ROLE OF METALLOGRAPHY IN METALLURGICAL FAILURE ANALYSIS

Mr. B K Shah (Quality Assurance Division)

BARC, Mumbai.

I. Introduction

It is now a well recognized adage that failures do not just happen, but they are
caused. Hence in modern technology, failure analysis assumes supreme
importance. There are two dominating reasons for investigating failures. Unless the
true cause of the failure is known, no remedial action is possible. Secondly, it helps
a lot in improving the reliability and safety of machinery and structural hardware
which form the heart of modern industries. In the last few decades, systematic
investigations carried out on several failed components have generated a fund of
useful information for taking suitable remedial measures to prevent similar
recurrences. A new science of Failure Analysis has emerged.

Metallurgists enjoy a good failure. A tough failure analysis problem not only poses
an intellectual challenge, but is reassurance that a part is not over designed, and is
also good job security. Failures will always be with us because designers are always
exploiting the ultimate capabilities of materials.

When confronted with a failure, the instinct of all but the metallurgist is that the
materials are at fault- either improper material specification or an actual defect.
Often the metallurgist’s first task is to determine whether the problem presented is
really metallurgical, or instead the result of abusive use in service, overloading, or
just poor mechanical design.

Failure analysis is an exceptionally challenging and exciting task. The investigator

conducting the failure analysis should have an aptitude and interest in carrying out
the investigation. Also, he should not have a preconceived notion of the cause of the
failure. The investigator must have a sound back ground in the areas of fracture
mechanics, fabrications technology and an analytical mind. He must not hesitate to
take the assistance of experts in other relevant fields. Every failure analysis will
have something new to convey, though it may have some semblance to similar
failures occurred earlier. The investigator should keep abreast of modern
developments in processing, fabrication etc.

II. Fundamental Sources of Failure

The analyst should carefully plan the sequence of tests for each investigation. A
wrong sequence can introduce artifacts or mask or even destroy important evidence
and thus defeat the purpose.
A part or assembly is considered to have failed under one of three conditions : (a)
when it becomes completely inoperable, (b) when it is still operable but is no longer
able to perform its intended function satisfactorily, or (c) when serous deterioration
has made it unreliable or unsafe for continued use, thus necessitating its immediate
removal from service for repair or replacement. The fundamental sources of failure
include :

46
 • Deficiencies in Design
• Deficiencies in Selection of material
• Imperfections in Material
• Deficiencies in Processing
- Cold forming and related operations
- Improper heat treatment
- Welding
• Errors in Assembly
• Improper Service Conditions
• Combination of above factors
• Sabotage

III. General Practice in Failure Analysis

Although the sequence is subject to variation depending on the nature of a specific

failure, the principal stages that comprise the investigation and analysis of a failure
are as follows :

1) Collection of background data and selection of samples

2) Preliminary examination of the failed part (visual examination and record
keeping)
3) Nondestructive testing
4) Mechanical testing (including hardness and toughness testing)
5) Selection, identification, preservation and/or cleaning of all specimens
6) Macroscopic examination and analysis (fracture surfaces, secondary cracks
and other surface phenomena)
7) Microscopic examination and analysis
8) Selection and preparation of metallographic sections
9) Examination and analysis of metallographic sections
10) Determination of failure mechanism
11) Chemical analyses (bulk, local surface corrosion products, deposits or
coatings, and microprobe analysis
12) Analysis of fracture mechanics
13) Testing under simulated service conditions (special tests)
14) Analysis of all the evidence, formulation of conclusions, and writing the report
(including recommendations).

IV. Modes of Failure

1) Mechanical
• Yielding
• Overload Ductile Fracture
• Brittle Fracture
• Creep & Stress Rupture
• Fatigue
• Wear

2) Corrosion
• Uniform Corrosion
• Pitting

47
 • Crevice Corrosion
• Galvanic Corrosion
• Selective Leaching
• Intergranular Corrosion

3) Corrosion – Mechanical
• Stress Corrosion Cracking
• Corrosion Fatigue
• Hydrogen Embrittlement
• Erosion Corrosion

