Materials and Design 31 (2010) 4258–4268

Contents lists available at ScienceDirect

Materials and Design

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

Failure analysis of bursting on the inner pipe of a jacketed pipe in a tubularheat exchanger

Yi Gong, Jing Zhong, Zhen-Guo Yang *

Department of Materials Science, Fudan University, Shanghai 200433, China

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

Article history:Received 10 November 2009Accepted 3 April 2010Available online 13 April 2010

Keywords:Failure analysisCorrosionMetal matrix

0261-3069/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.matdes.2010.04.010

* Corresponding author. Tel.: +86 21 65642523; faxE-mail address: zgyang@fudan.edu.cn (Z.-G. Yang)

In order to identify the causes of a bursting incident that occurred on the inner pipe of a jacketed pipe in atubular heat exchanger for synthesis of high pressure polyethylene, series of characterization analysiswere conducted. Metallurgical structure and chemical composition of the pipe’s metal matrix wereinspected by metallographic microscope (MM) and photoelectric direct reading spectrometer; scanningelectron microscope (SEM) and energy dispersive spectroscope (EDS) were applied to observe the micro-scopic morphology and micro-area composition on the ruptured surface; compositions of the coolant, i.e.the circulating cooling water were examined by inductively coupled plasma-atomic emission spectrom-etry (ICP-AES), ion chromatography (IC) and Fourier transform infrared spectroscopy (FTIR). In addition tothese, finite element analysis (FEA) was employed to study the erosive effect of on the pipe. Analysisresults revealed that interaction between corrosion and erosion both led by scaling, was the main causethat accelerated its thinning and eventually resulted in its premature failure. Finally countermeasuresand suggestions were proposed.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Heat exchangers with diverse types including shell and tube,plate, regenerative, adiabatic wheel, and plate fin, etc. are thefamiliar thermal equipment in a wide range of applications [1].Among them, the tubular heat exchanger, which generally refersto the shell and tube type, is the most widely used in industry[2]. In terms of the reasons for its popularity, advantages like largeratio of heat transfer area, excellent versatility, and convenientcleaning method may be ascribed to [3]. Furthermore, tubular heatexchanger particularly exhibits superior reliability in high temper-ature and high pressure applications as the heat is not directlytransferred by working fluid inside the heat tubes [4].

Although tubular heat exchanger is characterized by reliability,safety is still always one of the primarily concerned issues for it inengineering practice. In fact, some failure incidents on their tubeswere truly reported and analyzed in past applications, whereasmost of which were only involved with the equipments operatingunder normal temperatures and pressures and were usually led byfaults generated in design, service, maintenance, and materialsselection. Goyder [5] studied the thinning effect on heat exchangertubes from flow-induced vibration which was resulted from im-proper design and operation. Kuznicka [6] discovered that interac-tion between pitting and erosion both engendered by disqualified

ll rights reserved.

: +86 21 65103056..

cooling water was the main cause resulting in leakage of one heatexchanger tube. According to Ranjbar [7], improper material selec-tion accompanied with scaling that was led by low-velocity circu-lating cooling water containing large size of substances were thetwo main causes of erosion and then thinning on the heater ex-changer tubes. In our past works [8], it was determined that leak-age cases of titanium tubes in some heat exchangers of nuclearpower facility were caused by a variety of inappropriate ways indesign, manufacture and installation of the tubes. Particularly,mechanism of hydrogen blistering on titanium tubes under cir-cumstance of seawater was firstly discovered at home and abroad.

Actually, failure issues of tubular heat exchangers’ tubes underextreme conditions like elevated temperatures and pressures alsohave been reported in the past. However with respect to the failurecauses, they were mainly relevant to the cracking led by thermalfatigue [9–11]. Only in research from Kaul and Muralidharanet al. [12], crevice corrosion caused by expansion of tubes underconditions of 126–163 �C and 16–20 MPa was predominantlyblame for the leakage of tubes in the tubular heat exchangers.

