Failure analysis of heat exchanger tubes

A. Usman, A. Nusair Khan

*
Metallurgy Division, G.P.O. Box 502, Rawalpindi, Pakistan
Received 23 September 2006; accepted 20 November 2006
Available online 26 January 2007
Abstract
Primary waste heat exchanger tubes of material ASTM A213 grade T11 failed after operation of only three and a half
months. The heat exchanger was of the bayonet type with boiler water inside the tubes and secondary reformer outlet pro-
cess gas at the shell side. The heat exchanger environment was rich in hydrogen, carbon monoxide and nitrogen. The tem-
perature of the process gases was 960 C and the heat exchanger was producing steam at a temperature of 306 C and a
pressure of 1500 psig. The failed, used and new heat exchanger tubes were subjected to stereo/optical microscopy, chemical
analysis and hardness testing. The cause of the failure was thoroughly investigated using optical/scanning electron micro-
scope equipped with energy dispersive spectrometer. The study revealed that the material was exposed to thermal cycling
and excessive local heating. The same was also conrmed by simulated experimentation. These conditions lead to thermal
fatigue of the material with consequent failure.
2006 Elsevier Ltd. All rights reserved.
Keywords: Bulging; Dendrite; High temperature; Failure analysis; Heat exchanger
1. Introduction
Primary waste heat exchanger (C-101) was attached with the ammonia plant. The heat exchanger was of
bayonet type with boiler water in the tube side and secondary reformer outlet process gas was at the shell side.
The chemical composition of the process gas is given in Table 1. The temperature of the process gas was
1760 F (960 C) at the entrance and 740 F (393 C) at the exit. The temperature of the steam in the tubes
was 583 F (306 C). The heat exchanger was producing steam at a high pressure of 1500 psig. In normal oper-
ation, these temperatures and pressures remain the same.
The ammonia plant was revamped in the year 19891991 to enhance its ammonia production capacity by
25%, thus increasing the heat load on C-101 heat exchanger by about 22%. The C-101 was not modied in the
revamp. Before revamp, the maximum service life of C-101 was about 13 years which reduced to 2 years after
the revamp.
1350-6307/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engfailanal.2006.11.051
*
Corresponding author.
E-mail address: nusairy@yahoo.com (A.N. Khan).
www.elsevier.com/locate/engfailanal
Engineering Failure Analysis 15 (2008) 118128
The rst indication of the leakage in the bundle was observed only after 20 days of its induction but the
plant continued to run. After three months the plant faced emergency shut down due to sudden loss of Natural
gas pressure. During this shut down the bundle of tubes was removed for inspection. Circumferential cracks
were observed on ve tubes, whereas three had longitudinal cracks. The location of the tubes is shown in
Fig. 1. The standard tube material is 1.25Cr0.5Mo, ASTM A213, Grade T-11.
Various tests were performed including visual examination, chemical analysis, fractography and metallo-
graphic studies using optical and scanning electron microscopes to determine the cause of failure. Further
a simulated experimentation was also performed to conrm dierent hypotheses.
2. Nomenclature
For convenience, the following nomenclature will be used for the various tubes in this report.
Received tubes Nomenclature
New and un-used tube NT
Used but un-failed tube A-43
Failed tube No. 05 T-05
Failed tube No. A-24 A-24
3. Results
3.1. Visual examination
All the received tubes were observed with the help of a hand magnier and stereomicroscope.
Fig. 1. Sketch showing the position of failed tubes.
Table 1
Chemical composition of the outlet process gas on dry basis
Gases Ar N
2
CH
4
CO CO
2
H
2
Mole% 0.26 22.49 0.44 12.05 8.19 56.57
A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128 119
3.1.1. Tube NT
This tube had no scale or cracks on either of its inner or outer surfaces, Fig. 2. The appearance of the tube
was light brown pertaining to supercial iron oxide. The wall thickness of the tube was consistent throughout
the length. No blackening was observed on any surface of the tube.
3.1.2. Tube A-43
The tube was black in color from both inner and outer sides. The inner surface had a dark black scale with
some portions chipped away. A thin, adherent black scale was also present on the outer side of the tube. No
visible crack or defect was present on the surface of the tube.
3.1.3. Tube T-05
This tube was black on the inner surface, and ash grey at the outer surface, Fig. 3. The length of the tube
had a slight bend. The tube had two types of defects, i.e. ruptured portion with a bluge along the tube axis and
crack like openings across the tube length, Fig. 3. A portion of this tube was cut apart along the tube axis.
