Failure Analysis of Stainless Steel Heat

Exchanger Tubes in a Petrochemical Plant

D. N. Adnyana

Journal of Failure Analysis and

Prevention

ISSN 1547-7029
Volume 18
Number 2

J Fail. Anal. and Preven. (2018)

18:413-422
DOI 10.1007/s11668-018-0422-8

1 23
Your article is protected by copyright and
all rights are held exclusively by ASM
International. This e-offprint is for personal
use only and shall not be self-archived in
electronic repositories. If you wish to self-
archive your article, please use the accepted
manuscript version for posting on your own
website. You may further deposit the accepted
manuscript version in any repository,
provided it is only made publicly available 12
months after official publication or later and
provided acknowledgement is given to the
original source of publication and a link is
inserted to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at link.springer.com”.

1 23
 Author's personal copy
J Fail. Anal. and Preven. (2018) 18:413–422
https://doi.org/10.1007/s11668-018-0422-8

TECHNICAL ARTICLE—PEER-REVIEWED

Failure Analysis of Stainless Steel Heat Exchanger Tubes

in a Petrochemical Plant
D. N. Adnyana

Submitted: 9 November 2017 / in revised form: 21 December 2017 / Published online: 15 February 2018
 ASM International 2018

Abstract A shell and tube heat exchanger had failed its also experienced other multiple parallel cracks, but in its
four tubes after just over a year in service. One of these circumferential direction. None of these cracks had reached
tubes was analyzed in this examination. This heat the internal wall of the tube.
exchanger is used to transfer heat from steam on the shell
side to slurry on the tube side. The tube material was made Keywords Heat exchanger

of SA 249 TP 304L, a standard specification for welded Austenitic stainless steel tube

austenitic stainless steel tubes. A failure investigation was Stress-corrosion cracking (SCC)
Corrosion pits

carried out on the leaked tubes by performing a number of Transgranular cracks
Multiple parallel cracks
examinations including visual and macroscopic examina-
tion, chemical analysis, metallographic examination,
hardness testing and scanning electron microscopy equip- Introduction
ped with energy-dispersive spectroscopy analysis. Results
of the failure analysis showed that the leaked tubes of the Failures in heat exchangers are commonly associated with
heat exchanger had primarily experienced stress-corrosion methods of manufacturing of tubing, handling methods
cracking (SCC) in the longitudinal direction caused by the during fabrication, testing methods in the shop and in the
combined effects of tensile stress and corrosion. Most of field and the total environment to which the unit is exposed
the stress-corrosion cracks initiated at multiple sites of the after fabrication, including conditions during shipment,
external surface of the tubes where corrosion pits and other storage, start-up, normal operation and shutdown [1–3]. In
surface defects were present. Formation of surface defects addition, process fluids and cleaning methods applied to the
was most likely associated with some excessive contact heat exchanger may also influence the level of carryover
and/or collision between tube and adjacent tubes or with contaminants that contribute to the corrosion [4].
baffles or other support in the heat exchanger. In the early As in other equipment, stress-corrosion cracking (SCC)
stage, most of the cracks were not branched, but later can also occur in the heat exchanger as a result of corrosion
continued by cracks branching in transgranular manner. and stress at the tip of a growing crack. The stress level
Some of the trace elements found on the fracture surface necessary for SCC is below that needed for fracture with-
deposit may have been responsible for the formation of out corrosion, and the type of stress may be applied or
corrosion pits and SCC on the tubes, acting either alone or residual [4, 5]. Austenitic stainless steel heat exchanger
in combination. In addition, at location close to the mul- equipment can also fail by SCC in acidic chlorides and
tiple parallel cracks in the longitudinal direction, the tube fluorides or by polythionic acids if the microstructure is
sensitized [5, 6]. In the presence of chlorides austenitic
stainless steel, particularly the AISI type 304 stainless
D. N. Adnyana (&) steel, is generally susceptible to pitting corrosion [6]. In
Department of Mechanical Engineering, Faculty of Industrial
addition to chlorides, the SCC in austenitic stainless steel is
Technology, National Institute of Science and Technology
(ISTN), Jakarta, Indonesia also influenced by the presence of any caustic environment
e-mail: [email protected]