In this paper, a bursting incident on one heat pipe of a tubularheat exchanger in the high pressure polyethylene equipment wascarefully analyzed in a petrochemical complex in Shanghai. Thewhole equipment of the circulating cooling process, i.e. the tubularheat exchanger, is mainly designed for recycling unpolymerizedethylene gas for repolymerization. As illustrated in Fig. 1a, afterpolymerization reaction, mixture of unpolymerized ethylene gasand low molecular weight polyethylene oligomer is separated from

Fig. 1. Schematic diagrams of the tubular heat exchanger (a) circulating cooling process (b) jacketed pipe structure (c) external appearance and (d) conceptual scheme ofjacketed pipe.

Y. Gong et al. / Materials and Design 31 (2010) 4258–4268 4259

polyethylene slurry in separator V202. Then, in order to obtainpure ethylene gas, the mixture is cooled and separated throughthree similar two-step procedures successively, each involving acooler (tubular heat exchanger) and a separator. Actually, the tubu-lar heat exchanger in this case is exactly a peculiar one, which isdesigned with a jacketed pipe structure rather than the conven-tional conformation that tube bundles are encircled by a shell, asshown in Fig. 1b and c. The jacketed pipe structure is composedof two concentric pipes, one inside another [13]. Consequently,the two pipes are respectively named inner pipe and jacket pipe.The inner pipe transports high temperature and high pressure eth-ylene gas, while the jacket pipe conveys industrial circulating cool-ing water in the annular space, seen in Fig. 1d. Concretely in thiscase, the inner pipes were made of low alloy steels with caliberU of 61 � 12 mm, and their operating pressure and temperaturewere 30.0 MPa and 40–250 �C, respectively. The expected life ofthe heat exchanger was about 16 years, but some of their innerpipes in the jacketed pipes failed within just 4 years. Particularly,one of the inner pipes in the cooler namely E302 suddenly burstnear the tee pipe of the cooling water outlet in service and causedleakage of the ethylene gas within it. The detailed burst position ismarked with red rectangular in Fig. 1b as well. Such a severe failureincident not only brought about enormous economic loss, but alsothreatened the life of operative employees seriously. Hence, thor-ough investigations were urgently needed to determine the exactcauses of the failure to prevent similar failure incidents from hap-pening in the future. Thus detailed characterization methodsincluding metal matrix examination, process medium inspection,

and macro and micro morphology observation on the rupture sur-face were conducted. Apart from these, the finite element analysis(FEA) was also employed to calculate the fluid velocity near thebursting part of the inner pipe.

As the tubular heat exchanger with peculiar jacketed pipe struc-ture is seldom applied in engineering practice, this failure analysiscase conducted on the burst inner pipe of the jacketed pipe was apioneering attempt for safety evaluation on tubular heat exchang-ers with this structure. What’s more, failure mechanism as corro-sion–erosion interaction caused by scaling on the inner pipe inthe heat exchanger, which operated under extreme conditions ofhigh temperature and high pressure specific for polyethylene poly-merization, was also firstly reported. Meanwhile, the FEA methodwas adopted in the failure analysis of tubular heat exchangers un-der this service conditions as well. Finally, based on the analysis re-sults that bursting was caused by local thinning of the inner pipe’swall thickness under corrosion–erosion interaction, preventionmethods were suggested, which has an important significance infailure prevention for the similar inner pipes in tubular heatexchangers under extreme conditions in the future.

2. Experimental and results

2.1. Visual observation

Fig. 2a and b presents the serious corrosion morphologies onthe outside wall of the failure inner pipe near the outlet of the cool-ing water, including rufous corrosion products, scaling substances

Fig. 2. External appearances of the failure inner pipe (a), (b) total morphology (c) scaling on outside wall (d) thinning of rupture cross section (e) bursting position and (f)scaling on inside wall.

4260 Y. Gong et al. / Materials and Design 31 (2010) 4258–4268

and concave areas on its surface. As shown in Fig. 2c, some corro-sion pits can be detected as well. Meanwhile, parts of the corrosionproducts had scaled off the pipe surface. Moreover, the wall thick-ness of the inner pipe had been thinned badly, seen in Fig. 2d,which may have been resulted from the interaction between corro-sion and erosion and would result in the strength decrease of thepipes. The minimum thickness of the thinned pipe wall was only2.0 mm, compared with its original value of 12.0 mm. Conse-quently, plastic deformation occurred at the straight segment ofthe thinned inner pipe, and finally led to bursting, seen in Fig. 2e.Take a closer look at the inside wall of the inner pipe (seen inFig. 2f), lots of white ethylene oligomer deposits can also be found.Such a severe deposition and scaling phenomenon may affect theheat transfer efficiency of the cooler. Thus, regular descaling oper-ation and highly efficient detergent are urgently required for thetubular heat exchanger in service.

