03/08/2017

Engineering Failure Analysis

Josias van der Merwe

Course Background and Purpose

• The metallurgical engineer can be expected to operate in a very
wide field of operations. His activities may range from the
processing of minerals to the development of alloys and non‐
metallic materials. In anyone field he will be involved with the
application and performance of materials and structures. Such
materials and structures have a finite useful life but sometimes
premature failures occur for a variety of reasons. The metallurgical
engineer has to have a sound understanding of materials properties
and behaviour in order to assess the life of components and
structures or in order to analyse a premature failure.
•
• Since this is a 9 credit course, students are expected to spend at
least 70 hours working on the course outside of the timetabled
lectures and tutorials.

1
 03/08/2017

Course Objectives
• This course aims to give a basic approach to
analysing engineering failures, and will address
the following objectives:
– To develop skills to evaluate and appreciate the
properties of metals and other materials.
– To become familiar with the main modes of failure of
materials.
– To develop skills to analyse common failures through
the application of engineering and science principles.
– To develop skills of reporting on investigations.
– To develop skills in observation/measurement and
deduction

Course Outcomes
• Understand how to analyse typical failures.
• Design against fatigue failure and analyzing
fatigue failures and characteristics
• Describe fracture and determine fracture modes,
such as ductile‐to‐brittle transition, and describe
fractographic features and the role of stress
concentrations.
• Apply simple fracture mechanics and calculate
leak‐before‐break conditions.
• Handle failures related to high temperature
conditions.

2
 03/08/2017

ECSA Outcomes
Outcome 4: Investigations, experiments and Assessment Satisfactory
data analysis performance
Learning outcome: Demonstrate competence This outcome will be assessed by a The outcome will be
to design and conduct investigations and combination of assignments and lab satisfied with a final
experiments. experiments which will consist of a mark of more than 50%
proposal and lab report, as well as
Associated Assessment Criteria: relevant test and exam questions
The candidate executes an acceptable process
including but not restricted to:
1. Plans and conducts investigations and
experiments;
2. Conducts a literature search and critically
evaluates material;
3. Performs necessary analyses;
4. Selects and uses appropriate equipment or
software;
5. Analyses, interprets and derives
information from data;
6. Draws conclusions based on evidence;
7. Communicates the purpose, process and
outcomes in a technical report.

ECSA outcome
Exit Level Outcome 5: Engineering methods, skills Assessment Satisfactory
and tools, including Information Technology performance

Learning outcome: Demonstrate competence to use This outcome will be The outcome will
appropriate engineering methods, skills and tools, assessed by a be satisfied with a
including those based on information technology. combination of final mark of more
assignments and lab than 50%
Associated Assessment Criteria: experiments which will
The candidate: consist of a proposal and
1. Uses method, skill or tool effectively by: lab report, as well as
a) Selecting and assessing the applicability and relevant test and exam
limitations of the method, skill or tool; questions
b) Properly applying the method, skill or tool;
c) Critically testing and assessing the end‐results
produced by the method, skill or tool.
2. Creates computer applications as required by the
discipline.

3
 03/08/2017

Overview
• 1. Introduction
• 2. Plasticity and strengthening mechanisms
• 3. Fracture
• 4. Fracture Mechanics
• 5. High temperature deformation response of crystalline
solids
• 6. Transition temperature approach to fracture control
• 7. Microstructural aspects of fracture toughness
• 8. Environment assisted cracking
• 9. Cyclic stress and strain fatigue
• 10. Fatigue crack propagation
• 11. Fatigue of welded structures
• 12. Analysis of engineering failures

ASSESSMENT AND RULES

– Class test – 20%

– Assignment, tutorial and laboratory project – 20%
– June examination – 60%

4
 03/08/2017

Highly Recommended

• Hertzberg, R.W. Deformation and Fracture Mechanics

of Engineering Materials, John Wiley, 4th Edition,1996.
• Dieter, G.E. Mechanical Metallurgy, Mc Graw‐Hill, New
York, 3rd Edition, 1986.

