Recent Trends in Sturctural Integrity Assessment, Eds_: V R. Ranganath , S. Tarafder, A .

Bahadur,
18-19 January 2001, National Metallurgical Laboratory, Jamshedpur, India, pp. 1-17.

Remaining Life Assessment of Engineering Components

R. N. GHOSH
National Meterllurgical Laboratory, Jamshedpur-831 007

ABSTRACT

A structure or a component is expected to perform specific functions for a

minimum specified duration of time. If it ceases to do so, it is said to have
failed. The consequence of service failure can be tragic and expensive. There
are innumerable cases of engineering disasters resulting in loss of life and
property. Therefore, utmost care and attention are required to be given to
critical structures / components to ensure that such incidents do not take place.
Life assessment exercise performed at regular intervals is a means to ensure
absence of such tragic service failure. It gives us an estimate of remaining life
of the component, which is essential to plan future course of action so that it
could be used till the very end of its real life. Recent development of powerful
non-destructive techniques and availability of modern computational facility at
affordable cost has made this a routine exercise for many critical applications.
This paper describes the basic principles behind remaining life assessment of
engineering components and also presents some of our experiences.

INTRODUCTION

Although engineering structures and components are designed to last for a specified
period, premature failure does take place for a variety of reasons viz., design
deficiencies (lack of knowledge), material selection deficiencies, processing
deficiencies, assembly and installation error, operational and maintenance error, and
environmental impact. In addition one must realise that every design has a probability
of failure. This is because both the material property and the loading are random
variables having a mean and a standard deviation. Therefore, in spite of best effort
there would always be a few incidents when the loading may exceed the critical value
or the material used may have inferior property. However, depending on the level of
factor of safety used, the probability of failure is always very low (10-6/10-7), but it
increases with prolonged service.
Old components are likely to be more prone to failure primarily due to the natural
process of ageing and also due to the fact that so far no failure has taken place; a
characteristic of any stochastic process. There are several natural processes of ageing
of structural materials e.g. corrosion, fatigue, wear, creep, shock loading etc. Micro-
structural damages keep on accumulating in the component due to any one or a
combination of the above factors. As a result its load bearing capacity continues to
decrease. Failure takes place when it falls below a critical level determined by the
component geometry and loading. Fig. 1 gives a typical illustration of the nature of

1
X.N. Ghost:

defects that develop at various stages of the life of a creep test specimen subjected to a
given load and temperature . It is evident that even after certain types of defects are
seen to have nucleated , the specimen does still have a significant amount of life. Past
experience has shown that this analogy also holds good for components as well [1-3]
Careful non-destructive evaluation carried out at various stages can indeed help us
extract the full service potential of an expensive component without any compromise
on safety-

fracture
Damage
Action required
parameter
D A. None until next
scheduled outage
$ C B. Replica test at
A
specified inter-vat
C Limited service
until repair
D. Immediate repair
Exposure time

Fig. 1: A schematic diagram showing a relationship between the nature of defects that develop
at various stages of creep strain accumulation and actions required.

The consequence of failure of engineering structures / components can be tragic and

expensive . Therefore, all such cases are critically examined to provide valuable inputs
for future developments . The analysis carried out on Comet aircraft accident is a
classic example [4]. This did provide powerful insight into the actual problem and was
in some way responsible in making air travel one of the safest means of public
transport . Life assessment exercise is however performed much before a failure is
likely to occur . Its main objective is failure prevention . In addition it helps us delay
failure through timely action e.g. if some abnormality is detected, it is repaired before
being put into service again and use the component till the very end of its life. In view
of the ever-growing cost of setting up new plants and equipment , the current trend now
is to refurbish and extend the life of the existing units . Remaining life assessment is
the first major step in this exercise - This paper describes the basic concepts involved in
such Studies.

WHY DO COMPONENTS HAVE REMAINING LIVES?