123
 Author's personal copy
414 J Fail. Anal. and Preven. (2018) 18:413–422

containing corrosion agents such as sodium (Na) and/or In addition, metallographic examinations were per-
potassium (K), especially at high temperatures [7]. formed on the prepared samples using an optical
This paper presents a failure analysis performed on microscope at various magnifications. The metallographic
leaked tubes of a heat exchanger used in a petrochemical samples were mounted using epoxy and prepared by
plant. This heat exchanger had four tubes which fail in just grinding, polishing and etching. The etchant applied was
over 1 year in service. This heat exchanger was a typical aqua regia solution. A hardness survey was also carried out
shell and tube heat exchanger with a cross/counter flow on the same metallographic samples using the Vickers
arrangement used to transfer heat from steam on the shell hardness method at a load of 5 kg (HV5). Moreover,
side to slurry on the tube side. The heat exchanger tubes examination of some fracture surfaces of the leaked tube
laid out in a triangular 60 pattern and mounted together as samples was performed using scanning electron micro-
a bundle with tube sheets, baffles and tie rods, inside the scopy (SEM) to determine the fracture surface topography
cylindrical shell. The operating parameters of the heat and nature of the failure. This SEM was also equipped with
exchanger were as follows: operating pressure in the shell EDS (energy-dispersive spectroscopy) analysis to detect
side was 49.0 kg/cm2g, while in the tube side was the presence of any corrosion by-product.
112.2 kg/cm2g; operating temperature in the shell side was
261.5 C (inlet) and 261.2 C (outlet), while the operating
temperature in the tube side was 221.2 C (inlet) and Results and Discussion
249.0 C (outlet). The tube dimension was 19.05 mm in
diameter and wall thickness of 2.11 mm, and its material Visual and Macroscopic Examination
was SA 249 TP 304L, a standard specification for welded
austenitic stainless steel tubes. Tube piece A after being tested using dye penetrant clearly
The purpose of this failure analysis was to verify the showed two adjacent sets of cracks (see Fig. 1a). Elongated
material properties and to determine whether the material parallel cracks were observed in the longitudinal direction,
used for the heat exchanger tubes met the specification or and shorter cracks were also observed in the circumferen-
suitable for its operating condition. Furthermore, this fail- tial direction. Among the longitudinal parallel cracks
ure analysis was also aimed to establish the type, cause and observed, one large longitudinal crack appeared as the
mode of failure of the leaked tubes of the heat exchanger, primary crack since through this crack the tube leaking
and based on the determination, some corrective or reme- most likely occurred. In addition to the two sets of cracks,
dial action may be initiated that will prevent similar most of the external surface of tube piece A was severely
failures in future. damaged by bruises, scratches, wear, cuts, dents, etc. (see
Fig. 1b). Formation of these surface defects may have been
associated with collision, rubbing movement and/or other
Examination Details fretting contacts that may have occurred between the tube
and adjacent tubes during the heat exchanger operation.
In this failure analysis, one leaked tube was pulled out of For tube piece B, a similar crack pattern was also
the heat exchanger and cut away into two tube pieces, A observed after being tested using dye penetrant (see
and B each of about 750–800 mm in length for laboratory Fig. 2a). All the cracks in tube piece B were continuations
examination. The tube piece A exhibited multiple parallel of the cracks in tube piece A. Similarly, the cracks in tube
cracks on its surface either in the longitudinal direction B also consisted of elongated cracks in the longitudinal
along the tube or in the circumferential direction around direction of the tube and shorter cracks in the circumfer-
the tube. Tube piece B, which was cut about 2–3 m away ential direction. In addition, from the dye penetrant test on
from tube piece A, did not show any visual surface the tube piece B it was found that at a location approxi-
cracking. Both tube pieces A and B were initially tested mately 1800 from the primary sets of cracks, some
using dye penetrant to identify any crack(s) and subse- intermittent parallel cracks formed in the longitudinal
quently cut into several specimens at different locations for direction of the tube, see Fig. 2b. On this side of tube piece
laboratory examination. Macroscopic examination on the B, there were no circumferential cracks observed. Multiple
fracture surfaces of the leaked tube samples was performed circumferential cracks only formed along the weld seam of
using a stereomicroscope. Chemical analysis on the pre- the tube.
pared samples was carried out using an optical spark Macroscopic examination performed on the external
emission spectrometer. The purpose of this chemical surface around the tube circumference of tube piece A is
analysis was to determine whether the material used for the shown in Fig. 2c, showing the longitudinal tube weld seam
leaked tube met the specification. and the primary longitudinal crack, together with other
adjacent longitudinal cracks. The weld seam was located