Failure Mode Gross appearance Macroscopic Fractographic

appearance appearance
Ductile Large plastic Shear lips Dimples
Deformation at
fracture
Brittle No Plastic Flat Cleavage (river
Deformation patterns)
Fatigue No plastic Beach marks Striations
deformation
Intergranular No plastic - Rock candy
Fracture (TE, deformation
HE, SCC)

V. Tools of Metallurgical Failure Analysis

Optical Microscope (OM)

♣ Type of Microstructure
♣ Nature of Crack

Scanning Electron Microscope (SEM)

Fractography

Microchemical Characterization

♣ EPMA
♣ AES & ESCA

Failure analysis is incomplete without a microscopic examination. Because, it is at

this stage that information is gathered about the origin of the crack/fracture. A
common tool that is readily available is the optical stereo microscope. This has very
low magnification capability and enables the direct observation of fracture surface.
Reasonably large specimens can be handled and the surfaces can be seen as such
without any preparation. Once the origin is located, it is studied in greater detail at
higher magnifications using optical microscope to obtain further information like :

48
 (i) length of an existing crack,
(ii) presence of inclusions at the origin
(iii) presence of striations and determination of their spacing
(iv) length of a fatigue crack before it became critical
(v) Presence of any corrosion products and host of other information.

The scanning electron microscope has been of great help in the study of fracture
surfaces. It allows a direct study of actual fracture surfaces on specimens small
enough to be accommodated in the specimen chamber of SEM. A variety of signals
emanates from the specimen viz. secondary electrons, back scattered electrons,
characteristic X-rays, etc. The most common mode of imaging for Fractography is
the secondary electron imaging.

Non-Destructive and Mechanical Tests are extremely useful in failure analysis.

Because, the former (NDT) helps to reveal the presence of any surface and internal
flaws in the material and the latter to ascertain whether the material confirms to
specifications.

The conventional NDT Techniques are broadly classified as

(i) Those that detect surface and sub-surface flaws viz. LPT, MPT, ECT
(ii) Those that detect internal flaws viz. UT, RT

Important mechanical tests carried out during failure investigation include Tension
Test, Impact Toughness Test and Hardness Test. Tension Test indicates whether
strength of the material was adequate to withstand service stress. Impact Test
indicates whether toughness of material was adequate to resist brittle fracture.
Hardness Traverse indicate the Presence of any hard and brittle microstructure in
any zone(base metal/weld pool/HAZ).

VI. Failure Case Studies

A. Mechanical Failure

A.1 Sea Water Pump Shaft Failure

A broken piece of sea water pump shaft was received from a nuclear power station
to find out the cause of failure. The shaft was reported to have been fabricated at
site from austenitic stainless steel AISI 316 bar stock. It was informed that (i) shaft
top piece had broken at 10 cm from the top end, (ii) in addition to main fracture 5
more spiral cracks were seen on the broken pieces, and (iii) the pump impeller on
checking was found to be free to rotate indicating thereby that failure had not been
initiated due to jamming of the pump.

A.2 RMS Titanic Metallurgical Failure Analysis

i). Maiden Voyage of Titanic

• April 10, 1912 from Southampton, England to New York, USA
269 m long, 28 m wide, 18 m from water line to Deck, 5.3 m from keel to
funnel and gross weight of 46000 tons

49
 • April 12, 1912 at 1140 pm,
- it struck an ‘Iceberg’ 3-6 times its own mass
- 6 Forward compartment ruptured
- within 2 hours 40 minutes, the ship sank
- loss of more than 1500 lives

ii) Metallurgical Failure Analysis

• Steel sample collected from hull of ship during expedition in the North
Atlantic on August 15, 1996.
• Failure Analysis carried out at University of Missouri - Rola
• it involved : Chemical Analysis, Metallography, Fractography (SEM),
Tensile Test & Charpy Test

iii) Chemical Composition

Titanic Hull Plate ASTM A36

C 0.21 0.20
Mn 0.47 0.55
P 0.045 0.012
S 0.069 0.037
Si 0.017 0.007
Cu 0.024 0.01
O 0.013 0.079
N 0.0035 0.0032
Mn:S Ratio 6.8 : 1 14.9 : 1

♣ High oxygen & low Si - partially Deoxidized (Semi Killed) Steel

♣ High ‘P’, High ‘S’, Low ‘Mn’, & Low ‘Mn:S’ ratio
♣ Relatively High P, O & S - Tendency to embrittle at Low Temperature
♣ Steel Probably produced by Acid Open Hearth Process

iv) Metallography

♣Grain Size : 60 µm (longitudinal) & 42 µm (Transverse)