In order to thoroughly identify the causes of the bursting, sam-ple was cut from the failure inner pipe for further characteriza-tions. The bursting crevasse exhibited an elliptic shape, whichcovered nearly one-third of all the circle area of the pipe, seen inFig. 2e. Fig. 3 shows the three-dimensional schematic diagramsof the bursting position on the pipe, in which sample was cut atthe site A.

2.2. Metal matrix examination

Chemical compositions of the failure inner pipe’s metal matrixare listed in Table 1, which are in accordance with the require-ments of 20g boiler steels in GB713-1997 standard of China (equalsto ASME SA516 Gr.55 [14]). Additionally, the existence of Cu mayrefine grains of the steel, which could bring about the improve-ment of strength and toughness of materials.

Etched in agent of picric acid (2,4,6-trinitrophenol) 1.25 g, HCl20 ml, ethanol 10 ml and H2O 10 ml for 40 s, the metallographicstructures of the metal matrix are displayed in Fig. 4a, whichshows a typical low-carbon steel structure. This microstructureconsists of ferrite and pearlite equiaxed grains. Moreover, a verysmall amount of linear inclusions can be observed in part of themicroscopic area, seen in Fig. 4b.

2.3. SEM and EDS analysis of the rupture

2.3.1. Rupture surfaceFirstly, rupture surface of the inner pipe was observed with

SEM. As is shown in Fig. 5a, a hook-shaped fracture edge can beseen along the rupture. Unevenly distributed black and white cor-rosion agglomerations can be found on the edge surface, and most

Fig. 3. Schematic diagrams of sampling on the failure inner pipe.

Table 1Chemical compositions of inner pipes (wt.%).

Element C Si S P Mn Ni Cr Mo V Cu

Inner pipes 0.136 0.217 0.009 0.014 0.605 0.139 0.136 0.044 0.001 0.228SA 516 Gr.55 [14] 60.20 0.15–0.30 60.035 60.035 0.50–0.90 / / / / /

Fig. 4. Metallographic structures of inner pipe (a) 500� and (b) polished state 100�.

Y. Gong et al. / Materials and Design 31 (2010) 4258–4268 4261

of which appear to be loose and laminated morphologies, seen inFig. 5b. Fig. 5c displays various shapes of corrosion products at-tached on the rupture surface. More detailed information of themcan be found in a further magnified scheme in Fig. 5d, which re-vealed a clear gossypine configuration with formation of crackingwithin it.

Additionally, chemical compositions of the corrosion productswere detected by EDS. The three analysis sites were displayed inFig. 5c as well, and their chemical compositions were listed inFig. 6 and Table 2. According to the analysis results, black corrosion

products (sites 001 and 003) consist of four major elements as Fe,O, S and Cl, which identifies possible corresponding compounds asFeOH, Fe2O3, FeS, FeCl3 and so on. It was such a surprise that thecontents of chlorine element in the black corrosion products wereup to 1.66% and 6.25% (wt.%) respectively in sites 001 and 003.Hence, it can be inferred from the EDS results that corrosion ledby chloride ions may occur on the pipe. The white corrosion prod-ucts (site 002) were composed of three major elements of Fe, O andS, which demonstrates the possible corresponding compounds asiron oxides and sulfides.

Fig. 5. SEM morphology of the rupture surface (a) total morphology (b) magnification of the rupture edge (c), (d) further magnification of corrosion products.

Table 2Chemical compositions of corrosion products on the rupture surface (wt.%).

Element O Cl S Fe

Site 001 4.50 1.66 3.19 90.65Site 002 5.41 / 1.68 92.90Site 003 6.62 6.25 0.96 86.18

Note: ‘‘/” denotes the content lower than 0.5 wt.%.

