OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.

uk

Damage and remaining life estimation in high temperature plant with

variable operating conditions

H. C. Furtado1, L.H. de Almeida2 and I. Le May3

1
CEPEL, Rio de Janeiro, Brazil, 2COPPE, Universidade Federal do Rio de Janeiro, Brazil, 3Metallurgical
Consulting Services, Saskatoon, Canada

Dr Heloisa Cunha Furtado graduated from the Federal University of Rio de

Janeiro in Metallurgical and Materials Engineering, later obtaining M.Sc. and
D.Sc. degrees from the same university, dealing with materials for high
temperature steam generators. Since 1986, she has worked at CEPEL (Electric
Power Research Center), being responsible for the High Temperature Materials
Area. For more than 15 years, she has been responsible for structural integrity
evaluations of Brazilian thermal power plants.
Dr Iain Le May attended the Royal College of Science & Technology, Glasgow,
and the University of Glasgow, Scotland. He was on the faculty of the University
of Saskatchewan, from 1963 to 1985, being Professor of Mechanical
Engineering from 1970 to 1985. He established Metallurgical Consulting
Services Ltd., Saskatoon, in 1978; and MCS Associates, Inc., Pittsburgh, in
1991. He has more than 300 publications, including several books. He is an
Adjunct Professor in the post-graduate program of Dept. of Materials and
Metallurgical Eng., Federal University of Rio de Janeiro (COPPE/UFRJ),
Brazil. He has consulted extensively in Canada and other countries and has
appeared as an expert witness on many occasions. Dr Le May is a Fellow of
ASM, ASME, and the Royal Society of Arts, London (FRSA), and serves on many
society and international committees. In 2007 he was elected as a Member of the
Brazilian Academy of Sciences.

Dr Luiz Henrique de Almeida has been in UFRJ since 1976, Professor of

COPPE since 1997 (head of department 1988 - 1992, Lab. of Mechanical
Properties 1979 80; 1985 - 88 and 1996 to present, Electron Microscopy
Lab. 1988; 1992- 95 and 1998 to present..); Visiting Researcher, University
of Saskatchewan 1983 - 84, Visiting Professor University of British
Columbia 1987; Sub-secretary of Science, Technology and Innovation of
the Rio de Janeiro State 2000 - 02 and 2004- 06. CNPq Researcher 1-B.
CNPq Advisory Committee for Mine, Metallurgy e Materials Engineering
1992 and 1996/98; COPPETEC consultant for metallurgical problems and
mechanical failures; Awards: Luiz Dumont Villares from ABM 1988, Best
Paper Award (IMS/ASM 1994), Annual Buehler Technical Paper Merit
Award for Excellence, IMS, 1999; Teaching: Physical and Mechanical
Metallurgy, Materials Science and Electron Microscopy; Research:
Physical metallurgy and mechanical properties of steels and non-ferrous
alloys, analysis of structural integrity, development of heat resistant steels
and CrW ferritic steels for petrochemical industry; He has supervised 31
theses and 19 graduation projects; 140 Publications (33 in International
Journals), 1 patent.
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

Abstract

Estimating damage and the remaining life in power plant or high temperature pressurized
petrochemical equipment operating under essentially steady state conditions has become
relatively routine. The relations governing creep damage under such conditions are
relatively straightforward and well known. Life can be estimated based on parametric
methods for the particular conditions of temperature and stress and for the specific
materials involved. With varying periods of operation the accumulated damage can be
computed by addition of the damage occurring in the various operating intervals.
However, the situation is much less clear when there are many shutdowns and the
operating conditions vary considerably with time. Similarly, when there are local hot
spots caused by uneven flame temperature or other factors, there may be areas in which
damage is concentrated.

Thus it is necessary to utilize inspection techniques to survey the extent of damage,

although these may not always provide clear estimates of remaining life, being dependent
on judgment on the part of the inspector. Clearly one method is to remove samples of the
material and to conduct accelerated creep testing, but this is both expensive and time
consuming and the number of points evaluated is necessarily limited.