Design of components always aims at absence of failure- Therefore its size or

dimension is estimated on the basis of the minimum specified material properties,
whereas in reality it may have been made of materials having superior properties.
Consequently re-evaluation would indicate that the component has significant built-in
remaining life.
Life estimation of a component often requires long-term properties, which may not
be available at the time of design. Under such a situation the designer extrapolates
short-term material data to satisfy his requirement. Subsequently, as long term data

2
Remaining life assessment of engineering components

are generated, recalculation may exhibit availability of additional life- Design of power
plant components subjected to creep condition during the sixties, were based mostly on
100,000 hrs stress rupture tests. In view of the importance of long-term creep data in
estimating the life of critical components, many costly test programs involving
multiple laboratories were concurrently undertaken to generate long-term data.
Presently, for many grades of steels, stress rupture data for periods up to 250,000 hrs
are available. Re-evaluation of the life of many such components in the light of the
newly acquired data often suggests that these could be used far beyond the
recommended design life. Past operating experience also supports that service lives
considerably longer than design expectations are usually achieved in practice- In fact a
survey carried out by EPRI, USA indicate that during the mid eighties about 93,000
MW of power was being generated by plants which have been in operation for more
than 50 years [5].

Engineering components are designed on the assumption that they will experience,
to some extent, higher than normal operating conditions, whereas in reality that may
not have happened. Therefore, it is worthwhile to re-estimate the life of an ageing
plantbased on its actual operating data collected over its entire period of service life.
Over the years there has been significant development in the area of non-destructive
evaluation of engineering components. Modem non-destructive testing (MDT) tools
are capable of detecting much smaller defects with greater reliability. Therefore
periodic NDT studies carried out on critical components can help us to use them until
the very end of their useful life.

TECHNIQUES FOR REMAINING LIFE ASSESSMENT

Remaining life assessment technique would depend on the nature of damage or

defects that may have accumulated during service. Careful non-destructive evaluation
is always the first step in any integrity evaluation activity. Modern non-destructive
tools can not only detect fine defects / cracks but also provide an estimate of material
properties. The defect size determines the approach which would be most applicable,
e.g. for steel, if it is less than 1mm (i.e. microscopic), one should use damage
mechanics approach; whereas if it is greater than I mm, fracture mechanics based
method would be more appropriate [6,7].
The basic difference, between the two methods are described in Figs. 2 and 3. In
damage mechanics the failure criterion generally is:

a>ay ...(I)

where a is the applied stress ay is the yield strength. The applied stress being a
function of geometry, size and loading is likely to change with service exposure as
more numbers of defects accumulate. Likewise the yield strength may also change,
particularly in cases where the component is subjected to elevated temperature.
Therefore, life assessment of a component, using this approach would require
development of models that would describe evolution laws, for both cs and ay. The

3
K.N. Ghosh

point of intersection of the two gives the expected life of the component. The
component must therefore be inspected before it reaches this state to avoid service
failure.
Fracture mechanics based approach uses unstable crack extension as the failure
criterion, instead of plastic deformation (or yielding). The basic principle involved is
shown in Fig. 3. The condition for such a crack growth is defined in terms of a
parameter called stress intensity factor (K). When K exceeds a critical value (Kr)
failure takes place [8]. Stress intensity factor K gives the amplitude of the stress
distribution at. the tip of a crack. It is a function of applied .stress (or), crack length (a),
and component geometry. It is commonly expressed as:

K =af(a/WI ...(2)

Failure criterion
o > (Ty
or = }(load, geometry)
6
Y cry = Material property
Cy

Time Life

Fig_ 2 : A schematic showing the principle of life assessment using the

concepts of damage mechanics.

where W is the component thickness through which the crack is likely to propagate and
f(ufW) is a function dependent an the geometry and the type of loading . K,, which is
commonly known as fracture toughness, is a material property . As the crack grows in
service under the influence of the applied stress , K continues to increase. Likewise
depending on the environment to which the component is subjected , KK may also
change . Therefore, here also there is a need to establish laws governing evolution of K
as well as K . The point of intersection of the two is the safe operating life. The
inspection interval for component integrity evaluation should therefore be selected
accordingly.

DAMAGE MECHANICS BASED APPROACH

Service exposure often leaves permanent marks on components being investigated.