4262 Y. Gong et al. / Materials and Design 31 (2010) 4258–4268

2.3.2. Outside wall of the ruptureThe outside wall of the rupture was then also analyzed under

SEM and EDS. As is clearly shown in Fig. 7a, corrosion productswith various shapes but all in black and gray color can be detectedon the wall. Moreover, chains of undulating hills, i.e. the typicalsurface morphologies of crevice corrosion can also be inspectedin Fig. 7b. Within a large corrosion concave area (Fig. 7c), corrosionproducts have desquamated to pieces and many small white parti-cles can be found depositing on the smooth bottom of the area,which can be considered as the trace of erosion. What’s more, someloose and porous tiny corrosion products can be seen on the bot-tom of the concave area as well, seen in Fig. 7d.

Meanwhile, chemical compositions of three different corrosionproducts with three different morphologies (marked in Fig. 7a, band c) were detected by EDS, seen in Table 3 and Fig. 8. It can beclearly confirmed that these black and gray corrosion productswere composed of Fe, O, Cl and S elements. Particularly the con-tents of chlorine element were abnormally high, nearly up to1.36%, 3.23% and 1.83% (wt.%) respectively in sites 1, 2 and 3. Theseevidences correspond with the chemical compositions of the corro-sion products covering on the rupture surface detected before.Thus, it can be further inferred that crevice corrosion led by chlo-ride ions that may be contained in the circulating cooling water oc-curred on the inner pipe.

2.4. Inspection of circulating cooling water

Inspections were also conducted on the circulating coolingwater. For improvement of precision, the cooling water was firstly

distilled and condensed from 1000 mL to 50 mL. Finally, straw yel-low transparent solution with a small amount of gossypine depos-its was obtained. The gossypine substances may lead to scaling onthe outside wall of the inner pipes. Compositions of the solutionand the deposits were then successively detected.

2.4.1. ICP-AESICP-AES was firstly used to identify the element compositions

of the solution, the results of which were shown in Table 4. It isdisplayed in Table 4 that concentrations of Ca, Na, K elementsare relatively high, which are always considered to be the maincauses leading to scaling. What’s more, the 2.84 mg/L of Fe ele-ment is accounted for the corrosion substances, which mustraise special attention during the operation management in thefuture.

2.4.2. Ion chromatographIn order to further testify the ion compositions of the cooling

water, ion chromatograph was employed. It is displayed in Fig. 9and Table 5 that concentrations of the chloride ions is 534 ppm,i.e. 26.7 ppm in original water (with a comparison of the Cl� con-centration is limited under 20 ppm in the water supply accordingto its demanded value of design). Thus, the overproof concentra-tion of Cl�may result in serious crevice corrosion and should be ta-ken special care of. With respect to the sources of these largeamounts of Cl�, three paths may be blame for:

I. Directly using the industrial water containing Cl� to serve asthe circulating cooling water.

II. Neutralizing the alkaline circulating cooling water with HCl.III. Cleaning the scaling deposits on outside wall of the inner

pipes with industrial water in downtime.

2.4.3. FTIRThe gossypine deposits obtained from the condensed solution

sample were then detected by FTIR so as to help further confirmthe chemical compositions of the cooling water. As is shown inFig. 10, absorption peaks of AOH, C@O, C@C and SiAO radicalsare detected respectively at wave number of 3406.85 cm�1,

Fig. 6. Chemical compositions of corrosion products on the rupture surface (a) site 001 (b) site 002 and (c) site 003.

Y. Gong et al. / Materials and Design 31 (2010) 4258–4268 4263

1710.24 cm�1, 1622.11 cm�1 and 1116.05 cm�1, from which theexistence of siloxane can be inferred. Meanwhile, the flexuralvibration absorption peaks of CAH are observed at 602.02 cm�1.The FTIR result indicates that the deposits in the circulating coolingwater were a kind of mixture of organic compounds. Particularly,functional groups as AOH, C@C and SiAO, etc. were all active en-ough to adhere on the outside wall of the inner pipe, in otherwords, they were apt to cause scaling.

2.5. Finite element analysis

ANSYS is a universal computational simulation software foranalysis on physical fields like thermal field, force field, magneticfield, etc. and their coupled fields [15–19]. In this case, FLOTRANCFD, which is the subprogram specialized for fluid dynamics anal-ysis in ANSYS, was employed to calculate the fluid velocity near thescaled part on the inner pipe in the annular space of the jacketed

Fig. 7. SEM morphology of the outside wall of the rupture (a) total morphology (b) magnification of corrosion surface (c) concave area (d) porous corrosion products.