An alternative assessment method is to use measurements of hardness that can be

correlated with creep damage and remaining life, and this is demonstrated in the paper,
previous work having shown such correlations to be valid [1]. The hardness
measurements can also be correlated with non-destructive microstructural observations
made at the same time.

In the paper, a study is presented that has been made of specimens of 9Cr-1Mo steel from
accelerated creep tests made under different conditions of temperature and stress. The
results show that the measured hardness values recorded correlate well using the Larson-
Miller parameter, and that they can, in turn be used to approximately estimate the degree
of damage and the remaining life of the material.

The utility of the approach is that it can be applied to high temperature plant operating
under complex time/temperature/stress cycles, that it is non-destructive and that
evaluations can be made in a limited time with a minimum of disruption to plant
operation.
Key words: Hardness, Larson-Miller Parameter, Integrity Evaluation

1. Introduction
The principal degradation mechanism of steels in high temperature applications is creep,
which is time-dependent, thermally assisted deformation of components under stress.
Creep of Cr-Mo ferritic steels initially involves the spheroidization and structural
transformation of carbide precipitates and, at an advanced stage, results in the formation
of voids and microcracks. Since life expectancy of components that work at high
temperatures is, in most cases, based on the capacity of the material to maintain its creep
resistance at high temperature for a period of at least twice the design life, methods to

2
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

evaluate the creep resistance of the material based on physical changes taking place in it
as well as robust methods for data extrapolation are necessary

Nearly half a century has elapsed since the introduction of time and temperature
parameters for the correlation and extrapolation of creep data. The first work was done by
Larson and Miller [1], who introduced the concept of a parametric function to include
both stress and temperature as variables. This has the form:

=f(,t)T(C + log t) (1)

where is the stress, t the time for rupture and T the absolute temperature and C is a
constant. Larson and Miller developed a method to unify a family of lines showing
rupture time as a function of temperature at different stress levels such that a single curve
is obtained, when log(rupture time) is plotted as a function of the Larson-Miller
parameter. When this master curve is obtained, it can be used to estimate time to
rupture at different values of stress and temperature within the limits set by the
experimental data. This approach has been used to obtain interpolated values of time to
rupture, as well as extrapolated values for times significantly greater than those for which
experimental data are available.

Hardness measurements give an indication of a materials resistance to deformation and

can be used to estimate the extent of damage undergone by a material exposed to
temperature and stress for a long time. In this way, hardness has been used as a simple
mechanical, non-destructive test method to estimate loss of mechanical resistance and
creep resistance. Although empirical, it can be a useful complement to metallographic
methods of evaluation. Efforts have been made to correlate hardness measurements with
creep resistance [2], mainly on low alloy steels such as those used in power plants, where
loss of resistance to deformation occurs through morphological changes to carbides
present in the matrix of the steel.
The process of recovery of a strain-hardened or quench-hardened steel can also be
described in terms of a time-temperature parameter, namely the Hollomon-Jaffee
parameter [3], which is equivalent to the Larson-Miller parameter used for creep. A
relationship between the Larson-Miller parameter, P = (C + logtf), and the changes in
hardness can be developed: if the initial hardness (H0) at time t = 0 is known, the material
can be aged at temperature T, and from the values of H obtained at different times, the
relation H = f(P) can be determined for the particular material. For field measurements of
hardness of components that have been in service for long periods of time, the effective
or main value of temperature to which they were exposed can be estimated if the
quantities P and H0 are known. As in most cases H0 is unknown, it can be estimated
through known data for similar materials. Another application for hardness measurements
is to relate them to microstructural observations. Locally degraded regions will have a
lower hardness, and such measurements can supply additional useful indications of the
material damage caused by creep.
The object of the present study was to examine the correlation between hardness
measured in creep specimens deformed under different conditions and the Larson-Miller
parameter. The study was made on 9Cr-1Mo P91 steel. Accelerated creep tests were
made, and subsequently the test specimens were analyzed by means of hardness tests, and
characterized by light microscopy, SEM and TEM.