These may be in the form of micro - structural / dimensional changes, accumulated
strain, nucleated defects or cracks etc. Many of these can be detected by non-
destructive means. However there are certain other characteristics (e.g. mechanical
properties , extent of embrittlement etc) which can be estimated only by conducting
tests in the laboratory on service exposed materials-

4
Remaining life assessment of engineering components

K,
Failure criterion
K> Kc
K =.load, geometry)
Kc = Material property

Time

Fig. 3: A schematic showing the principle of life assessment using the concepts of
fracture mechanics.
Therefore sometimes it may be necessary to remove a part of the component from
existing units for such studies . Over the years , based on past experience as well as
experimental and theoretical studies, predictive models have been developed . Current
methods for life assessment of ageing components are primarily based on these. Most
extensive work so far has been carried out for components , which experience high
temperature and pressure . The methods used can broadly be classified in two
categories viz. direct and indirect . The former is based on direct estimation of life
from the present condition of the unit , e.g. dimensional / micro- structural changes,
extent of cavitation etc [8 ]. Whereas the latter uses measurements like oxide scale
thickness , hardness , etc. to calculate the average metal wall temperature which, in
conjunction with the operating pressure and the master rupture plot of the material,
gives an estimate of the life [2]. The following examples illustrate the basic approach
being followed at present.

Creep cavitation is a common failure mechanism for heat affected zones (HAZ) in
welded super heater/reheater tubes or steam pipes [ 1,21. Evolution of cavitation could
be followed by in - situ metallographic examination using plastic replication technique
or direct examination under a portable microscope during the shut down . Fig. 1 shows
the nature and the distribution of cavities at various stages of the life of a component.
Based on past experience and laboratory tests a number of thumb rules have been
devised for life estimation . For example the detection of macro cracks formed due to
coalescence of cavities would require immediate repair and /or replacement of the
component . Established empirical models are also used to arrive at a more direct
estimate . One of these relates a cavitation parameter (A), defined as the fraction of
grain boundaries containing cavities , with the life fraction used up ( t/tr) as follows
[1,21:

A=0.5( tt")-0.095 ...(3)

Using the concept of damage mechanics and laws governing the evolution of creep
strain , a simple expression relating remaining life fraction (RLF) with the accumulated
strain can be derived. One of these is as follows:

5
(ihosh

ln(RLF)= 1nr1- Xt')= ( EX)ln(I - eer) ...(4)

Where, t, is the rupture life, cis rupture strain (--10%), E a material constant (-3%), r
is accumulated creep strain at a time t [S]. The above expression could directly be
used to calculate the remaining life from strain measurement. A simple calculation
would show that to attain an accuracy of 10% in prediction life strain measurement
should be within 0.01%. The only way this could be attained is to weld markers made
of heat resistant steel at critical locations on selected components and continue to
measure the distance between these at every shut down.

The above illustrations are examples of direct estimation of remaining life.

However, in certain cases it may not be possible to get a correct estimate of the metal
wall temperature the component has seen in its life because of a variety of reasons.
For example, due to load fluctuation and oxide scale growth, metal wall temperature of
reheater / superheater tubes or steam pipes in a power plant boiler, is unlikely to
remain constant. In such cases indirect methods are used to get an approximate
estimate of the same from a set of simple non-destructive tests. Micro-structural
examination, hardness and oxide scale thickness measurements provide valuable inputs
for such studies. Each one of these is a function of both time and temperature of
exposure. Master plots are now available for describing the expected variations in
microstructure, hardness or scale thickness as functions of both time and temperature.
Using these and the total length of service, the average metal wall temperature can be
calculated. Subsequently on the basis of the operating pressure or the stress the unit is
likely to experience in future, the remaining life can be estimated from the master
rupture plot of the material the component is made of [5].

Methods described above essentially depend on information collected from non-

destructive tests. In view of the importance of life assessment of ageing high
temperature components extensive destructive tests (e.g. creep) have been carried out
on samples removed from components such as steam pipes, turbine discs etc. Analysis
of these data indicates that matching of the shape of the creep curves of a service
exposed material with that of the virgin could be quite dependable. There are two
basic approaches. The one developed by Wilshire et al. [9], commonly known as
0 projection, is purely empirical in nature. It uses a specific algebraic equation to
represent creep strain as function of time. The material constants are evaluated from
experimental data on the virgin material and their dependence on stress and
temperature is also established experimentally. Material constants are now available
for several grades of low alloy steels used in power plants. Using these, creep curves
at any stress and temperature can be generated. Fig. 4 indicates a typical plot at a
relatively higher temperature (Ta,,,,) and stress (cr p) so that it takes reasonably short
time to conduct the test. On this an actual experimental plot obtained from the service
exposed material has been superimposed. The creep strain from which the matching
appears to be satisfactory is a measure of the damage that has accumulated in the
material because of service exposure. This is projected on to the estimated creep curve

6
Remaining life assessment of engineering components

under the actual operating condition (or T) as shown in the Fig. 4 to estimate the
remaining life.