♣Banded Microstructure (longitudinal)
♣Sulphide Inclusions elongated in the direction of banding

v) Tensile Testing
Titanic AISI 1020
UTS 417 Mpa 379 Mpa
YS 193 Mpa 207 Mpa
% Elongation 29 26
% RA 57 50

50
 vi) Charpy Impact Testing

Titanic Steel A36 Steel

DBTT (20J) 32oC (Long) - 27oC
56oC (Trans.)
o
FATT (50%SF) 49 C (Long) - 3oC
59oC (Trans.)

♣ At elevated temperature, the impact energy values for longitudinal Titanic steel is
much higher than transverse specimen indicating the effect of banding
♣ Sea Water temperature at the time of collision was - 2oC .
♣ SEM Fractography showed cleavage Brittle Fracture on fractured longitudinal
charpy impact test specimen tested at 0oC.

vii) Conclusion

1 Titanic suffered Catastrophic Brittle Fracture

DBTT (32-59oC) much higher than sea water temperature (-2oC)
2 Steel used in construction was probably the best available plain carbon steel ship
plate during 1909 to 1911, but not acceptable as per current practice.
3 Suitable Navigational Aids did not exist in 1912 to sight ‘Iceberg’ from a greater
distance to allow more time for evasive action.
4 Olympic, built from similar steel, by same shipyard and same design, served
more than 20 years. The only difference was a ‘Big Iceberg’

A.3 Failure of Rocket Motor Casing during Hydro Test

An excellent example to explain the importance of selection of material on the

toughness consideration is the failure of 0.6 m diameter rocket motor case during
hydro-test. The motor case failed during hydro test at a pressure 38 kg/cm2, which
was about 56% of proof pressure. Post failure examination showed that failure
originated in a weld defect (0.56 cm deep and 3.6 cm long).
Design specification and Failure Data

Proof pressure 63.4 kg/cm2

Failure occurred at 38 kg/cm2
Material 250 Grade maraging steel
Thickness of plate 1.85 cm
Process Submerged arc automatic welding
Yield strength 16.800 kg/cm2

Membrane stress at failure 7,000 kg/cm2

51
The real lesson in this failure is the lack of design knowledge that went into the
material selection Even though the material selected (Grade 250 maraging steel) has
high yield strength of 16.800 kg/cm2, its plane strain fracture toughness is only 8,938
kg.cm-3/2. At the design stress of' 11,200 kg/cm 2
, a critical defect of only 0.2 cm in
depth could have caused catastrophic failure. Had they used grade 200 maraging
steel. whose yield strength is only 15,400 kg/cm2 (about 10 percent lower than Grade
250) but the toughness is about 16,800 kg.cm-3/2 (about double of Grade 250), the
chamber would not have failed. At a design stress 200 kg/cm2, the Grade 200 steel
would take a flaw 1.42 cm in depth to cause plane strain fracture and this is a very
large flaw. In all probability, a flaw this large in this type of' material would be
arrested under plane stress condition anyway as soon as it grew through the
thickness. Thus, the Grade 200 maraging steel would have been a more reliable
material for the chamber than Grade 250 maraging steel.

In summary, using a lower strength steel with higher toughness(a large KIC/óY) even
at a design stress which was a higher percentage of the yield strength would have
been better.

B. Corrosion Failure

B.1 Failure of Spin Head (Used in Synthetic Fibre Industry) during Hydro Test
This case relates to the failure of a spin head component used in synthetic fibre
industry for converting molten polymer into fibres. Fabrication involved welding,
stress relieving (for dimensional stability) and pickling. Fabricated component gave
way during commissioning hydro test. Austenitic stainless steel AISI 316 was used
in place of recommended material stabilised grade AISI 321. After welding,
component was stress relieved at 550 C/100 min. and acid pickled. Welding and
stress relieving resulted in sensitised material and acid pickling attacked it
intergranularly ultimately causing leakage during hydro test.