123
 Author's personal copy
J Fail. Anal. and Preven. (2018) 18:413–422 415

Fig. 1 Dye penetrant test (a) and surface damage appearance of the tube piece A (b)

Fig. 2 Dye penetrant test on the tube piece B (a) and (b) and macroscopic examination on the external surface of the tube piece A (c)

123
 Author's personal copy
416 J Fail. Anal. and Preven. (2018) 18:413–422

approximately 2 mm from the primary crack. This indi- or other surface defects that formed on the outer surface of
cated that formation of the multiple longitudinal cracks of the tube. It is also seen in Figs. 4 and 5 that after the stress-
the tube was not associated with the welding of the tube. corrosion cracks had reached a certain length, the crack
Adjacent to the primary crack, several parallel cracks subsequently continued in a transgranular manner with
formed over the weld seam in the circumferential direction some branching [4–7]. In the sample obtained from tube
(see Fig. 2c). piece B (see Fig. 5), the microstructures obtained were
The typical appearance of the fracture surfaces of the very much similar with those obtained from samples A.
primary crack of tube piece A is shown in Fig. 3. Most of Although somewhat less extensive, all of these parallel
the fracture surface had a brittle appearance and exhibited cracks were typical of SCC. All of these parallel cracks
crack topography consistent with crack propagation originated from some corrosion pits or surface defects on
through the tube wall from the external surface to the the tube external surface until the crack paths had reached
internal surface of the tube. In addition, oxides and cor- some certain length and subsequently continued by crack
rosion products were observed on most of the fracture branching in a transgranular manner.
surface shown in Fig. 3. In addition to the longitudinal parallel cracks formed
adjacent to the weld seam of the tube, the tube pieces A and
Chemical Analysis B had also experienced other cracks located approximately
180 or at the opposite side from the primary cracks, see
The results of chemical analysis obtained from the material Fig. 6. The topography of these cracks was very much
used for the tube pieces A and B in comparison with the similar to the primary cracks in which the crack(s) were
standard material are given in Table 1. It can be seen that originally formed from corrosion pits at the external
all of the elemental contents of the tube pieces A and B are
within or very close to the material specification of SA 249 Table 1 Results of chemical analysis of the tube material of the heat
TP 304L, a standard specification for welded austenitic exchanger in comparison with the standard material
stainless steel tubes [8, 9]. Composition (wt.%)
Measured value
Metallographic Examination and Analysis
Tube Tube Standard material (SA
A number of specimens were prepared from tube piece A at Element piece A piece B Average 249 TP 304L)
different locations. All the macrostructures and Iron 68.8 69.1 68.95 Balance
microstructures obtained were similar from one sample to Carbon 0.0351 0.0244 0.0298 0.03 (max)
another. The macrostructures shown in Fig. 4a and b Manganese 1.53 1.53 1.53 2.00 (max)
clearly exhibit weld seam structure and one large through Silicon 0.289 0.256 0.273 1.00 (max)
wall crack that had caused leakage in the tube. This large Chromium 19.7 19.6 19.65 18.0–20.0
crack appeared to be the primary crack located at Nickel 8.40 8.35 8.375 8.0–12.0
approximately 2 mm from the weld seam. According to the Phosphorus 0.0289 0.0399 0.0344 0.045 (max)
microstructures shown in Fig. 4a and b, this primary crack Sulfur 0.0058 0.0072 0.0065 0.03 (max)
topography was typical of stress-corrosion cracking (SCC) Molybdenum 0.153 0.155 0.154 …
[1–5]. The appearance of this SCC was brittle, and most of Cobalt 0.175 0.168 0.172 …
the cracks were transgranular. The other parallel longitu-
Copper 0.283 0.287 0.285 …
dinal cracks adjacent to the primary crack also showed
Niobium 0.332 0.323 0.328 …
similar topography typical of SCC. As seen in Figs. 4 and
Vanadium 0.139 0.143 0.141 …
5, most of these parallel cracks initiated from corrosion pits