Time
Fig. 4 A schematic diagram showing the principle of remaining life assessment based
on creep test performed on a service exposed material-

The method suggested by the high temperature materials group at NPL, Teddington
uses a set of coupled differential equations to represent simultaneous evolution of
creep strain and microstructural damages. These have been built on our current
understanding of the underlying principle of high temperature deformation. The
procedure has been implemented in the form of a computer based design aid called
CRISPEN [10]_ The approach being model based, it has been possible to extend its
scope to describe behaviour of anisotropic material [I I], the effect of complex loading
conditions [12] and the influence of environmental interaction [13]. Even though in
principle this approach could also be applied to estimating remaining life of
engineering components, subjected to creep, it has so far not been explored as
extensively as the 0 projection. One of the reasons is that the present form of
CRISPEN primarily deals with materials that exhibit strain softening whereas most
engineering structures are made of steel whose creep behaviour is best described by
time softening model [14].

FRACTURE MECHANICS BASED APPROACH

Unlike damage mechanics , here a specific macroscopic defect of known size and
geometry determines the life of a component. Therefore its growth characteristics
under the given condition of loading is required to be followed. Under sustained
loading in neutral environment, unstable crack growth takes place only if the
dimension of a pre-existing defect is greater than a, (critical crack length). For a given
loading condition a, is estimated by substituting the value of material toughness in
place of K in eqn. (2).
The major challenge before NDE (non-destructive evaluation) is therefore to ensure
that any crack exceeding ac, does not go undetected. The choice of NDE technique to

7
be employed would depend on material toughness. Lower the toughness, shorter is the
critical crack length. Such material would therefore demand for more sensitive
equipment to be employed. Past experiences reveal that prolonged service exposure
leads to embrittlement or loss of toughness. Therefore old/ageing components are
required to be evaluated more carefully.

Subcritical cracks in a component can grow in service either due to fatigue loading
or due to prolonged exposure to corrosive environment. Figs. 5 and 6 show crack
growth behaviour in such cases. Under fatigue loading this is best described in terms
of a parameter called stress intensity factor range (AK) defined as the difference
between K,,,u, and K,,,;,,; the maximum and the minimum stress intensity factors
respectively [7].

Stage I Stage II
infinite life finite life

Stage III
A K ,h
little life

log (A K)

Fig. 5 An approach to fracture mechanics based life assessment for components subjected to
fatigue loading

When AK< AK,h, a threshold value, the crack growth rate is so low that for all
practical purposes the component can be said to have an infinite life (stage I).
Similarly when K,,,,, approaches KK the crack growth rate is so high that the life is very
limited (stage 111). Meaningful life estimation is possible only during the second stage
where the crack growth law could be described as:
da
dN = A(AK)n ...(5)

where N is the number of cycles, A and n are material parameters. Since AK is a

function of crack length, a numerical integration scheme is required to estimate life. In
cases where amplitude of loading varies, integration should be performed in units of
sub-blocks consisting amplitude loading [7,15).
Life estimation procedure under sustained loading is also very much the same.
Crack growth law during the second stage in this case is given by:

da = aK .^
...(6)
dt

8
Remaining life assessment of engineering components

where A and n are material constants. The time it takes for the crack to grow from a to
ac is the useful life of the component. K being a function of crack length here as well,
numerical integration will be required . As in fatigue here also there is a threshold
stress intensity factor , Ktscc such that when K <Ktscc there is no crack growth and the
component has an infinite life . Likewise in stage III as K approaches KK there is little
life left (Fig. 6)

log (K)
Fig. 6 An approach to FM based life assessment for components subjected to sustained
loading in corrosive environment.

