malfunction. Fe was within limits; it was commented that water was black on the previous visit (on 11-02-04). Sulphite was low; recommended to increase Scalrid 66 to remove O2. Deaerator was said not to be effective since PRV diaphragm was missing (removed for repairs). Fe had become high again. Scalrid 66 was recommended to be increased because of deaerator malfunction and consistently low SO3. Sulphite had now become high. Fe was within limits; water appearance was clear. Sulphite was low again; dissolved O2 was detected to be 8 ppm with an O2 probe meter. Scalrid 66 was further increased. Fe was within limits.

10-Nov-03 17-Dec-03 30-Dec-03 06-Jan-04 13-Jan-04 11-Feb-04 19-Feb-04 04-March-04

Dissolved Fe was at 11.2 ppm. 26 2040 22 <5 Inspection by DOSH (boiler regulatory body) 22 2040 15 <5 23 2040 15 <5 24 2040 33 <5 Partial handover to Client 12 2040 1.5 <5 2040

17-April-04

13

21-May-04 27-May-04 28-May-04

18 No results given 55 2.3 13

<5

2040 <5 2040

10-June-04 12-June-04

23-July-04 27-Aug-04 17-Sept-04 2022-Sept-04 14-Oct-04 25-Oct-04 04-Nov-04 17-Dec-04 15-Jan-05

2.0 <5 Email from Operator to Consultant, asking for advice regarding the claim from the deaerator designer that proper operation of deaerator requires feed water temperature to be 500 C, minimum, which the system cannot fulll. 75 2040 Sulphite had now become high again. 10 <5 Fe was rising again. Fe in feedwater = 0.3 ppm. Fe in storage tank = 0.35 ppm. Fe in raw water = 2.3 ppm. 65 2040 Sulphite was high. 1.1 <5 Fe was within limits; water was slightly brownish. 12 2040 Sulphite was low again. 1.2 <5 Fe was within limits. 65 2040 Sulphite was high. 1.3 <5 Fe was within limits. Leaking was detected at some tubes; a pipe was tted to bypass the economiser. 15 2040 Sulphite was low. 0.6 <5 Fe was within limits. 70 2040 Sulphite was high. 0.8 <5 Fe was within limits. 35 3050 Sulphite was within limits. 0.5 <5 Fe was within limits. 12 2040 Sulphite was low. No reading <5 20 2040 Sulphite was within limits. 0.5 <5 Fe was within limits. Water appearance was clear.

3.4.2. Nature of external wastage The characteristics of the wastage are typical of that due to acid condensation attack on cold surfaces due to frequent stoppages. A fuel additive was added, which was claimed to be able to reduce such attacks but it did not seem to have worked. The external wastage was not the major cause of the present tube thinning, though the maximum value of 0.85 mm loss for the 11 months of operation is abnormally high. In the absence of internal wastage, tube thickness would reduce to 50% of its original value within 2 years of service.

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210 Table 3 SEM/EDX analysis. No Element Composition (wt%) External deposits 1 2 3 4 5 6 7 8 O2 Si P S K Ca Mn Fe 19.14 1.52 0.30 0.31 0.24 0.60 0.62 77.28 Internal deposits 19.00 3.21 Not detected Not detected Not detected Not detected Not detected 77.79