Week no Lecture date Course work Lab/Tut Lab Assessment

Date
1 24‐28 July Introduction ‐What constitutes failure; Tensile
response
2 31‐4 Aug Strengthening Mechanisms 3 Aug Introduction ‐ Proposal
3 7‐11 Aug Fracture Lab 1 ‐ Proposal

4 14‐18 Aug Fracture toughness

5 21‐25 Aug Fracture mechanics 24 Aug Sample Preparation

Hardness profile
6 28‐1 Sep High temperature deformation response of Tutorial
crystalline solids
7 4‐8 Sep Impact properties 7 Sep Tutorial
Recess (28‐31 Mar)
8 18‐22 Sep Microstructural aspects of fracture toughness

9 25‐29 Sep Environment assisted cracking 28 Sep Failure Lab 2 Group report

10 2‐6 Oct Cyclic stress and strain fatigue 18 Apr – TEST

RW225

11 9‐13 Oct Fatigue crack propagation 12 Oct SEM Assignment 1

12 16‐20 Oct Analysis of engineering failures and Case
studies
13 23‐26 Oct Welding failures 25 Oct Presentation Assignment

5
 03/08/2017

Overview
Theme Reference Topics Reference
(Hertzberg) (Dieter)

1. Introduction
What constitutes Elastic vs ductile deformation, Chapter 1
failure? definition of terms

Tensile response of Chapter 1 Stress‐strain curves

materials
2. Plasticity and Strengthening Mechanisms
Plasticity Chapter 3
High temperature Chapter 6 Creep; Temperature‐stress‐strain‐rate Chapter 13
deformation response relations; Deformation mechanisms;
of crystalline solids

Overview (cont.)
3. Fracture
Fracture Chapter 7 Theoretical cohesive strength; Stress Chapter 7
concentrations; Notch strengthening;

Fracture Mechanics Chapter 8 Plane strain fracture toughness testing; Chapter 11

Design philosophy;
Transition temperature Chapter 9 Charpy Impact testing; Additional fracture Chapter 14
approach to fracture control test methods;

Microstructural aspects of Chapter 10 Generalities; Cleanliness; Microstructural

fracture toughness refinement;

Environment assisted Chapter 11 Fracture mechanics test methods;

Embrittlement models;
cracking

6
 03/08/2017

Overview (cont.)
4. Fatigue

Cyclic stress and strain Chapter 12 Stress vs strain controlled fatigue; Chapter 12
fatigue Avoidance of fatigue damage; Fatigue
crack initiation mechanisms;

Fatigue crack Chapter 13 Macroscopic fracture modes;

propagation Influence of load interactions;

Fatigue of welded
structures
5. Analysis of engineering failures
Chapter 14 Problem solving

Tutorials
• 3 Aug RW5 (Introduction, Proposal)
• 24 Aug RW5 (Lab 1, Hardness profile)
• 7 Sep RW5 (Lab 2, Failure)
• 28 Sep RW5 (SEM)
• 12 Oct RW5 (Failure)
• 26 Oct RW5 (Presentation)

7
 03/08/2017

Why do we do failure analysis?

• Health and safety
• Limit costs
• Functionality
• Insurance claims
• Mechanism

de Havilland DH 106 Comet ‐

the world's first commercial jetliner

8
 03/08/2017

• Developed and manufactured by de Havilland at

its Hatfield Aerodrome, Hertfordshire, United
Kingdom headquarters,
• Comet 1 prototype first flew on 27 July 1949.
• It featured an aerodynamically clean design with
four de Havilland Ghost turbojet engines buried
in the wings, a pressurised fuselage, and large
square windows.
• For the era, it offered a relatively quiet,
comfortable passenger cabin and showed signs of
being a commercial success at its 1952 debut.

Origins
• On 11 March 1943, the Cabinet of the United Kingdom
formed the Brabazon Committee to determine the UK's
airliner needs after the conclusion of the Second World
War
• De Havilland chose to challenge the widely held
scepticism of jet engines as too fuel‐hungry and
unreliable
• In 1945 no turbojet engine manufacturer in the world
was drawing up a design specification for an engine
with the thrust and specific fuel consumption, that
could power an aircraft at the proposed cruising
altitude

9
 03/08/2017

Testing and prototypes

• From 1947 to 1948, de Havilland conducted an extensive research and
development phase, including the use of several stress test rigs at Hatfield
for small components and large assemblies alike.
• Sections of pressurised fuselage were subjected to high‐altitude flight
conditions via a large decompression chamber on‐site, and tested to
failure.
• However, tracing fuselage failure points proved difficult with this method,
and de Havilland ultimately switched to conducting structural tests with a
water tank that could be safely configured to increase pressures gradually.
• The entire forward fuselage section was tested for metal fatigue by
repeatedly pressurising to 19.0 kPa overpressure and depressurising
through more than 16,000 cycles, equivalent to about 40,000 hours of
airline service.
• The windows were also tested under a pressure of 83 kPa, 32.8 kPa above
expected pressures at the normal service ceiling of 11,000 m
• One window frame survived 690 kPa, about 1,250 percent over the
maximum pressure it was expected to encounter in service.