The above approach , often called LEFM (linear elastic fracture mechanics ), would
strictly be applicable to materials having relatively low toughness [7,16]. The crack
tips of the defects in tougher materials become blunt and cracks do not propagate in an
unstable manner. This provides them ability to withstand higher loads . LEFM based
life prediction for such materials is therefore likely to be highly conservative.
Alternate fracture criteria viz, J.JJ [7,17]; 3:88 [ 18] could be used for such classes of
materials . These have been derived using elasto-plastic analysis. Likewise , for time
dependent crack growth analysis , parameters like C and C * are thought to be more
appropriate [ t,191. The derivation of the expressions to be used to estimate the
appropriate fracture parameters is often quite complex - It may require the use of high
speed computers . This is the reason why the application of fracture mechanics based
concept for integrity assessment of engineering components is of rather recent origin.
Application of the principles of fracture mechanics in life assessment is quite simple,
provided solutions for the stress intensity factor (SIF) or other appropriate fracture
parameters for cracked components are available . Amongst the various methods for
the evaluation of K for cracked components, finite element method (FEM) has become
the most popular . With the availability of high speed computational facility as well as
commercial FEM packages for stress analysis it is now quite simple to arrive at
expressions for K for a variety of component , and crack geometry . Handbooks for SIF
provide such expressions for a variety of components under different loading
conditions [ 20]. Therefore this approach is now gaining popularity and has been used
to solve many life assessment problems , e.g. this approach has been applied for the life
prediction of boiler drums exhibiting ligament cracking [ I].

9
R.N. Ghosh

ROLE OF NON-DESTRUCTIVE TESTING

In the light of the above background, it is now possible to illustrate the role that
NDT is expected to play. This has been presented in the form of a schematic diagram
(Fig. 7). There are broadly two kinds of expectation viz. Current physical dimensions
and material properties. The former helps us estimate the stress (or SIF) the
component is likely to experience and the latter provides a value of material strength
(or toughness) to test whether it is still safe to use. Varieties of simple, easy to use
tools are now available for the measurement of physical dimensions, e.g. ultrasonic
gauges are regularly used for thickness measurement of pressure vessels. Service
exposure leads to loss of section thickness due to interaction with en"-ironment.
Therefore old vessels are likely to experience higher stresses which can be estimated
from such measurements. However for life assessment we need to compare it with the
current material strength which may have changed because of service induced
embrittlement. This indeed is a difficult task. In the absence of suitable non-
destructive techniques the practice currently followed is to carry out WFMPI (wet
fluorescent magnetic particle inspection) on the surface which is exposed to the
corrosive environment. Detection of a crack is considered to be a sign of embrittlment.
The decision to repair or retire the vessel is based primarily on the nature and the
extent of the defects. Sometimes an acoustic emission test may also be conducted to
check whether the defects are likely to grow during service. However such a test
might be very expensive, as it needs the use of a costly microprocessor based
equipment available only in a few organisations in India. Nevertheless non-destructive
evaluation of material property still remains a major problem to be solved for reliable
life assessment of ageing components.

Non-destructive Case History

Testing

Remaining Life
Assessment (RLA)

Service Material 'Virgin

Exposed Database

Fig- 7: Role of non-destructive testing in remaining life assessment

In certain areas some novel techniques are being attempted to estimate the material
property. Life assessment of boiler tubes based on oxide scale thickness measurement
using special ultrasonic probes is an example worth looking at. The thickness of the
oxide layer (x) that develops on the inner wall of a boiler tube depends both on the
metal wall temperature (T) and the length of service exposure (t). Several expressions

10
Remaining life assessment of engineering components

describing the relationship between these are now available . One of these is given
below:

log(x) = A + B(C + log I )T --- (7)

where A, B and C are material constants . These have been evaluated for several grades
of Cr-Mo steels [1,5]. Using the above equation , therefore , an estimate of the average
metal wall temperature can be obtained. Subsequently master rupture plot of the
material can be used for life estimation at any given operating stress.
The NDT ( non-destructive test ) technique to be used for life assessment is often
determined by the most likely mode of failure . Steam pipes , for example, are prone to
failure due to creep , which is a time dependent deformation .. The defects usually
nucleate at the outer surface since it is subjected to a higher stress and they propagate
inwards . Thus the surface where defects could be seen is easily accessible. In-situ
metallography is therefore the best technique to be used . It can detect defects much
before they attain the critical size in comparison to other methods 1.

