Failure Analysis investigation of Cracking at Weld Joint of LOOP Boiler Feed Water (BFW) Preheater
LOOP BFW preheater was in service for approximately 10 months, the tube side process medium was Synthesis gas (including H2 gas).
The failure was noticed as leakage from circumference weld joint (Tube sheet to head) during regular plant patrolling. This was first time
failure in welds of heat exchanger; however, the plant was in operation till the shutdown was undertaken.
Material of Construction for Tube Sheet is SA-336 F22 CLASS 3 and for Head is SA-387 GR.22 CLASS 2. Circumference weld Joint was
welded using approved WPS with SMAW welding process (SFA 5.5, E9015-B3) from inside and SAW welding process (SFA 5.23, EB3R+
UV420TTRW) from outside.
The details of design parameters are given in Table 1.
PAUT/Manual UT, PT and RT was carried out at site to identify the Quantum of cracks in Circumferential weld joint. And the Defect mapping
was done shown in Figure. 2 & 3.
All other 2.25Cr-1Mo Welds (i.e., weld joint for Nozzles to dished head and Manway to Dished head) were also examined by Manual UT,
MT/PT to verify the soundness of weld (i.e., free from any cracks). Also, in-situ Hardness test was performed on these welds to check the
Hardness.
Figure 1: General drawing indicating weld joint failure location.
Figure 2: Defect location Figure 3: Defect location
mapping from outside mapping from inside
FEA analysis was performed to confirm mechanical integrity of Loop BFW Preheater Exchanger for operating condition piping load on Nozzle A1 & B1.
MANUFACTURERS DATA REVIEW
Manufactures data provided vide vendor was verified for the compliance to the requirements of the Project and following are the observation
for the same.
A. Material Certificate for Dished head, Tube sheet & Welding Consumable were reviewed to check compliance with API 934A for mechanical,
Chemical and temper Embrittlement control. Mechanical Properties, Chemical Properties reported in MTC were found meeting the requirements
of API 934A. Also welding consumables were Tested for Step cooling (temper embrittlement screening) test and were found meeting the API
934A requirements.
B. The Circumferential weld joint was Local PWHT’ed as specified in Drawing. For local PWHT extension of Soak band, Heat Band & Insulation
band were found not meeting the requirements addressed in ASME sec VIII Div 2 Ed. 2017, Clause No. 6.4.3.1 and WRC 452 referred therein.
Nevertheless, PWHT chart was verified for Circumferential weld joint and PWHT cycle reported were in line with requirement. Also, Hardness
result after PWHT for Circumferential weld joint were reported within the limits of the API 934A requirements.
C. Absence of pre-existing Manufacturing defect could be verified by thorough review of NDE reports only. Vendor to provide RT film for Review
but same was not available for verification.
ANALYSIS
To find the root cause of failure, boat sample was removed from the crack location. A sample from the OD side weld crack (location 45o to 90o)
was cut. And a crack sample from ID side of the weld was removed from approachable cracked location of ID Weld (near to location 270o to 315o).
These boat samples were sent to Metallurgical Lab for further analysis. In addition, Insite in-situ Metallographic examination was performed to
identify the Microstructure and the crack morphology.
The investigation was carried out based on the background data, history of failure with available photographic evidence, visual examination,
wet fluorescent magnetic particle inspection (WFMPI), low magnification examination, chemical analysis, scanning electron microscopy (SEM),
microstructural examination, tensile test and hardness measurement Based on the investigative findings the root cause of the problem has been
identified.
Details of the test / Examination that were performed vide insitu
Metallographic examination & Lab testing are as Follows:
IN-SITU METALLOGRAPHY AT SITE
Crack with dished end side PM from Inside
Figure: 4 Weld (100X) Figure: 5 HAZ (200X)
Crack with Tube sheet side PM from Outside
Figure: 6 Weld (100X) Figure: 7 HAZ/PM (500X)Inference: Inter dendritic cracks, branching in nature and filled with scale are observed at weld & HAZ region from outside and inside.
INVESTIGATION STEPS – INNER WELD:
A. Visual Examination:
Visual examination was carried out on sample received for investigation along with photographic evidence as under:
Figure: 8 Shows boat sample of inner weld in as-received condition.
Crack is observed in the marked region. The defect location mapping
show’s location from where boat sample was removed.
Figure: 9 Sectional view of the inner weld sample. Crack
is seen on weld as seen in the marked region.
B. Wet Fluorescent Magnetic Particle Inspection (WFMPI):
To find out any surface defects/auxiliary cracking, WFMPI test was carried out on the sample.
Equipment Used: Controls and equipment make Yoke Type Electromagnetic Crack detector. Current Used: AC
Reference Standard: As per ASTM E 709.