Table 3Chemical compositions of corrosion products on outside wall of the rupture (wt.%).

Element O Cl S Fe

Site 1 3.25 1.36 0.25 95.14Site 2 4.39 3.23 0.36 92.03Site 3 3.75 1.83 0.35 94.07

4264 Y. Gong et al. / Materials and Design 31 (2010) 4258–4268

pipe. The FEA result will visually display the erosive effect from thefluid around the scaled part on the pipe.

Fig. 11 displays the FEA model of the jacketed pipe structurewith scaling of 2 mm on the inner pipe. The result of the computa-tional analysis is presented in Fig. 12a, which shows the fluidvelocity distribution around the scaled part on the inner pipe. Itis obvious that density of the high-speed fluid (denoted with redcolor)1 suddenly increases near the scaled part. Quantitatively,maximum velocity of the center fluid is 4.693 m/s. That the high-speed center fluid accumulates around this part inevitably en-hances the erosive effect, which will consequently result in thin-ning on the pipe near this part.

In order to further study the effect of scaling thickness on thefluid velocity, the scaling thickness was then defined as 4 mm.The FEA result is shown in Fig. 12b, from which it can be learnedthat density of the high-speed center fluid before the scaled partis greatly lower than that in Fig. 12a. This means that the densitydifference in Fig. 12b is even larger around the scaled part, whichwill lead to severer erosive effect here. However, the absolute in-crease of maximum fluid velocity is not obvious, only from 4.693to 4.756 m/s. Thus, it can be concluded that the erosive effectmay be proportional to the scaling thickness.

3. Discussion

3.1. Dissolved oxygen corrosion

The circulating cooling water used in this case was acquiredfrom the industrial water, which commonly contained a relatively

1 For interpretation of color in Figs. 12, the reader is referred to the web version ofthis article.

high content of dissolved oxygen in it. Hence, the inner pipe, par-ticularly the outside wall of the inner pipe was apt to be subjectedto the dissolved oxygen corrosion. As the electrode potential of me-tal was lower than that of oxygen, the dissolved oxygen corrosionon the metal in water belonged to electrochemical corrosion, inwhich iron was corroded as anode while oxygen was deoxidizedas cathode. The electrochemical reaction formulas are shownbelow:

anode reaction : 2Fe! 2Fe2þ þ 4e

cathode reaction : O2 þ 2H2Oþ 4e! 4OH�

total reaction formula : 2Feþ O2 þ 2H2O! 2FeðOHÞ2 ð1Þ

Then, the original corrosion products, i.e. the Fe(OH)2 may fur-ther react with water and the dissolved oxygen in it, and would fi-nally be transformed into Fe(OH)3, seen in Eq. (2) [20]:

4FeðOHÞ2 þ 2H2Oþ O2 ! 4FeðOHÞ3 ð2Þ

Actually, Fe(OH)3 would also react with Fe(OH)2 to form ferro-ferric oxide, seen in Eq. (3):

FeðOHÞ2 þ 2FeðOHÞ3 ! Fe3O4 þ 4H2O ð3Þ

Meanwhile, as two unsteady compounds, the Fe(OH)2 andFe(OH)3 decomposed into FeO and Fe2O3 under wet conditionaccording to Eqs. (4) and (5).

FeðOHÞ2 ! FeOþH2O ð4Þ

2FeðOHÞ3 ! Fe2O3 þ 3H2O ð5Þ

Thus, the corrosion products on the outside wall of the innerpipe were mixtures of iron compounds. Among them, the blackones were Fe3O4 and FeO, while the rufous ones were Fe2O3. As amatter of fact, corrosion products as iron compounds were loose,which has been testified under SEM in Figs. 5 and 7. Thus, the ferricions Fe3+ and ferrous ions Fe2+ in them may diffuse outwards. Un-der this condition, fresh corrosion products would be engenderedon the existent corrosion product layer when Fe3+ and Fe2+ contactOH� or oxygen in the water, which then increased the thickness of

Fig. 8. Chemical compositions of corrosion products on outside wall of the rupture (a) site 1 (b) site 2 and (c) site 3.

Y. Gong et al. / Materials and Design 31 (2010) 4258–4268 4265

Fig. 11. FEA model of the scaled part on the inner pipe (a) total geometry and (b)scaled part.