Likewise, the limitations of current NDT techniques to detect cracks/defects may

lead to premature failure of components . Fatigue failure of Comet aircraft is an
excellent example . Later analysis has shown that the critical crack size for the material
of which the aircraft was made was smaller than the diameter of the bolt head [7].
Therefore, visual examination could not detect them . Development of materials
having higher critical crack length ( fracture toughness ) and better joining techniques
were needed to overcome such failures-

CASE STUDIES

Ever since the establishment of creep testing facility at the National Metallurgical
Laboratory (NML) in 1974, a number of its scientists had the opportunity to look at
various problems related to life assessment of power plant components- The major
emphasis initially has been on components subjected to high temperature and pressure.
In view of the growing importance of the topic, a multi -sponsor thrust area project
called Component Integrity Evaluation Programme (CIEP) was initiated primarily to
augment facility and expertise to undertake such work on a much larger scale. The
first phase of this programme has just been completed . A few of the recent case
studies included would illustrate how the above principles could be applied to solve
real life problems. The methods described are quite varied and component specific. In
certain cases the computation can be time taking as well . A number of computer
software has been developed to assist such work . These are being used for component
integrity evaluation activity at NML.

Steam pipes
Steam pipes of power plants are designed on the basis of 100 ,000 hr rupture strength
data. Therefore, as these approach this critical value, it may be necessary to evaluate
whether they are still safe for use . We had the opportunity to generate stress rupture
data on several service exposed steam pipes made of a CrMoV steel. Tests were

11
R.N. Ghosh

conducted in air environment at either higher stress or temperature to cut down the
testing time . Temperature accelerated tests are supposed to provide a more realistic
estimate of life. A simple comparison of the stress rupture data with those obtained
from virgin pipes did not indicate any loss of strength . Rather it appeared that service
exposed material may be better than the virgin . The above conclusion certainly looked
unrealistic . Later, a more rigorous analysis was carried out using a computer software
(CLIP), that provides a frame work for Creep Life Prediction of engineering materials
with the help of three commonly used time temperature parameters viz., Larson-Miller
(LMP), Sherby-Dorn (SDP) and Manson.Haferd (MHP) parameters [21]. Table 1
gives a summary of the analysis. Rupture life at 550°C / 64 MPa and 100,000 hr,
rupture strength at 550°C were estimated for virgin as well as service exposed pipes.
Table 1: Remaining life and 100,000 hr rupture strength of steam pipe material at 550° C as
function of.service exposure.
Service Remaining life (hr) Rupture Strength {MPa)
Exposure (hr) LMP SUP MHP LMP SDP MHP
0 (Virgin) 161000 164000 98000 97 97 102
56,000 184000 199000 105000 73 73 67
100,000 91 000 97 000 89 000 68 67 60

It is revealed that while the rupture strength decreases with time, the estimates of
remaining life has no correlation with service exposure. This is primarily because in
case of the latter an attempt has been made to extrapolate the data beyond its domain
of validity. Therefore it is important to know the limits of extrapolation. It would be
more rational to use rupture strength as the criterion for integrity assessment rather
than to estimate its remaining life [22],

A more detailed analysis did show that prolonged service exposure had changed the
microstructure significantly [23] and it has a definite correlation with the minimum
creep rate. This has been established through the measurement of a parameter called
threshold stress whose magnitude, in relation to the applied stress may give an idea of
the condition of the pipes. It has been reported that it decreases with service exposure
like the rupture strength [24]. This too supports the above observation.

Heater tubes
Heater tubes are used in coking/visbreaking units of petroleum refineries to raise the
temperature of reduced crude oil (RCO) to 500 C. These are made of Cr-Mo steel and
are required to withstand high temperature and pressure. During the heating stage,
coke deposition takes place on the inner wall of the tube. Consequently the metal wall
temperature is required to be raised to maintain the same level of heat flux to bring
down the viscosity of RCO. Besides, the reduced cross sectional area due to coke
deposition would demand higher pressure drop to maintain the flow rate. Therefore, it
is of utmost importance to monitor skin temperature and pressure drop across the tube
length, Whenever these go beyond a specified limit, the unit is shut down for cleaning.
Usual length of a normal coking cycle varies between 150-240 days. During shut
down, the furnace tubes are cleaned either by mechanical means with the help of a