2209

3.4.3. Internal pitting The pits were lled with brownish oxides that only occurred at locations where the darkish protective magnetite (Fe3O4) were not present; these are typical characteristics of oxygen attack. Oxygen dissolved in the feed water in contact with a clean Fe surface will attack the surface by an electrochemical reaction, where the Fe is oxidised at the anode and released into the water. The electrons from oxidation are released and absorbed by O2 at the cathode, to eventually form Fe(OH)2. In the absence of free reactive oxygen and at elevated temperatures, the Fe(OH)2 is converted to a dense, protective layer of magnetite (Fe3O4), darkish in colour; this layer keeps the corrosion rates of boiler tubes to very low levels. However, in the presence of free oxygen, the protective magnetite will oxidise to the loose, porous and less protective Fe2O3. When breaks occur at the magnetite layer due to this mechanism (or to other mechanisms), the freshly exposed metal would become anodes in an electrochemical cell, with small anodic sites in conjunction with large cathodic areas; such a situation would result in pitting attack of the anode [13,14]. The presence of high oxygen has been directly conrmed by dissolved oxygen analysis on a particular day and indirectly conrmed by the frequent requirement of increased sulphite injection into the feed water. The signicant presence of Fe ions in the water has been directly conrmed. The wastage and pitting at the internal surfaces of the tubes have been the major causes of thinning and tube leakage. However, the welding was of poor quality and it is possible that in some of the leaks at least, a signicant contribution was from a lack of root penetration (Section 3.2.3). 3.5. SEM/EDX analysis The compositions of the internal and external deposits of Tube A, from SEM/EDX analysis, are presented in Table 3. The internal deposits were oxides of Fe and Si; the oxides of Fe were products of corrosion and is expected; the oxides of Si were from silica dissolved in the feedwater and not properly removed, also expected. The external deposits came from corrosion/ oxidation products of Fe as well as from combustion byproducts of the biomass fuel, all expected. It is seen that the EDX results did not reveal any damage mechanism other than those identied in Section 3.4 above. 4. Conclusions The three tube samples supplied were found to have suffered from abnormally high wastage at both the external and internal surfaces; the latter surfaces were more badly affected and in addition suffered from deep pitting. Wastage at the external surfaces had been due to acid condensation attack of the cold surfaces during stoppages. Wastage and pitting at the internal surfaces had been due to oxygen attack. The presence of oxygen in the feedwater had very likely been due to improper function of the deaerator over long periods. Internal wastage/pitting and external wastage, leading to excessive tube thinning, were deemed to have been the major causes of the leakages, though in some cases poor welding could have contributed. SEM/EDX analysis did not reveal any damage mechanism other than the above. Acknowledgements The authors gratefully acknowledge the support from Centre for Vehicular Technology, Universiti Tunku Abdul Rahman and Profmetal Sdn. Bhd. References
[1] [2] [3] [4] [5] Kakac S. Boilers evaporators & condensers. 1st ed. New York: Wiley-Interscience; 1991. Barer RD, Peters BF. Why metals fail. 1st ed. New York: Gordon and Breach Science Publishers; 1970. Becker WD, Shipley RJ. ASM handbook, failure analysis and prevention, 10th ed., vol. 11. ASM international; 2002. Lpez-Lpez D, Wong-Moreno A, Martinez L. Carburisation processes involved in boiler-tube failures. Corros Sci 1993;35:11515. Chaudhuri S. Some aspects of metallurgical assessment of boiler tubes basic principles and case studies. Mater Sci Eng A 2006;432:909.

2210

H.C. Qua et al. / Engineering Failure Analysis 18 (2011) 22012210

[6] Lee NH, Kim S, Choe BH, Yoon KB, Kwon D. Failure analysis of a boiler tube in USC coal power plant. Eng Fail Anal 2009;16:20315. [7] Ahmad J, Purbolaksono J, Beng LC, Rashid AZ, Khinani A, Ali A. Failure investigation on rear water-wall tube of boiler. Eng Fail Anal 2009;16:232532. [8] Saha A, Roy H, Shukla AK. Investigation into the probable cause of failure of economiser tube of a thermal power plant. J Fail Anal Preven 2010;10:18790. [9] Dhua SK. Metallurgical investigation of failed boiler water-wall tubes received from a thermal power station. Eng Fail Anal 2010;17:15729. [10] Spurr JC. Signicance of copper deposits associated with a boiler tube failure. Anti-Corros Method M 1959;6(8):2337. [11] Hutchings FR, Unterweiser PM. Failure analysis: the British Engine technical reports. Metals Park (OH): American Society for Metals; 1981. [12] Szab A, Varga K, Nmeth Z, Rad K, Oravetz D, Mak K, et al. Effect of a chemical decontamination procedure on the corrosion state of the heat exchanger tubes of steam generators. Corros Sci 2006;48:272749. [13] Cramer SD. ASM handbook, corrosion: materials, 10th ed., vol. 13B. ASM international; 2005. [14] Trethewey KR, Chamberlain J. Corrosion for science and engineering. 2nd ed. UK: Longman; 1995.