Table 4ICP-AES results of circulating cooling water.

Element Ca Fe K Mg Na Zn P

Concentration (mg/L) 159.4 2.84 40.3 31.1 287.6 7.15 5.88

Fig. 9. Ion chromatograph results of circulating cooling water.

Table 5Ion contents in the circulating cooling water.

Ion Cl� SO2�4

F�

Content (mg/L) 534 331 /

4266 Y. Gong et al. / Materials and Design 31 (2010) 4258–4268

the loose corrosion products layer in turn. Consequently, thicknessof the pipe’s metal matrix became being thinned continuously andcorrosion products layer with increasing thickness was formed in-stead. Thus, it can be now concluded that this mechanism of dis-solved oxygen corrosion was one of the initial reasons forultimate thinning of the inner pipe.

3.2. Crevice corrosion

It was stated that the gossypine substances in the cooling waterwere apt to scale on the outside wall of the inner pipes. But scalingmechanism of this type was not relevant to corrosion but con-cerned with adsorbability of organic groups. Moreover, learned

Fig. 10. FTIR result of gossypine depo

from Table 4, the high contents of Ca, Na, K elements may alsoaggravate the extent of scaling. Consequently, scaling layer gradu-ally became thicker and thicker.

With the generation of the thick scaling layer on the outsidewall of the inner pipe, process that dissolved oxygen contactedthe metal matrix may be slowed down or even obstructed underthe layer. In other words, effect of dissolved oxygen corrosionwas no longer the predominant degradation cause on the innerpipe. Due to this block effect, the oxygen concentration of theupper layer (the scaling layer) would be higher than that of thedeeper layer (metal matrix under the upper layer). Thus the upperlayer became the anode while the deeper one served as the cath-ode. This was exactly the mechanism of crevice corrosion, underwhich condition corrosion grew deeper and deeper and would fi-nally thin the wall thickness of the pipe. Meanwhile, presence ofthe chloride ions Cl� contained in the circulating cooling watermay aggravate localized corrosion for its strong permeability,adsorbability and migratory aptitude [21]. This aggravation mech-anism is expressed as follows:

Mþ þ Cl� !MCl ð6Þ

And MCl is easy to hydrolyze [22]:

sits in circulating cooling water.

Fig. 12. Fluid velocity distribution around the scaled part on the inner pipe (a) 2 mm thick scaling and (b) 4 mm thick scaling.

Y. Gong et al. / Materials and Design 31 (2010) 4258–4268 4267

MClþH2O!MOHþHþCl� ð7Þ

Thus, aggressive hydrochloric acid HCl was generated in a cer-tain small area between the upper layer and the deeper layer,which could accelerate the dissolution of metal in return. What’smore, due to the localizedly auto-catalyst effect of Cl�, the rate ofcrevice corrosion was much faster than that of dissolved oxygencorrosion. Seen in Table 5, the circulating cooling water used in thiscase contained a considerable amount of Cl�, up to 26.7 ppm,which provided a critical supply for the environment of crevicecorrosion. As shown in Fig. 2b and c, different sizes of concaveareas on the outside wall of the inner pipes were just mainlycaused by crevice corrosion.

3.3. Erosion

As shown in Fig. 1b, it is clear that bursting occurred near theposition of the tee pipe, which is always used to change the flowchannel within the jacket pipe. Whereas, presence of tee pipe usu-ally supplies a favorable environment to engender scaling, espe-cially near the turning part of the tee pipe. As was discussedabove, severe scaling had already occurred on the outside wall ofthe inner pipes, hence its extent may be even aggravated nearthe tee pipe. Scaling commonly brings about partial obstructionof the flow channel in the jacket pipe. In order to keep consistencyof the entire flow in the pipe, fluid velocity through the scaled partwill be inevitably increased abruptly, which may result in a severeerosive effect from fluid turbulence on the unscaled but corrodedpart near the scaled part. Sometimes, it will even bring about im-pact flow, which may cause magnificent impact effect on the pipewall as well. As the corrosion products were loose in characteristicshere, they would then be flushed apart from the outside wall of in-ner pipe [23,24]. In fact, trace of erosion was exactly found in thecorrosion products, as shown in Fig. 7c. Subsequently, the freshmetal matrix of the inner pipe without scaling of corrosion prod-ucts was exposed to the corrosive environment, and then engen-dered corrosion products on its surface once again. That wasexactly the periodical interaction between corrosion and erosionon the outside wall of the inner pipe, under which condition thethickness of the inner pipe was thinned continuously.