Fig. 3 Appearance of the fracture surfaces of the leaked tube piece A

123
 Author's personal copy
J Fail. Anal. and Preven. (2018) 18:413–422 417

(a) surface of the tube and then propagated by branched SCC

in a transgranular manner through the austenitic grains of
the tube material.
The tube also experienced multiple parallel cracks in its
circumferential direction (see Figs. 1 and 2). In order to
reveal these circumferential cracks in more clearly, some
samples as shown in Fig. 7 were prepared. Most of the
circumferential cracks observed (see Fig. 7a) were trans-
granular and formed over the weld seam of the tube. All the
cracks initiated from the tube external surface either from
corrosion pits or from other stress concentration. Some of
these stress concentrations likely influenced by formation
of bruises, cuts or scratches on the tube surface. In order to
determine the crack length or crack depth of those cir-
cumferential cracks, another sample shown in Fig. 7b was
also prepared and the resulting microstructures are seen in
Fig. 7b. It is seen that the cracks that formed in the cir-
cumferential direction were also probably typical of stress-
corrosion cracking. However, the crack paths tended to be
bulbous with a number of lobes, and the crack tips them-
selves appeared somewhat blunted and filled with oxides
(b) and/or corrosion product [10]. None of these cracks had
reached the inner wall of the tube. Moreover, none of these
type cracks were observed elsewhere on the other tube
surface, including the opposite side of the tube.

Hardness Test and Analysis

Hardness testing on all specimens at different test locations

obtained from tube pieces A and B was performed using
Vickers hardness method at a load of 5 kg (HV5). From the
results obtained, the average hardness values of the weld
metal and the base metal for each sample are summarized
in Table 2.
All the test results shown in Table 2 clearly exhibit that
the average hardness values obtained from all specimens
were ranging from 206 to 233 HV. This hardness level is
approximately higher than the hardness value of the tube
material according to the standard material of SA 249 TP
304L which is at 90 HRB (max) [8, 9]. This relatively high
hardness value obtained most likely influenced by cold
Fig. 4 (a) Macrostructure and microstructures obtained from some working introduced during fabrication and/or operation of
specimen of tube piece A at location (a), showing one large crack and the heat exchanger. In addition, the relatively high hardness
other cracks propagated in transgranular manner through the
austenitic grains of the tube material. Most of the cracks originated values obtained may have been associated with the pres-
from the tube external surface where some corrosion pits were ence of some dispersed oxides or inclusions and/or other
present. The microstructures also containing several twins and some second-phase particles within the austenitic matrix phase
dispersed oxides or inclusions and/or other second-phase particles. In introduced during the tube manufacturing process [9]. In
addition, no grain boundary thickening due to formation of chromium
carbides or sensitization was observed. Etched with aqua regia general, the tube material having high hardness values may
solution. (b) Macrostructure and microstructures obtained from some have reduced its ductility and could promote the material
specimen of tube piece A at different location from that obtained in susceptible to cracking. Although most of the cracks that
Fig. 4a, showing similar microstructures with that shown in Fig. 4a. had caused the tube failure were due to SCC, the presence
Etched with aqua regia solution
of such imperfections in the tube material may have also
contributed to the acceleration of tube failure in service.

123
 Author's personal copy
418 J Fail. Anal. and Preven. (2018) 18:413–422

Fig. 5 Microstructures obtained from some specimen of tube piece dispersed oxides or inclusions and/or other second-phase particles. In
B, showing several cracks initiated from multiple sites where addition, no grain boundary thickening due to formation of chromium
corrosion pits were present. Most of the cracks propagated in carbides or sensitization was observed. Etched with aqua regia
transgranular manner through the austenitic grains of the tube solution
material. The microstructures also containing several twins and some