Severe scaling was observed on the inner side of the tube; with cracking in the scale, Fig. 4. Moreover, chan-
neling was also observed inside the tube, Fig. 5. A cross-section of this channel is shown in Fig. 5, depicting the
reduction in wall thickness of the tube.
Fig. 2. As received tube-NT.
Fig. 3. Tube-T05 with two types of defects, i.e. (a) cracks across the tube without bulging and (b) ruptured portions with bulging.
120 A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128
Many cracks were observed around the ruptured portion of the tube, Fig. 6. The fracture surface had par-
allel channels like appearance, Fig. 6. The cracks across the tube, Fig. 3, had smoother appearance with no
scaling at the outer side of the tube, Fig. 7a. The same cracks at the inner side of the tube were lled with scale
and blister around their geometry, Fig. 7b and c.
3.1.4. Tube A-24
Unlike tube T-05 this tube had cracks only across the tube axis with no bulging, Fig. 8. There were two
prominent cracks almost parallel to each other on the outer surface of the tube. More cracks were revealed
on the inner surface of the tube when it was cut open, Fig. 9. It was also observed that the cracks were emerg-
ing from the inner surface of the tube as some were present only on the inner surface but were not visible on
the outer surface.
Fig. 4. Internal surface of the tube T-05 with cracks and scale.
Fig. 5. (a) Stereograph of a channeling defect (marked) inside tube T-05, (b) cross section of the same channel.
A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128 121
4. Chemical composition
SEM (scanning electron microscope) equipped with EDS (energy dispersive spectrometer) and C/S analyzer
was used to determine the chemical composition of the material. The chemical composition of the dierent
tubes with its nearest equivalent standard is mentioned in Table 2.
Fig. 6. Inner surface of tube T-05 with a ruptured crack along the tube axis, other cracks are also associated with it.
Fig. 7. Cracks across tube T-05 axis. (a) external surface of the cracks, (b) internal surface of a crack with blister around it and, (c) internal
surface of a crack, lled with scale.
122 A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128
5. Optical microscopy
Samples were cut from the cross section and along the length of the tubes. The samples were studied in the
as-polished and etched condition. All the tubes were found to be seamless. Various other observations on the
tubes are as follows.
5.1. Tube NT
The general structure of the samples after etching revealed ferrite and pearlite. The grain size of the tube
was 15 7 lm. On the outer surface of the tube, a band of ferrite was observed, Fig. 10, but the micro-hard-
ness showed no dierence between the bulk and the outer surface. Cavities/shrinkage like defects were also
observed in the samples, Fig. 11.
Fig. 8. Two parallel cracks were observed on the outer side of the tube (location encircled). No bulging was observed.
Fig. 9. Inner surface of the Tube A-24 with cracks across the tube axis.
Table 2
Chemical composition of dierent tubes along with its equivalent
Elements Composition (wt.%)
NT A-43 A-24 T-05 ASTM A213, Grade T-11
Cr 1.27 1.34 1.25 1.25 1.01.5
Mo 0.64 0.65 0.51 0.55 0.440.65
Si 0.74 0.78 0.65 0.68 0.51.0
Mn 0.52 0.37 0.43 0.56 0.30.6
C 0.088 0.088 0.10 0.102 0.050.15
S 0.003 0.003 0.003 0.001 0.025 max
Fe Balance Balance Balance Balance Balance
A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128 123
5.2. Tube T-05
Several portions of the tubeT-05 were studied, refer to Fig. 3.
5.2.1. Bulged region
In the as-polished condition inclusions were observed in the samples, Fig. 12. The concentration of these
inclusions was up to 1.5%. A thick oxide scale was observed on the inner and outer surface of the tube which
was measured up to 90 10 lm and 130 5 lm, respectively. After etching the samples revealed ferritepearl-
ite structure with spheroidized carbides in pearlite, Fig. 13.
Fig. 10. Ferritic band on the outer surface of the new tube (tube NT).
Fig. 11. Defects in dierent portions of the tube NT.
Fig. 12. Cavities in the bulged area of tube-T05, in the as-polished condition.
124 A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128
5.2.2. Area close to the vertical cracks
Ferritepearlite was observed in etched samples with spheroidized carbides, Fig. 14b, the average grain size
was 16 6 lm. Chromium rich precipitates were also observed on the grain boundaries, Fig. 14a.