Figure: 10 Shows inner weld sample in WFMPI tested condition.
Linear indications are seen.
Figure: 11 Sectional view of the sample in WFMPI tested condition.
Linear indication is seen in marked region extending in the cross-section.
C. Low Magnification Examination: -
Low magnification examination was carried out on crack surface to reveal more details about failure mechanism.
Figure: 12 Crack surface low magnification view showing brittle
nature of crack. 10X
D. Chemical Analysis: -
Table 2 Results of chemical analysis obtained through optical emission spectroscopy
Inference: Results show that the material meets with the requirements as per the specification of ASME II Part C SFA 5.5 E9015 B3 & API 934A.
*Required as per API 934 A (Conventional steel).
E. Scanning Electron Microscopy (SEM): -
SEM was conducted at crack surface to reveal more details about failure mechanism. The comments are given next to the individual photograph.
Figure: 13 Crack surface SEM view showing Interdendritic
nature of cracks.
F. Microstructural Examination: -
Microstructural examination was carried out at various locations as listed below. Initially, the examination was done in as-polished condition and
thereafter in etched condition.
• Longitudinal cross section at crack
• Longitudinal cross section at crack tip
• Crack surface
Longitudinal cross section at crack
Figure 14: Sample in Figure 15: Crack edge as-polished
as-mounted view showing branched nature of
cracks, 200X
Figure: 16 Crack Edge Figure: 17 Crack Edge 500X200X
Figures 16 & 17 are crack edge microstructures of dendritic structure of bainite/martensite with carbides. Intergranular cracks, branching in nature
and filled with scale are seen.
Longitudinal cross section at crack tip
Figure 18: Sample in Figure 19: As-polished view
as mounted Condition showing branching nature of
crack, filled with scale. 500X
Figure 20: Panoramic views of the crack. Branched nature of
intergranular crack, filled with scale crack is seen initiating from
inner edge. Microstructure is dendritic structure of bainite/
martensite with carbides.
Crack surface morphology :
Figure 21: Sample in Figure 22: Surface as-polished
asmounted condition. view Showing, 200X
Figure: 23 Surface Figure: 24 Surface 200X200X
G. EDS After Metallography:
EDS analysis was conducted inside crack, to find out presence of different elements. The results of analyses are reported in Table 3.
Table 3. Results of the EDS analysis: Inference: EDS
analysis inside crack reveals presence of iron oxide.
H. Tensile Test:
Tensile test was carried out on the test specimen drawn
from sample. The results are reported in Table 4.
Table 4: Tensile test result.
Inference: Results shows that the material does not meet with the requirement as per the specification API 934 A (conventional steels)
and elongation requirement as per the specification ASME Sec II C SFA 5.5 E9015-B3. The strength values are very high.
I. Hardness Measurement:
Hardness was measured at different locations. The results are
reported in Table 5 & 6.
Table 5 Hardness test results:
Inference: Weld and HAZ values are higher than PM.Table 6 Hardness test results:
Inference: Results shows that the material does not meet with the requirement as per the specification API 934 A (conventional
steels). The hardness values are very high.
J. Microhardness Profile: -
Microhardness test was conducted at various locations on and around crack, the result is reported in Table 7.
Table 7 Microhardness profile test result: Near crack
Inference: Hardness value seems to be on higher side.
INVESTIGATION STEPS – OUTER WELD
A. Visual Examination: -
Visual examination was carried out on sample received for investigation along with photographic evidences as under:
Figure 25: Shows failed boat sample of outer weld in
as-received condition. Crack is observed on the OD
weld region which is highlighted. The defect location
mapping show’s location from where boat sample was
removed.
Figure: 26 Sectional views of the sample. Crack is seen in the
marked region.
Figure: 27 Shows Sound weld sample of outer weld in
as-received condition. Visually no significant damage
is seen.
B. Wet Fluorescent Magnetic Particle Inspection (WFMPI):-
To find out any surface defects/auxiliary cracking, WFMPI test was carried out on the sample. Equipment Used: Controls and equipment make
Yoke Type Electromagnetic Crack detector.
Figure: 28 Shows failed outer weld sample in
WFMPI tested condition. Linear indication is seen
Figure: 29 Sectional views of the failed outer weld
sample in WFMPI tested condition. Linear indication
is seen in marked region extending in the cross-section.
Figure: 30 Shows sound outer weld in WFMPI tested
condition. No significant indication is seen.
C. Low Magnification Examination: -
Low magnification examination was carried out on fracture surface to reveal more details about failure mechanism and corrosion pattern.
Figure: 31 Crack surface low magnification view showing
brittle nature of crack.
D. Chemical Analysis: -
Table 8 Results of chemical analysis obtained through optical emission spectroscopy:
Inference: Results show that the material meets with the requirements as per the specification of ASME II C SFA 5.23 EB3R & API 934A.