3
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

2. Material and methods

The material studied was P91 9Cr-1Mo ferritic steel, whose chemical analysis is given in
Table 1.
Table 1-Chemical composition of 9Cr-1Mo steel (% by weight)

Element Nominal [4] Analysis

C 0.06-0.15 0.12
Mn 0.25-0.66 0.52
P 0.025 max 0.008
S 0.012 max 0.005
Cu - 0.16
Si 0.18-0.56 0.33
Ni 0.43 max 0.20
Cr 7.90-9.60 8.57
Mo 0.80-1.10 0.92
V 0.16-0.27 0.20
Al <0.05 < 0.01
Nb 0.05-0.11 0.085
N 0.025-0.080 0.046

Tensile and hardness tests were made at room temperature on specimens from the as-
received plate. The results are presented in Table 2. According to the steel manufacturer
the 9Cr-1Mo plate was normalized at 1065C and quenched at 787C. The hardness
measurements are within the range expected from the nominal specification of the
material.

Table 2 Mechanical properties of steel ASTM A387 Gr91

Nominal
Mechanical Properties Results
[4]
Tensile strength (MPa) 585-760 691+6
0,2%Yield Stress (MPa) >415 539+9
Elongation (%) >18 22.6+0,4
Reduction of Area(%) - 71+2
Hardness Longitudinal - 226+1
(HV) Transverse - 229+1

Accelerated creep tests were made on specimens machined according to ASTM E8M [5],
as shown in Figure 1.

Figure 1 Reduced conventional test specimen for creep test

Table 3 shows the chosen conditions for creep tests on the 9Cr-1Mo steel plate. All tests
were of creep rupture type. Vickers hardness was measured with a WPM Vickers/Brinell
machine with a 40kg load. Hardness tests rather than microhardness tests were preferred

4
 OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

as they have the advantage of acting over a greater area and, consequently, the influence
of local effects of from heterogeneous distribution of carbides in the microstructure is
less significant. On the other hand, owing to the greater dimensions of the hardness
impression, measurements were taken only in the central part of the test specimens.

Table 3 Conditions of creep tests realized on 9Cr-1Mo steel sheet.

Test Specimen Test temperature (oC) Stress (MPa)

01 566 230
02 566 210
03 566 190
04 579 190
05 593 160
06 593 175
07 593 190
08 593 210
09 607 145
10 607 160
11 607 190
12 621 135
13 621 145
14 621 160
15 621 175
16 621 210
17 649 125
18 649 135
19 649 135
20 649 145
21 649 160
22 649 160

3. Results
Table 4 presents creep and hardness results on reduced test specimens. The hardness
measurements were made adjacent to the creep fractures. Generally, six readings were
taken from each test specimen

Table 4. Creep and hardness test results for 9Cr-1Mo steel.

Stress (MPa) Temperature (oC) 566 579 593 607 621 649
Time to Rupture (h) 100.43
125
Hardness(HV) 200+2
Time to Rupture (h) 744.14 45.88
135
Hardness(HV) 201.3+0,6 204+1
Time to Rupture (h) 961.54 177.39 18.14
145
Hardness(HV) 205+2 212+2 213+2
Time to Rupture (h) 1,448.06 299.37 45,35 6.68
160
Hardness(HV) 204+4 210+13 210+2 214+5
Time to Rupture (h) 502.94 19.31
175
Hardness(HV) 207+2 215+7
Time to Rupture (h) 5,127 515.68 129.04 21.12
190
Hardness (HV) 206+3 207+4 213+3 221+3
Time to Rupture (h) 1,093.6 26.24 1.93
210
Hardness(HV) 209.7+0,6 219+10 226+4
Time to Rupture (h) 167.09
230
Hardness(HV) 219+3

5
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

4. Discussion

When Larson and Miller developed their parameter, they used the value of C=20 because
that was the value that fitted best for the steel they were studying. Subsequently, it
became common to use this value for Cr-Mo steels, since it is an adequate approximation
for most steels that are used at high temperature. However the use of an optimized value
for C for each material results in more adequate predictions.