Prototypes
• The first prototype DH 106 Comet (carrying
Class B markings G‐5‐1) was completed in
1949, and was initially used to conduct ground
tests and brief early flights
• The second prototype was registered G‐ALZK
in July 1950 and it was used by the BOAC
Comet Unit at Hurn from April 1951 to carry
out 500 flying hours of crew training and route
proving

10
 03/08/2017

Design
• The Comet was an all‐metal low‐wing cantilever
monoplane powered by four jet engines; it had a four‐place
cockpit occupied by two pilots, a flight engineer, and a
navigator.
• The clean, low‐drag design of the aircraft featured many
design elements that were fairly uncommon at the time,
including a swept‐wing leading edge, integral wing fuel
tanks, and four‐wheel bogie main undercarriage units
designed by de Havilland
• Two pairs of turbojet engines (on the Comet 1s, Halford H.2
Ghosts, subsequently known as de Havilland Ghost 50
Mk1s) were buried into the wings
• One of the most striking aspects of Comet travel was the
quiet, "vibration‐free flying" as touted by BOAC. For
passengers used to propeller‐driven airliners, smooth and
quiet jet flight was a novel experience

Fuselage
• Diverse geographic destinations and cabin pressurisation alike on
the Comet demanded the use of a high proportion of alloys,
plastics, and other materials new to civil aviation across the aircraft
in order to meet certification requirements.
• The Comet's high cabin pressure and fast operating speeds were
unprecedented in commercial aviation, making its fuselage design
an experimental process.
• At its introduction, Comet airframes would be subjected to an
intense, high‐speed operating schedule which included
simultaneous extreme heat from desert airfields and frosty cold
from the kerosene‐filled fuel tanks, still cold from cruising at high
altitude
• The Comet's thin metal skin was composed of advanced new alloys
and was both riveted and chemically bonded, which saved weight
and reduced the risk of fatigue cracks spreading from the rivets

11
 03/08/2017

Propulsion
• The Comet was powered by two pairs of turbojet
engines buried in the wings close to the fuselage.
Chief designer Bishop chose the Comet's
embedded‐engine configuration because it
avoided the drag of podded engines and allowed
for a smaller fin and rudder, since the hazards of
asymmetric thrust were reduced.
• The engines were outfitted with baffles to reduce
noise emissions, and extensive soundproofing
was also implemented to improve passenger
conditions

Operational History
• The earliest production aircraft, registered G‐ALYP ("Yoke
Peter"), first flew on 9 January 1951 and was subsequently
lent to BOAC for development flying by its Comet Unit
• In their first year Comets carried 30,000 passengers
• Flights on the Comet were about 50 percent faster than on
advanced piston‐engined aircraft such as the Douglas DC‐6
(490 mph for the Comet compared to the DC‐6's 315 mph),
and a faster rate of climb further cut flight times.
• In August 1953 BOAC scheduled nine‐stop London to Tokyo
flights by Comet for 36 hours, compared to 86 hours and 35
minutes on their Argonaut piston airliner. (Pan Am's DC‐6B
was scheduled for 46 hours 45 minutes.)
• The five‐stop flight from London to Johannesburg was
scheduled for 21 hr 20 min

12
 03/08/2017

Accidents
• On 26 October 1952, the Comet suffered its first hull loss when a
BOAC flight departing Rome's Ciampino airport failed to become
airborne and ran into rough ground at the end of the runway. Two
passengers sustained minor injuries, and the aircraft, G‐ALYZ, was a
total loss.
• On 3 March 1953, a new Canadian Pacific Airlines Comet 1A,
registered CF‐CUN and named Empress of Hawaii, failed to become
airborne while attempting a night takeoff from Karachi, Pakistan, on
a delivery flight to Australia. The aircraft plunged into a dry drainage
canal and collided with an embankment, killing all five crew and six
passengers on board.
• The accident was the first fatal jetliner crash, as well as the Comet's
first accident to result in fatalities.
• In response, Canadian Pacific cancelled its remaining order for a
second Comet 1A and never operated the type in commercial
service
• Both early accidents were originally attributed to pilot error