*Required as per API 934 A (Conventional steel).
E. Scanning Electron Microscopy (SEM):
SEM was conducted at crack surface to reveal more details about failure mechanism. The comments are given next to the individual photograph.
Figure: 32 Crack surface SEM view showing interdendritic
nature of cracks. 100X
F. Microstructural Examination:
Microstructural examination was carried out at various
locations as listed below. Initially, the examination was done in
as-polished condition and thereafter in etched condition.
• Longitudinal cross section at crack
• Longitudinal cross section at crack tip
• Crack surface
• Transverse cross-section- Sound Weld sample
Longitudinal cross section at crack
Figure 33: Sample in Figure: 34 Crack edge polished view showing
asmounted Condition branched nature of cracks, filled with scale. 100X
Figure 35: Crack edge Figure 36: Crack edge 500X100X
Figures 35 & 36 shows crack edge microstructures of dendritic structure of bainite/martensite with carbides. Intergranular cracks, branching in
nature and filled with scale are seen.

Longitudinal cross section at crack tip
Figure 37: Sample Figure: 38 As-polished view
in asmounted showing wide open crack,
Condition mildly branching in nature,
filled with scale, 100X
Figure: 39 Panoramic views of the crack. Intergranular wide
opening crack, branching in nature and filled with scale.
Microstructure is dendritic structure of bainite/martensite with
carbides, 100X
Crack Surface
Figure: 40 Sample Figure: 41 Surface as-polished
in asmounted view showing branched nature
condition. of cracks. 100X.
Figure: 42 Surface, Figure: 43 Surface, 500X100X
Figures 42 & 43 are surface microstructures of dendritic structure
of bainite/martensite with carbides. Intergranular cracks,
branching in nature are seen.
Transverse cross-section- Sound Weld Sample
Figure: 44 Sample Figure: 45 Weld microstructure
in asmounted of dendritic structure bainite/
of condition martensite with carbides. No
significant cracking is seen.
100 X
G. EDS After Metallography: -
EDS analysis was conducted inside crack, to find out presence of different elements. The results of analyses are reported in Table 9.
Table 9 Results of the EDS analysis:
Inference: EDS analysis reveals presence of iron oxide.H. Tensile Test: -
Tensile test was carried out on the test specimen drawn from sample. The results are reported in Table 10.
Table 10 Tensile test result.
Inference: Results shows that the material does not meet with the requirement as per the specification API 934 A (conventional steels).
The strength values are very high.
I. Hardness Measurement:
Hardness was measured at different locations. The results are reported in Table 11, 12 & 13.
Table 11 Hardness test results: Cracked Weld location
Inference: Results shows that the material does not meet with the requirement as per the specification API 934 A (conventional steels).
The hardness values are very high.
Table 12 Hardness test results: Sound Weld location
Inference: Weld and HAZ value seems higher.Table 13 Hardness test results: Sound Weld Location
Inference: Results shows that the material does not meet with the requirement as per the specification API 934 A (conventional steels).
The hardness values are very high.
J. Microhardness Profile: -
Microhardness test was conducted at various locations on and
around crack, result is reported in Table 14.
Inference: Hardness value seems to be higher.NDE and Hardness Check on Channel Side Nozzles Joints:
a) Reading for In-Situ hardness measurements for Nozzle joints are as follows.
b) Nozzles (A1, B1 & M1) to dished head joint were examined by MT (Outside) & PT (Inside) to detect the surface defect, and manual
UT examination was performed to detect volumetric defects. No unacceptable indications / discontinuity observed in MT, PT & Manual
UT examination for the Nozzles to dished head welds.
FEA for the Local Nozzle (A1 & B1) Load on Chanel side of
S10-E-5002:
Static structural analysis has been carried out to check the mechanical integrity of the equipment. The induced stress in the equipment
is found within the code allowable limit. Hence, equipment is safe for various load conditions.
DISCUSSION:
Leakage was noticed from the Circumferential weld joint of Heat Exchanger after 10 months of service. On inspection, multiple transverse
cracks were observed at the weld joint. 34 cracks were predominantly visible from inside and 12 cracks were through and through seen
from outside most of the cracks are within 270o - 0 o – 90 o (upper half portion of the Circumferential Weld). Boat samples were removed
from outer weld (Crack location and Sound weld location) as well as a sample was removed from inner weld (crack location) for investigation.
Inner weld
The inner weld sample is meeting with chemical composition requirements as per the specification of ASME II Part C SFA 5.5 E9015-B3 & API
934A. The X factor of weld is also matching with the requirements. However, the tensile and hardness values are not meeting with the specified
requirements of API 934 A (conventional steels). The hardness values in weld and HAZ are higher than the parent metal. The weld shows
dendritic microstructure of bainite/martensite with carbides.