12
Remaining life assessment of engineering components

mandrel or by steam air decoking . Since the latter process is more efficient and
provides cleaner surface, it is being practised in most of the units. Necessary steps are
taken to ensure that the temperature does not go beyond 700°C during the decoking
cycle [25].
These tubes are designed based on their 100,000 hr rupture strength. Many of these
are in use for the past 20-25 years. Few of these were removed from service to carry
out a detailed life assessment exercise. The study revealed that although no significant
change in the strength of these tubes were detected there was a considerable loss of
toughness and rupture ductility as a result of prolonged exposure. Carburization at the
inner wall of the tube where coke deposition takes place during normal operation was
identified as the main cause of this degradation [25]. This is also likely to impair
weldability of tubes making weld repair more difficult and susceptible to cracking.
Therefore it is of considerable relevance to evolve a strategy of tube replacement based
on material degradation so far detected. Since there is an evidence of embrittlement, a
fracture mechanism based integrity assessment was thought to be more appropriate.
Fig. 8 gives an idea of the kinetics of the growth of the carburized layer. This layer is
expected to be brittle and the unit is likely to experience thermal shock during the
normal coking and decoking cycle. Therefore it may be justified to assume that cracks
of length equal to the depth of carburization do exist in the component. Consequently
it is expected that an integrity assessment exercise would provide an estimate of the
critical crack length and state when it is likely to cause unstable crack extension in the
component.
The major problem has been to get an idea of the fracture toughness of the material.
The only estimate that could be obtained was Charpy V-Notch (CVN) value from tests
performed on sub-size specimen because of limited thickness of the tube. Therefore
the use of standard expression available in the literature to convert the same to K1c [1]
is the only alternative. With the assumption that such relationships are valid, here also
the LEFM approach could be used to generate plots of the type given in Fig. 3. The
effect of service exposure on the CVN value for the heater tube material is given in
Fig. 9. Using the data presented in Figs. 8 and 9 and the eq. (2), a plot of the type
given in Fig. 3 has been generated. This is given in Fig_ 10. This gives us an idea
about the depth of carburization that can be tolerated. Failure due to poor toughness of
the service-exposed tube may take place only during the mandatory hydrotests
performed just before putting these, back to service after steam air decoking. Life
assessment study provides valuable input to take correct decision to avoid failure at
this critical stage.

CONCLUDING REMARKS

A broad overview of two distinct approaches for remaining life assessment of

engineering components has been presented. A complete life assessment exercise on
any equipment is no doubt going to be expensive and time taking. Therefore, for
convenience the entire exercise is divided into three levels [1]. Level I, looks at the
past operating conditions in the light of the minimum specified material property data

13
R,N. Ghosh

currently available. As mentioned earlier, often the design is based on a relatively

short-term material database. Reassessment may therefore exhibit additional
remaining life.

5
O experimental 9
4
Ocalculated
31 8
2

0 0
0 1 2 3 4
t exp (- QIRT)

Fig.8: Depth of carburization expressed as function of temperature (T) compensated

time (t). Q is the activation energy of the process and R is the universal gas
constant

100

0 1 2 3 4

t exp (-CYR1)
Fig. 9: The observed Charpy impact value ( CVN) expressed as function of temperature
(T) compensated time (t). Q is the activation energy of the process and R is the
universal gas constant.
Level II assessment procedure is based on non-destructive measurements carried out
on components during major shut downs. A large variety of techniques are now
available to interpret these data. Such an exercise is usually taken up only when level I
analysis shows that more than 70-80% of the life has been used up_

Level III analysis is based on post exposure destructive test data obtained from
samples removed from critical locations of the component. Such an exercise is
recommended when level II studies reveal presence of certain categories of defects or
provide indications of sufficient loss of material strength and toughness. Here the life
of components are reassessed based on the short term properties collected from
laboratory tests performed on service exposed materials.

14
Remaining life assessment of engineering components

0 2 4 6 8 10

Crack length (a), mm

Fig. 10: Stress intensity factor (Ki) and toughness (K,c) as function of crack length
Based on the work so far carried out, excellent guidelines to characterise flaws and
their acceptance criteria in structures and components are now available [26,27].
Current practice for life assessment of engineering components, are primarily based on
these. Although most of the information required may be easy to obtain using suitable
non-destructive tools, some will be difficult to get. One such property is material
toughness. It is known to deteriorate with service exposure. This has been established
from tests carried out on service exposed material. Mostly these are either simulated
laboratory tests or studies conducted on miniature test samples removed from actual
components. However, such data are very limited. At present there is no established
non-destructive technique, which could give an idea about the toughness of a material.
In addition , often due to the limitation of the size and geometry of the component, it
becomes difficult to conduct valid fracture mechanics tests on service exposed
material. Keeping in view the importance of cylindrical pressure vessels and piping
there is need to explore the possibility of using are bend type specimens [28] for
testing and derive expressions for SIFs [29) for various crack geometry so that more
reliable fracture mechanics based integrity assessment can be carried out.
Concurrently, more efforts are required for non-destructive detection, sizing and
classification of defects. Recent work has shown that the measurement of certain
magnetic characteristics of service exposed material can provide information on
structural degradation of components made of steel [30]. Certainly there are many
unexplored possibilities for non-destructive testing. Clearly these are the areas where
more attention is required to fully exploit the capabilities of critical structures and
components.