C. Corrosion - Mechanical Failure

C.1 Heat Exchanger Shell Failure

A heat exchanger shell failed during a hydrotest in a fertilizer plant. The shell was
fabricated of type 304 (UNS S30400) plate, 8-mm thick. Fabrication involved
bending of the plates, longitudinal welding to make sub-shells, and finally,
circumferential welding between sub-shells. The shell had passed the hydrotest at
the fabricator’s shop. After about three months, when the hydrotest was being done
at the fertilizer plant, water jetted out near the weld, and the shell could not be
pressurized.

Failure analysis involved LPT, radiography, chemical analysis, delta-ferrite

measurement, and other tests. Microscopic examination indicated the presence of a
sensitized heat-affected zone (HAZ) near the weld and intergranular cracks in the
HAZ starting from the outside surface (Figures 3 through 5).

These observations clearly indicated it was intergranular stress corrosion cracking

(IGSCC). For IGSCC to initiate and propagate, there must be the simultaneous

52
presence of three factors : sensitization, tensile stress, and a specific corrosive
environment. In this case, welding of type 304 provided the first two contributing
factors: sensitized HAZ and tensile residual stress. The acid solution HNO3-HF-H2O,
used to remove the weld discoloration, provided the specific corrosive environment.
A hydrotest at the fabricator’s shop was performed immediately after acid cleaning,
i.e., immediately after IGSCC initiation, and, therefore, the shell passed this test. A
subsequent hydrotest at the fertilizer plant was performed after about three months.
During this period, IGSCC continued to propagate from the wled-induced residual
stress and the acid residue retained in the IGSCC-attacked HAZ. Ultimately, IGSCC
propagated through the wall and the shell could not withstand the hydrotest.

C.2 Poison Pipe for Nuclear Power Plant

This case refers to failure of poison pipe stainless steel AISI 304 in a nuclear power
plant. The poison pipe is meant for carrying sodium meteorite poison to the reactor
in case emergency shut-down is needed. However, since commissioning of the
reactor, such an occasion has never arisen and the pipe is normally filled with
reactor water as it is directly connected to the reactor. This pipe is covered with
asbestos insulation. Leak was noticed in the pipe, away from reactor, near the guard
pipe to poison pipe weld. The guard pipe is also of stainless AISI 304. Failure
analysis involved LPT, radiography, microscopic examination and micro hardness
measurements. LPT indicated crack location. Radiography confirmed LPT
indications and did not show any other defects. Microscopic examination of a
transverse section of pipe at most prominent crack location showed that (i) cracks
were starting from outside surface, some of them terminating within the wall and
some going through and through the wall thickness and (ii) cracks were following
Transgranular path. Oxalic acid etch test as per ASTM A262 practice A showed
acceptable dual structure indicating that material was not susceptible to IGC.
Microhardness measurements, on a section very near crack location, showe3d 230-
280 VPN near OD, 170-205 VPN near ID and 155-165 in the middle of wall
indicating the presence of some amount of cold work on the outside surface. Cracks
were originating from outside surface and were following Transgranular path. The
asbestos insulation on outside surface of leaching showed about 485 ppm chloride.
Therefore, the cracks in the pipe were attributed to chloride stress corrosion cracking
of austenitic stainless steel.

VII. Concluding Remarks

From the instances of failures in austenitic stainless steels described above, it is

evident that these could have been prevented by careful selection of suitable
materials and fabrication procedures. A recent NASA study on failures of
engineering structures has very aptly concluded that most failures were not caused
by the lack of new information but by the failure to use effectively the existing
knowledge. Also, there are very few failures due to causes which required new or as
yet undeveloped materials. Analysing the failures in metallic structures to bring out
causes for failures and to suggest suitable remedial action is a repetitive exercise.
However, it is an important exercise as it focuses the attention of all concerned to the
causative factors of such failures.

53
While a variety of tools are available for the investigation, the proper choice of a
particular tool is left to the analyst. In a majority of cases, the magnification available
in optical microscope is sufficient for the analysis. However, the need to have a
higher magnification becomes imperative in a few cases. In view of high cost
involved in a scanning electron microscope observation, its essentiality has to be
justified. Also the short time, available to the analyst in cases where resumption of
production in a plant is at stage is a point of restraint. Above all, the intuitive feeling
and the ingenuity of the analyst determines the success of a failure analysis.

54