SEM Fractography and EDS Analysis surface of the tube. Most of the fracture surfaces covered or
filled by oxides and/or corrosion product. From the results
Samples for SEM fractography and EDS analysis were cut of EDS analysis obtained from the second sample (Fig. 9),
from tube piece A at two different locations. The first most of the trace elements found in the deposits formed
sample was used to examine the fracture surface of the around the outside and inside diameter of fracture surface
primary longitudinal crack of the leaked tube from the included Ca, Na, Mg, S and K.
outside diameter to the inside diameter. The second sample Some of the trace elements found in both samples of
was obtained by opening the primary longitudinal crack to Figs. 8 and 9 may have been responsible for the occurrence
examine its crack surface similar to the first sample. This of SCC on the tubes, acting either alone or in combination.
second sample was located a few centimeters away from These trace elements may have entered into the shell side
the first sample. steam of the heat exchanger due to accidental carryover of
A SEM microfractograph of the through wall fracture dissolved solids or trapped in the heat exchanger after the
surface of the first sample is presented in Fig. 8. Most of tubing was periodically cleaned by chemical solution
the fracture surface topography exhibited a flat and cleav- containing such trace elements. Another possible source of
age appearance, typical of brittle fracture appearance. This the trace elements may have been the use of contaminated
SEM microfractograph further confirms that the leaked water during hydrostatic testing.
tube experienced SCC during its operation. The deposits
present on the fracture surface were analyzed using the
EDS technique, and the results obtained are presented in Conclusions
Fig. 8. The deposits generally contained a number of ele-
ments from which the tube was made such as Fe, Cr, Ni, From the results of the analysis, a number of conclusions
Mn, Si and some Al. The deposits also contained oxygen can be drawn as follows:
and carbon as elements that may be coming from some
1. The material used for tubing of the heat exchanger is
oxides and/or other surface contamination. From the results
very much close to the material specification of SA
of EDS analysis in Fig. 8, most of the trace elements found
249 TP 304L, a standard specification for welded
in the deposit included Na, Ca, Cl, Mg, K and S. Similar
austenitic stainless steel tubes. The microstructures
test results were also obtained from the SEM/EDS analysis
obtained showed austenitic grains containing several
of the second sample (see Fig. 9) showing the crack
twins. In addition, no any grain boundary thickening
topography initiated from the corrosion pits at the outside

123
 Author's personal copy
J Fail. Anal. and Preven. (2018) 18:413–422 419

Fig. 6 Microstructures obtained from some specimens of tube pieces A (a) and B (b) at the opposite side from location where the primary crack
formed. Similar microstructures also obtained from that shown in Figs. 4 and 5. Etched with aqua regia solution

due to formation of chromium carbides or sensitization calcium (Ca), chloride (Cl), magnesium (Mg), sulfur (S)
was observed. and potassium (K). Some of these trace elements may
2. The leaked tubes of the heat exchanger had primarily have been responsible for the occurrence of SCC in the
experienced stress-corrosion cracking (SCC) caused by tube, acting either alone or in combination. The corrosive
the combined effects of tensile stress and corrosion. trace element(s) may have entered into the heat
All the stress-corrosion cracks obviously initiated from exchanger in the shell side steam due to accidental
the external surface of the tube where corrosion pits carryover of dissolved solids or trapped in the heat
and other surface damage were present. The stress- exchanger after the tubing was periodically cleaned by
corrosion cracks had propagated toward the tube chemical solution containing such element(s). Another
internal wall in a transgranular manner with branching possible source for the presence of any corrosive agents
through the austenitic grains of the tube material. In in the heat exchanger may have been from the application
addition to SCC, the leaked tubes of the heat exchanger of contaminated water during the hydrostatic testing.
had also experienced some fretting wear on most of its 4. The primary crack that had caused the tube leaks
external surfaces. Formation of this surface damage clearly revealed large and longitudinal cracks extended
was most likely caused by excessive contact and/or along the length of the tube. At locations adjacent to
collision between the tube and adjacent tubes or with the primary longitudinal crack, other parallel cracks
baffles or other supports in the heat exchanger and may also formed in the longitudinal direction of the tube.
have acted as initiation sites that could promote the All of these longitudinal cracks were apparently
acceleration of SCC failure in the tubing. located outside of the weld seam of the tube, the
3. The trace elements that were found on the fracture nearest being the primary fracture which was approx-
surface deposit of the leakage tube include sodium (Na), imately 2 mm apart from the weld seam.