Such a relationship between scaling and erosive effect has beenverified through the FEA results. It is clearly shown in Fig. 12a thatdensity of the high-speed center fluid was indeed increased aroundthe scaled part, which would then enhance the erosive effect onthe corroded part that was near the scaled part on the pipe. Themechanisms are illustrated below.

If we define Q as the volume flow quantity, u as the fluid veloc-ity, �u as the average fluid velocity and A as the cross section area ofthe pipe, their relationship can be expressed as below:

Q ¼Z Z

Au dA ¼ �uA ð8Þ

Around the scaled part, as the cross section area is reduced, ifthe volume flow quantity remains unchanged, the average fluidvelocity will be consequently increased. In other word, it meansthat density of the high-speed fluid is increased.

Qualitatively, the erosive effect on the pipe can be described asEq. (9) [25]:

W ¼Ww þ kvn ð9Þ

In which, W denotes the total weight loss of erosion, Ww de-notes the weight loss of pure erosion, k is the constant relevantto the angle and type of erosion, n is the material parameter, andv is the velocity of erosion. Actually, within the jacket pipe in thiscase, the v can be redefined as the velocity difference Dv, whichmeans the difference of the velocities between the unscaled andthe scaled parts on the pipe. Thus, relationship between the erosiveeffect and fluid velocity difference can be expressed as Eq. (10):

W ¼Ww þ kðDvÞn ð10Þ

Particularly, the Dv is proportional to the density difference ofthe high-speed fluid. Hence, erosive effect around the scaled partwas especially enhanced. What’s more, according to the FEA re-sults, the erosive effect was also increased with the growth of scal-ing thickness. Thus, severe thinning was finally observed near thepart of scaling.

In summary, conclusions can be put forward that the dissolvedoxygen corrosion initially occurred on the outer side wall of the in-ner pipes and appreciably thinned their thickness. And then thecrevice corrosion that was caused by scaling from gossypine sub-stances and aggravated by chloride ions both contained in the cir-culating cooling water further thinned the inner pipes. Meanwhile,the unscaled but corroded part of the pipe was thinned as well un-der erosion which was caused by scaling. Under the interaction be-tween corrosion and erosion, weight loss ratio of the inner pipeincreased, which finally resulted in thinning of wall thickness ofthe whole pipe. Eventually, the thinned pipe suddenly burst underthe pressure within it when its wall thickness decreased below acritical value.

4. Conclusions

1. Metal matrix of the inner pipe of the jacketed pipe in the tubu-lar heat exchanger was in accordance with the requirements ofSA 516 Gr.55 specifications and it was standard low-carbonalloy steel used for boilers.

2. Dissolved oxygen corrosion and crevice corrosion were the twomain corrosion degradations on the outside wall of the inner

4268 Y. Gong et al. / Materials and Design 31 (2010) 4258–4268

pipe and then led to thinning of the pipe. Particularly, the latterone was brought about by the scaling of gossypine substances,and was accelerated by the overproof chloride ions in the circu-lating cooling water.

3. FEA method verified the aggravated erosive effect on theunscaled part near the scaled part on the inner pipe. And thiserosive effect would be further enhanced with growth of thescaling thickness.

4. Under the interaction between corrosion and erosion on theoutside wall of the inner pipe, the wall thickness was thinned,and eventually resulted in bursting under pressure within it.

5. Recommendations

1. The key factor leading to the excessive content of chloride ionsin the circulating cooling water should be verified and securitymanagement of acidification for neutralization of cooling watershould be enhanced.

2. Effectiveness of the in-use scale inhibitor, dispersant and pre-servative in the circulating cooling water under current opera-tion condition should be carefully studied; meanwhile othereffective agents could also be took into account.

3. Scaling deposits on both sides of the inner pipe must beremoved exhaustively in every cleaning process during down-time in order to enhance heat exchange efficiency and preventcrevice corrosion and erosion.

4. Surface coating protective measure such as nickel and phospho-rus plating layers or high temperature resistant organic coat aso-cresol epoxy resin are recommended to be applied in the newinner pipes in the future.