Second fatal accident

• occurred on 2 May 1953, when BOAC Flight
783, a Comet 1, registered G‐ALYV, crashed in
a severe thundersquall six minutes after taking
off from Calcutta‐Dum Dum (now Netaji
Subhash Chandra Bose International Airport),
India, killing all 43 on board. Witnesses
observed the wingless Comet on fire plunging
into the village of Jagalgori, leading
investigators to suspect structural failure

13
 03/08/2017

Investigation
• A large portion of the aircraft was recovered and
reassembled at Farnborough, during which the break‐up
was found to have begun with a left elevator spar failure in
the stabilizer.
• The inquiry concluded that the aircraft had encountered
extreme negative G forces during takeoff; severe
turbulence generated by adverse weather was determined
to have induced down‐loading, leading to the loss of the
wings. Examination of the cockpit controls suggested that
the pilot may have inadvertently over‐stressed the aircraft
when pulling out of a steep dive by over‐manipulation of
the fully powered flight controls.
• Investigators did not consider fatigue as a contributory
cause

Comet disasters of 1954

• Rome's Ciampino airport, the site of the first Comet
hull loss, was the origin of more disastrous Comet
flights just over a year later. On 10 January 1954, 20
minutes after taking off from Ciampino, the first
production Comet, G‐ALYP, broke up in mid‐air while
operating BOAC Flight 781 and crashed into the
Mediterranean off the Italian island of Elba with the
loss of all 35 on board.
• With no witnesses to the disaster and only partial
radio transmissions as incomplete evidence, no
obvious reason for the crash could be deduced

14
 03/08/2017

Abell Committee Court of Inquiry

• Media attention centred on sabotage,[88] while other
speculation ranged from clear‐air turbulence to an
explosion of vapour in an empty fuel tank.
• The Abell Committee focused on six potential aerodynamic
and mechanical causes: control flutter (which had led to
the loss of DH 108 prototypes), structural failure due to
high loads or metal fatigue of the wing structure, failure of
the powered flight controls, failure of the window panels
leading to explosive decompression, or fire and other
engine problems.
• The committee concluded that fire was the most likely
cause of the problem, and a number of changes were made
to the aircraft to protect the engines and wings from
damage that might lead to another fire

Another disaster
• On 8 April 1954, Comet G‐ALYY ("Yoke Yoke"), on charter to South
African Airways, was on a leg from Rome to Cairo (of a longer route,
SA Flight 201 from London to Johannesburg), when it crashed in the
Mediterranean near Naples with the loss of all 21 passengers and
crew on board.
• The Comet fleet was immediately grounded once again and a large
investigation board was formed under the direction of the Royal
Aircraft Establishment (RAE).
• Prime Minister Winston Churchill tasked the Royal Navy with
helping to locate and retrieve the wreckage so that the cause of the
accident could be determined.
• The Comet's Certificate of Airworthiness was revoked and Comet 1
line production was suspended at the Hatfield factory while the
BOAC fleet was permanently grounded, cocooned and stored

15
 03/08/2017

Cohen Committee Court of Inquiry

• With the recovery of large sections of G‐ALYP from the Elba crash and
BOAC's donation of an identical airframe, G‐ALYU, for further examination,
an extensive "water torture" test eventually provided conclusive results.
• This time, the entire fuselage was tested in a dedicated water tank that
was built specifically at Farnborough to accommodate its full length.
• Stress around the window corners was found to be much higher than
expected and stresses on the skin were generally more than previously
expected or tested
• This was due to stress concentration, a consequence of the windows'
square shape, which generated levels of stress two or three times greater
than across the rest of the fuselage.
• In 2012 a finite element analysis was carried out to find the stress values
in a digital model of the Comet's cabin window loaded to a pressure
differential of 8.25 psi. In this model the maximum stress level at the
margin of one of the outer row of rivet holes near the corner of the
window was almost five times greater than in the areas of skin remote
from the windows