5.2.3. Defect free region
The thickness of the scale on the inner and outer regions was 70 10 lm and 60 10 lm, respectively.
Pearliteferrite structure was observed in the etched samples.
5.2.4. Tube A-24
The cross section of the vertical cracks was studied in the as-polished and etched conditions. In the as-pol-
ished condition a number of small oxide-lled cracks were seen to emerge at the tube wall. These cracks ran
from inside towards the outside surface, Fig. 15. In the etched condition, these cracks were found to be trans-
Fig. 13. Bulged region microstructure from outer to inner surface of tube T-05.
Fig. 14. (a) Chromium rich precipitates on the grain boundaries and (b) spheroidized pearlite in the structure.
Fig. 15. Cross section of the vertical cracks in tube A-24.
A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128 125
crystalline in appearance and typical of thermal fatigue [1]. After etching, the samples revealed the presence of
ferrite and pearlite phases, Fig. 16, with spheroidized carbides, Fig. 17. Further, chromium rich precipitates
were also observed on the grain boundaries, Fig. 18.
Fig. 16. Cross section of the vertical cracks after etching.
Fig. 17. Spheroidized carbides (arrows) close to the crack in tube A-24.
Fig. 18. Chromium rich precipitates at the grain boundaries in tube A-24.
126 A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128
6. SEM examination
Cracks across the tubes were opened and studied under the SEM but no information was revealed due to
the presence of oxide scale on the surface of the crack. The porosities in the material contained MgO and CaO
impurities.
7. Conrmation of results
The obtained results were conrmed by designing the experiments at laboratory scale. For this purpose a
healthy pipe was selected and the two ends were welded with a thick ridge plate. The pipe was then thermal
cycled from 850 C to room temperature. An oxy-acetylene torch was utilized as a heating source, whereas a
water bath was utilized for quenching purpose. After 100 thermal cycles, the cross section of the pipe was
examined. The micro ssures observed in the samples (Fig. 19) had a the close resemblance with cracks
observed in the received failed pipes.
8. Discussion
The chemical composition of all the tubes was according to ASTM A213 grade T-11. Two types of defects
were observed in the failed tubes: localized bulging with ruptured crack along the tube axis and cracks across
the tube axis. Shrinkage/porosity cavities, some lled with MgO and CaO, were observed in all the tubes
implying manufacturing defects.
The cracks across the tube axis had features typical of thermal fatigue [13]. This maybe due to repeated
temperature swings causing stresses in the tube wall. The temperature variation maybe due to poor water cir-
culation during the plant operation. The thermal fatigue cracks were also conrmed by simulated experiments
in the laboratory. The presence of spheroidized carbides and chromium rich precipitates may indicate expo-
sure of the failed tube to high temperature, since these features were not found in the new tube (tube NT) and
the tube having satisfactory service life (tube A-43).
The rst failure may have occurred due to thermal fatigue causing cracking in the tubes and consequent
leakage. The presence of only the circumferential cracks with no bulging in tube A-24 supports this argument.
The leakage disrupted the coolant circulation with consequent rise in temperature of the vicinity. As in the
case of tube T-05, both the circumferential cracks and the bulged areas exist in each others vicinity the pres-
ence of spheroidized carbide in the microstructure of the bulged area indicates a probable rise in temperature.
9. Summary
Two types of problems were observed:
cyclic heating and cooling caused thermal fatigue which resulted in circumferential cracks;
exposure to higher than permissible temperature caused bulging with longitudinal cracking.
Fig. 19. (a, b) Cracks observed after simulated experiments in the laboratory.
A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128 127
The probable sequence of events are the following:
1. over a period of time thermal fatigue developed in the tubes;
2. circumferential cracks propagated under conditions of thermal fatigue with consequent leakage of water;
3. insucient cooling due to water leakage caused the tubes to bulge and crack longitudinally.
10. Recommendations
Timely inspection of tubes for thermal fatigue.
Timely replacement of aected tubes.
References
[1] French DN. Metallurgical failures in fossil red boilers. New York: Wiley-Interscience Publication; 1983. p. 103.
[2] Becker WT. Thermal fatigue in a vaporizer handbook of case histories in failure analysis, vol. 2. USA: ASM International; 1996. p.
1118.
[3] Cliord C. Short term failure of carbon steel boiler tubes, hand book of case histories in failure analysis, vol. 2. USA: ASM
International; 1996. p. 2146.
128 A. Usman, A.N. Khan / Engineering Failure Analysis 15 (2008) 118128