123
 Author's personal copy
420 J Fail. Anal. and Preven. (2018) 18:413–422

Fig. 7 Microstructures obtained from some specimens of tube piece A that revealed multiple parallel cracks in the circumferential direction: (a)
at the external surface and (b) at the transverse section of the tube

Table 2 Results of hardness test obtained from tube pieces A and B approximately around 180 or at the opposite side
using Vickers hardness method from the primary fracture.
Test location Tube piece A (HV) Tube piece B (HV) 6. In addition to the multiple parallel cracks that extended
in the longitudinal direction along the length of the tube,
Base metal (BM) 206–210 214–218 there were also other multiple parallel cracks observed
Weld metal (WM) 221 225–233 in the circumferential direction at a location close to the
primary crack. These cracks formed around the weld
seam of the tube. However, none of these cracks had
5. In addition to the primary fracture and multiple parallel reached the internal wall of the tube. No such cracks
cracks in the longitudinal direction of the tube, there
were observed elsewhere on the other tube surfaces,
were also other longitudinal cracks, but less extensive,
including on the opposite side of the tube.
observed on the tube external surface located

123
 Author's personal copy
J Fail. Anal. and Preven. (2018) 18:413–422 421

Element Mass Percentage %

C 18.66
O 23.25
Na 3.49
Mg 0.17
Al 0.44
Si 0.68
S 0.28
Cl 1.53
K 0.29
Ca 0.56
Cr 24.58
Mn 0.98
Fe 22.09
Ni 3.00

Fig. 8 SEM microfractograph of the through wall fracture surface of first sample obtained from the tube piece A and the corresponding EDS
spectrum of elements

Element Mass Percentage %

C 6.62
O 26.52
Na 0.36
Mg 0.29
Al 1.18
Si 2.40
S 0.07
Cl 0.02
K 0.12
Ca 3.29
Cr 34.40
Mn 1.28
Fe 20.86
Ni 2.60

Fig. 9 SEM microfractograph of the through wall fracture surface of second sample obtained from the tube piece A and the corresponding EDS
spectrum of elements

123
 Author's personal copy
422 J Fail. Anal. and Preven. (2018) 18:413–422

7. All the tube materials that had been tested exhibited 2. K.E. Perumal, Stress-corrosion cracking of stainless steel heat
relatively higher hardness values compared to the exchanger tubes in a fertilizer plant, published in handbook of
case histories in failure analysis, vol. 1 (ASM International,
allowable maximum hardness value according to the Material Park, 1994), pp. 126–127
tube material specification of SA 249 TP 304L. This 3. Solving Heat Exchanger Problems. http://www.chemengservices.
relatively high hardness of the tube material was most com/exchanger-problems.html (2014)
likely influenced by cold working introduced during 4. D.N. Adnyana, Stress-corrosion cracking in a nickel-base alloy
pre-heater expansion bellows. J. Sci. Technol. Metall. 29(3), 235–
fabrication and/or operation of the heat exchanger and 244 (2014)
could promote the material susceptible to cracking. 5. API RP 571, Chloride Stress Corrosion Cracking, 1st edn.
(American Petroleum Institute, Washington DC, 2003), pp. 4-130
to 4-134
Acknowledgments The author wishes to express his gratitude to the 6. ASM Handbook, Corrosion, vol. 13, 19th edn. (ASM Interna-
Head and Members of Department of Mechanical Engineering, Fac- tional, Materials Park, 1987), pp. 245–282
ulty of Industrial Technology of the National Institute of Science and 7. API RP 571, Caustic Stress Corrosion Cracking (Caustic
Technology (ISTN), for their support and encouragement in pub- Embrittlement), 1st edn, (American Petroleum Institute, Wash-
lishing this work. ington DC, 2003), pp. 4-138 to 4-143
8. Advanced Materials and Processes, Guide to Engineered Mate-
rials (ASM International, Materials Park, 2000), pp. 63–80
9. Metals Handbook, Properties and Selection: Irons, Steels, and
References High-Performance Alloys, vol. 1, 10th edn. (ASM International,
Materials Park, 1990), pp. 841–905
1. ASM Handbook, Failure Analysis and Prevention, vol. 11, 6th 10. API RP 571, Corrosion Fatigue, 1st edn, (American Petroleum
edn. (ASM International, Materials Park, 1998), pp. 628–642 Institute, Washington DC, 2003), pp. 4-135 to 4-137

123