Cohen Committee
• In water tank testing, engineers subjected G‐ALYU to repeated re‐
pressurisation and over‐pressurisation, and on 24 June 1954, after 3,057
flight cycles (1,221 actual and 1,836 simulated), G‐ALYU burst open.
• Hall, Geoffrey de Havilland and Bishop were immediately called to the
scene, where the water tank was drained to reveal that the fuselage had
ripped open at a corner of the forward left escape hatch cutout. A further
test reproduced the same results.
• Based on these findings, Comet 1 structural failures could be expected at
anywhere from 1,000 to 9,000 cycles. Before the Elba accident, G‐ALYP
had made 1,290 pressurised flights, while G‐ALYY had made 900
pressurised flights before crashing.
• Dr P.B. Walker, Head of the Structures Department at the RAE, said he was
not surprised by this, noting that the difference was about three to one,
and previous experience with metal fatigue suggested a total range of nine
to one between experiment and outcome in the field could result in failure

16
 03/08/2017

Conclusions
• The punch rivet construction technique employed in the Comet's
design had exacerbated its structural fatigue problems; the
aircraft's windows had been engineered to be glued and riveted,
but had been punch riveted only.
• Unlike drill riveting, the imperfect nature of the hole created by
punch riveting could cause fatigue cracks to start developing around
the rivet.
• Principal investigator Hall accepted the RAE's conclusion of design
and construction flaws as the likely explanation for G‐ALYU's
structural failure after 3,060 pressurisation cycles
• The Cohen inquiry closed on 24 November 1954, and although the
inquiry had "found that the basic design of the Comet was sound",
de Havilland nonetheless began a refit programme to strengthen
the fuselage and wing structure, employing thicker gauge skin and
replacing all square windows and panels with rounded versions

• In June 1956 some more wreckage from G‐ALYP was

accidentally trawled up from an area about 15 miles
south of where the original wreckage had been found.
• This wreckage was from the starboard side of the cabin
just above the three front windows.
• Subsequent examination at Farnborough suggested
that the primary failure was probably near to this area
rather than at the rear automatic direction finding
window on the roof of the cabin as had been
previously thought. These findings were kept secret
until the details were published in 2015

17
 03/08/2017

Introduction
• Basic assumptions
– Continuous body
• A body that does not contain voids or empty spaces
– Homogeneous
• If it has identical properties at all points
– Isotropic/Anisotropic
• With respect to some property when that property
does not vary with direction or orientation

What constitutes failure?

• Excessive elastic deformation – such as
buckling
• Yielding of excessive plastic deformation
– During pressure testing of a pressure vessel
– Consider room temperature properties compared
to elevated temperature properties
– Does it depend on the material?
• Fracture

18
 03/08/2017

Fracture
• Sudden brittle fracture
– Brittle fracture is almost always sudden and
unpredictable making it very dangerous
• Fatigue or progressive fracture
• Delayed fracture ‐ Hydrogen

• Can fracture be predictable or not?

Concept of stress and the types of

stress
• Definitions of stress and strain
• Engineering stress and strain
–

–
• True stress and strain
–

19
 03/08/2017

• Where:
• Ai = Initial area
• Li = Initial length
• Lf = Final length
• A0 = Original area

20
 03/08/2017

Relationship between true stress and

strain and engineering stress and
strain
• 1
• ln 1
• These relationships are based on the
unchanging volume of a solid
• If you pull a rod in its length then the cross
section area will decrease:

• ∆ 0
• . .
•

21
 03/08/2017

Distinctions concerning true strain and

engineering strain
• For true strain:
– The sum of the total of the strain increments =
The total strain (Not true for engineering strain)
– Same type of strain for a rod extended to twice its
length or compressed to half its length (Not true
for engineering strain)
• Engineering strain:
• Is similar to true strain when the strain is small

Example
• Consider the following increments in
elongation and determine the true and
engineering strain for the increments as well
as the total:
Increment Length of rod
1 50 mm
2 55 mm
3 60.5 mm
4 66.55 mm

22
 03/08/2017

Example
• A 25 cm long rod with a diameter of 0.25 cm is
loaded with a 4500 N weight. If the diameter
decreases to 0.22 cm, compute the following:
• a) The final length of the rod
• b) The true stress and strain at this load
• c) The engineering stress and strain at this
load