The inner weld sample received for the investigation showed multiple transverse cracks in WFMPI. The crack surface shows interdendritic nature
in SEM. Microstructural examination revealed interdendritic cracks, branching in nature, filled with scale. EDS analysis indicates presence of oxide
scale in the cracks. Microhardness profile reveals higher hardness values at the weld.
Outer weld
The outer weld sample is meeting with chemical composition requirements as per the specification of ASME II C SFA 5.23 EB3R & API 934A.
The X factor of weld is also matching with the requirements. However, the tensile and hardness values are not meeting with the specified
requirements of API 934 A (conventional steels). The hardness values in weld and HAZ are higher than the parent metal. The weld shows
dendritic microstructure of bainite/martensite with carbides. The outer location sample showed transverse cracking at weld joint. Interdendritic
cracks, branching in nature, filled with scale are seen in microstructural examination. EDS analysis indicates presence of oxide scale in the
cracks of outer weld. The micro hardness values are also very high at this location.
The sample from sound weld did not show any indication in WFMPI tested condition. However, high hardness values are observed in weld and
HAZ region compared to Parent Metal like the observation made from the cracked welds. This is indicative of presence of high residual stresses in
the weld.
Similar type of cracking was observed in the weld joints in insitu metallography examination at site. All the data and the tests done indicate that the
cracking seen in Circumferential weld joint is likely to have been caused due to inadequate local PWHT and the resultant higher local hardness. The
presence of interdendritic cracks, filled with scale along with the high hardness and tensile strength values in weld clearly indicate possibility of
inadequate PWHT. Weld joints with inadequate PWHT having high hardness possibly carry undetectable level of micro cracks. Further development
of such nascent micro cracks is likely over a period when the weld joint is subjected to operating conditions.
Transverse cracking in weld is generally associated with weld metal that is significantly higher in strength compared to the base metal. It can occur
because of longitudinal residual stress. The tensile test results of the weld indicate presence of higher stresses. During welding, weld metal shrinkage
in longitudinal direction may not have been accommodated, as the nearby base metal (which in this case is tube sheet with thickness around 350mm),
might have resisted it. As a result, the weld metal might have developed longitudinal stresses along the welding direction which led to eventual cracking
in transverse direction. These residual stresses may not have been adequately relieved during PWHT as seen from the tensile and hardness test results.
In design where large local heat sink exists like this case with tube sheet of 350mm just adjacent to Circumferential joint within the vicinity of 50mm Local
PWHT is generally not effective as the heat distribution (PWHT temperature and Time) along the through thickness of the weld could not be achieved
uniformly as large amount of heat would have been thermally conducted to this large heat sink resulting in deficient relieving of the residual stresses in
weld joint. Possibility of furnace PWHT could have been checked where entire equipment gets heated up uniformly to achieve uniform temperature though
out the Equipment and no local portion act as a heat sink thus providing proper stress relieving of high restrained joints like Circumferential weld joint.
On weld joints other than Circumferential weld joint, as such no discontinuity was observed in MT, PT & Manual UT examination. Also, Hardness observed
in these intact welds were well below the acceptable limits. As these welds were Post weld heat treated in the Furnace at vendor’s shop, and the results of
NDE and Hardness testing are found within limit, it is reasonable to infer that the PWHT performed for these joints was effective.
CONCLUSION
a. For Circumferential weld joint The OD weld & ID weld sample chemistry matches with specification of ASME II C SFA 5.23 EB3R and ASME II Part C SFA
5.5 E9015-B3 respectively. The X factor values are also satisfactory.
b. The tensile and hardness test results are not meeting with the requirement as per the specification API 934 A (conventional steels).
c. The crack morphology is intergranular, branching in nature and filled with oxide scale, that is the principal feature of Hydrogen assisted cracking.
Circumferential seam was a closing joint for the equipment and fabrication sequence necessitated that local PWHT only be feasible for this joint.
Inadequate local PWHT can lead to non-uniform stress relieving and thus result into some areas with higher hardness and undetectable nascent micro cracks.
This makes the weld joint more susceptible to further propagation of such micro cracks during service.
The cracking in Circumferential weld joint is observed to be intergranular in nature. Such cracks are likely to have been caused in the areas with higher hardness
in the weld with further assistance from Hydrogen trapped at the grain boundaries during service. Also, cracks are limited to the Circumferential Joint only and
integrity of the entire equipment other than Circumferential Joint is found to be in order.
RECOMMENDATION
Proper controls for Local PWHT shall be established to achieve uniform heat distribution throughout the weld thickness, where large heat sink is available near
the weld.