ACKNOWLEDGEMENT

The author would like to express his gratitude to all the team members of the CIEP,
its sponsors and Shri K P Prabhakaran for his help in preparing the manuscript. He
would also like to thank Prof. P. Ramachandra Rao, Director, NML for his kind
permission to publish this work.

15
R. N. Ghosh

REFERENCES

1. R. Viswanathan, Damage Mechanisms and Life Assessment of High Temperature

Components, ASM Int. Metals Park, Ohio ( 1989).
2. R. Viswanathan and G.M. Gehl, J of Metals, 44 (1992), 34
3. G.A. Webster and R.A_ Ainsworth, High Temperature Component Life
Assessment, Chapman & Hall, London (1994)
4. D.R.H. Jones, Engineering Materials , Vol. 3, Pergamon Press, (1993)
5. S.R. Paterson and T.W. Rettig, Conf. Procs., Life Extension and Assessment of
Fossil Power Plants, (Eds. R. B. Dooley and R. Viswanathan), EPRI, Palo Alto,
1987
6. J. L. Chaboche, High Temp. Tech. 5 (1987), 59
7. R.W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials,.
John Wiley, New York, (1983)
8. B.J. Cane and J.A. Williams, Int. Mater. Rev., 32 (1987), 241
9. R.W. Evans, J.D. Parker , and B . Wilshire, Int. J. Pres. Vess. and Piping, 50
(1992), 147.
10. B.F. Dyson and M McLean, ISIJ Int, 30 (1990), 802.
It- R. N. Ghosh and M. McLean, Acta Metall and Mater, 40 (1992), 3075.
12. R. N. Ghosh, R.V. Curtis and M. McLean, Acta Metall, and Mater.. 38 (1990).
1977.
13. N. Roy and R. N. Ghosh, Scripta Mater, 36 (1997), 1367.
14. S. Chaudhuri, N. Roy and R . N. Ghosh, Acta Metall. and Mater, 41 (1993), 273.
15. ASTM Stadards E-647, Vol. 03.01 (1989).
16. ASTM Standards E-399, Vol_ 03-01 (1989).
17. ASTM Standards E-913, Vol. 03-01 (1989).
18. British Standards Institution, BS-5762, (1979).
19. P.K. Liaw, A. Saxena and J. Schaefer, Engg. Fracture Mechanics, 32 (1989), 675
and 704.
20. Y. Murakami (ed.), SIF Handbook, Vol. 1 & 2, Pergamon Press (1987).
21. R. N. Ghosh, S. Chaudhuri, N. Roy and L. Chattopadhayay, Sadhana, 20 (1995)
331
22. L. Chattopadhyay and R .N. Ghosh, Engg. Fracture Mechanics, 54 (1996), 71.
23. R. Singh and S. Banerjee , Mat. Sci. and Engg. 132 (1991), 203.
24. R. Singh and S . Banerjee , Acta. Metall. and Mater, 40 (1992 ), 2607.
25. R. N. {Gosh et at., Remaining Life Assessment of Process Heater Tubes of
Petroleum Refineries, IOC/NML Interactive Project Report, ( 1994).

16
Remaining life assessment of engineering components

26. British Standards Institution, PD-6493, (1991).

27. L_ Milne, R. A. Ainsworth, A.R. Dowling and A.T_ Stewart, Init. J Pres. Vess. &
Piping, 32 (1988), 73.
28. S. Tarafder, M. Tarafder and V.R. Ranganath, J. of Testing and Evaluation,
ASTM, 24 (1996), 73.
29. S. Tarafder, M. Tarafder, V.R. Ranganath and R. N. Ghosh, Int. J. Pres. Vess. &
Piping, 70(1997). 127-
30. D.K. Bhattacharya, J. Non-Destructive Evaluation, 17 (1997), 1.

17