Constant volume and true strain

• Because of an assumption of constant volume
of a solid:
• 0
•
•
• Engineering strain:
• 1 . 1 . 1 1

23
 03/08/2017

Normal stress vs shear stress

•

Normal strain vs Shear strain

Shear strain:
≅

24
 03/08/2017

Stress‐strain curves
Type I – Elastic response
• – (Elastic spring)
• ‐ (Hooke’s law)

E – Young’s modulus

Experimental measurement of E –
Young’s modulus

25
 03/08/2017

Does the Young’s modulus change with

metallurgical condition of material?
• Esteel is not f(heat treatment, grain size, cold
work, etc.)
• Esteel = 200 GPa
• Strain gauge measurements:
• Wheatstone bridge:

Dependence of elastic stiffness on

interatomic spacing

26
 03/08/2017

Generalised Hooke’s Law

• ; – Shear modulus
Negative shear stress Positive shear stress

Shear diagonal

Different stresses

, , , , ,
Also:
, ,

Direction

Plane (normal to the direction)

27
 03/08/2017

Elastic stress strain relations

• ; ;
• ;
• If you apply then:
•
• ≡

Stress applied Strain (x) Strain (y) Strain (z)

28
 03/08/2017

Resiliency
• The resiliency of a material is a measure of the
amount of energy that can be absorbed under
elastic conditions and which is released
completely when the loads are removed

• .

Material with large energy absorption

(spring)
High yield strength and a low E

Elastic‐Homogeneous Plastic Response:

Type II

29
 03/08/2017

True stress: Empirical relationship

• (Empirical)
• n – strain hardening coefficient
• K – material constant

• n = 1 – ideal elastic
• n = 0 – ideal plastic behaviour

Plastic instability and necking

0.2% Proof stress:

30
 03/08/2017

True strain vs n
Prove that the true plastic strain at necking instability is numerically equal to the
strain‐hardening coefficient
at Pmax

Displacement

Elongation of tensile specimen

31
 03/08/2017

Tensile dimensions

Necking – triaxial stress state

• In addition to the necking strains, a triaxial
stress state exists in the vicinity of the
neck. The radial and transverse stresses
that are induced are developed as a result
of a Poisson effect. In effect, the more
highly stressed materials within the neck
wishes to pull in to accommodate the large
local extensions. Since the material
immediately adjacent to the necked area
experiences a much lower stress level,
these regions will resist such contractions
by exerting induced tensile stresses that
act to retard deepening of the neck.
Consequently, the triaxial stress field acts
to plastically constrain the material from
deforming in the reduced area. To provide
for such plastic flow, the axial stress must
be increased.

32
 03/08/2017

Final fracture appearance

Test temperature vs ductility

33
 03/08/2017

Toughness
• Resistance to fracture
• Large expenditure of energy in the fractur
process
•

Elastic‐Heterogeneous Plastic‐
Homogeneous Plastic Response:
Type III
A

C
B

To avoid it remove the interstitials, nitrogen

and carbon

34
 03/08/2017

Luder’s bands

Strain ageing

35
 03/08/2017

Elastic‐Heterogeneous Plastic
Response:
Type IV

Elastic‐Heterogeneous Plastic
Homogeneous Response – Type V

36
 03/08/2017

Composite Materials Tensile Response

• Variation of stress along a fibre

Normal stress – in the matrix

Stress

Shear stress – in the fibre

Fibre strengthening
• Materials of high strength, and especially high
strength‐to‐weight ratio, can be produced by
incorporating fine fibres in a ductile matrix
• Fibres – high strength and high elastic modulus
• Fibres – alumina, graphite, boron, metal wires
• Might be long and continuous or discontinuous
• Metals and polymers can be used as matrix
materials
• The high modulus fibre carries all the load

37
 03/08/2017

Strength and moduli of composites

• Assume that the matrix and fibre strain equally
• = , therefore:
•
• Where Af and Am are the cross‐sectional areas of
fibre and matrix
• The average composite strength:
• where

Volume fraction
• represent the fraction of total cross‐section
taken up by fibres, multiplied by the length of the
composite (assuming continuous fibres)
• This then represents the volume fraction of fibres
• ‐ volume fraction of fibre
• ‐ volume fraction of matrix
• ‐ volume fraction of composite

38
03/08/2017

39