5. Inspection methods such as resistance probe, potential probe orcomprehensive application of these methods are suggested toimplement real time monitoring and control of the corrosionextent.

Acknowledgements

This work was supported by both Shanghai Petrochemical Co.,Ltd. and Shanghai Leading Academic Discipline Project (ProjectNumber: B113).

References

[1] Kakaç S, Liu HT. Heat exchangers: selection, rating and thermal design. BocaRaton, FL, USA: CRC Press; 2002.

[2] Costa ALH, Queiroz EM. Design optimization of shell-and-tube heatexchangers. Appl Therm Eng 2008;28:1798–805.

[3] Wolverine Tube Inc. (Huntsville, US). Wolverine tube heat transfer data book II.Wolverine Tube Inc., Research and Development Team; 2001. p. 32.

[4] Tu ST, Zhang H, Zhou WW. Corrosion failures of high temperature heat pipes.Eng Fail Anal 1999;6:363–70.

[5] Goyder HGD. Flow-induced vibration in heat exchangers. Chem Eng Res Des2002;80:226–32.

[6] Kuznicka B. Erosion–corrosion of heat exchanger tubes. Eng Fail Anal2009;16(7):2382–7.

[7] Ranjbar K. Effect of flow induced corrosion and erosion on failure of a tubularheat exchanger. Mater Des 2010;31(1):613–9.

[8] Yang ZG. Failure analysis and countermeasure of heat exchangers in nuclearpower facility. Heat Treat Met 2007;32(S1):28–35 [in Chinese with an Englishabstract].

[9] Usman A, Khan AN. Failure analysis of heat exchanger tubes. Eng Fail Anal2008;15:118–28.

[10] Azevedo CRF, Alves GS. Failure analysis of a heat-exchanger serpentine. EngFail Anal 2005;12:193–200.

[11] Otegui JL, Fazzini PG. Failure analysis of tube–tubesheet welds in cracked gasheat exchangers. Eng Fail Anal 2004;11:903–13.

[12] Kaul R, Muralidharan NG, et al. Failure analysis of carbonate reboiler heatexchangers. Eng Fail Anal 1995;2:165–74.

[13] Pieve M. High temperature jacketed piping: review of the suited thermalmodels. Appl Therm Eng 2009;29:431–8.

[14] ASME boiler and pressure vessel code. section II; 1996A.[15] Wang YF, Yang ZG. Finite element analysis of residual thermal stress in

ceramic-lined composite pipe prepared by centrifugal-SHS. Mater Sci Eng A2007;460–461:130–4.

[16] Wang YF, Yang ZG. A coupled finite element and mesh free analysis of erosivewear. Tribol Int 2009;42:373–7.

[17] Wang YF, Yang ZG. Finite element model of erosive wear on ductile and brittlematerials. Wear 2008;265:871–8.

[18] Brooker DC. Numerical modelling of pipeline puncture under excavatorloading. Part I. Development and validation of a finite element materialfailure model for puncture simulation. Int J Press Vess Piping 2003;80:715–25.

[19] Kaptan A, Kisioglu Y. Determination of burst pressures and failure locations ofvehicle LPG cylinders. Int J Press Vess Piping 2007;84:451–9.

[20] Groysman A. Corrosion for everybody. Netherlands: Springer; 2010. p. 24.[21] Gong Y, Cao J, Meng XH, Yang ZG. Pitting corrosion on 316L pipes in

terephthalic acid (TA) dryer. Mater Corros 2009;60:899–908.[22] O’Brien TF, Bommaraju TV, Hine F. Handbook of chlor–alkali technology, vol. I.

In: Fundamentals, New York, USA: Springer; 2005. p. 1338.[23] Hales Crispin, Kelley J, Stevens, et al. Boiler feedwater pipe failure by flow-

assisted chelant corrosion. Eng Fail Anal 2002;9:235–43.[24] Dooley RB, Chexal VK. Flow-accelerated corrosion of pressure vessels in fossil

plants. Int J Press Vess Piping 2000;77:85–90.[25] Yang ZG, Gong Y, Meng XH. Failure analysis and countermeasure of dryers in

PTA facility. Phys Testing Chem Analys Part A: Phys Testing 2009;45(S1):6–16[in Chinese with an English abstract].