Extrapolations carried out from creep rupture data using several established parametric
methods have shown that correlations between predicted and experimental rupture times
on the parametric method used. The analysis made of the present data showed that the
best correlation was obtained using the Larson-Miller parameter with C = 35.73 [8]. This
value was obtained from optimization using the Manson and Medelson [6] method of
least squares for the experimental data. Figure 2 shows the master curve for Larson-
Miller parameter with C = 35.73 drawn using the equation of Spera [7], as described
elsewhere [8]. The axis is G, a function of stress and defined according to Spera [7] and
Manson and Ensign [9].

Figure 2 - Master curve for Larson-Miller parameter with C = 35,73.

Figure 3 presents isothermal plots of time to rupture obtained from the experimental data
using the parameter of Larson-Miller with C=35.73.
log(stress, MPa)

log(rupture time, h)
Figure 3 Logarithmic curves of stress versus time to rupture for experimental data based on the
Larson-Miller parameter with C= 35.73.

6
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

Hardness clearly decreases as the duration of the creep test increases as may be seen in
Table 4. Figure 4 shows the data of Table 4 for hardness in graphical form as a function
of stress and rupture time. The graph shows that for a given time to rupture, the higher
the stress, the greater the hardness.

Hardness, HV

Log(time to rupture, h)

Figure 4 - Isostress lines for the hardness of 9Cr-1Mo steel tested to rupture at different
stress levels.

Figure 5 shows the hardness data from Table 4 plotted as a function of rupture time and
temperature. As may be seen, the hardness at the time of rupture is related to three
variables: stress, time and temperature.
Hardness, HV

Log(time to rupture, h)

Figure 5 - Isothermal lines of hardness for 9Cr-1Mo steel tested at different temperatures.

As already discussed, one of the first works relating to time-temperature parameters was
performed in 1943 by Hollomon and Jaffee [3], in which these two variables were related
to the changes in hardness of quenched steels. Because of the similar dependencies
between hardness after creep rupture and time to rupture with respect to time and
temperature, it should be possible to correlate hardness with creep resistance or to relate
the hardness at time of rupture with the Larson-Miller parameter for a particular steel.
Figure 6 shows the hardness at the time of rupture plotted against the Larson-Miller
parameter for the 9Cr 1Mo steel in the present study. The correlation is good.

7
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

Hardness, HV

Figure 6 - Graph of hardness against the Larson-Miller parameter for 9Cr-1Mo steel tested at
different stresses and temperatures.

Figure 7 shows the relation of Larson-Miller parameter, which correlates temperature and
time with test hardness and stress. This allows one to correlate all variables in concert and
confirms the validity of Figure 6.

Figure 7 - Graph of Larson-Miller parameter versus hardness and versus stress for 9Cr-1Mo
steel.

The measurement of hardness in situ on power plants is not a simple matter and wide
variations can result from improper procedures. The authors have successfully used

8
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

hardness measurements of in-service components over a number of years but the surface
preparation is very important. It has been our practice to make the measurements on
areas which have been prepared for replication and thus have been ground and polished
to show the underlying microstructure. Additionally the applied load must be small and it
has been found that the only satisfactory method is the use of an ultrasonic hardness
tester, which is also insensitive to its orientation. If the steel has been oxidized and has
lost carbon from the surface layer, the reading will be low. Combining metallographic
observations with the hardness testing is very helpful in interpreting the extent of
damage.

Clearly the method will not detect damage such as Type IV cracking at welds. It is of
primary use in assessing the state of damage and the remaining life of the parent material
and can be valuable in detecting lower potential life in particular regions subjected to
variable temperature, and in particular to overheating.
6. Conclusions
From the results and analysis, it can be concluded that:

There is a similar relationship between hardness at failure of a creep specimen of

9Cr-1Mo steel, time and temperature as is found in correlating rupture time, stress
and temperature with the Larson-Miller parameter. Thus if the appropriate
Larson-Miller Parameter is known for a particular steel together with its initial
hardness and the operating temperature, it should be possible to estimate the
extent of damage and remaining life based on hardness measurements.

Thus the hardness changes for high temperature units operating with varying load
and temperature cycles can be used to provide a first estimate of the damage and
remaining life under such conditions. The estimation depends on careful hardness
measurements, avoiding decarburized surface areas and these should be made on
ground and finely polished surfaces.

The method may be applied to parent material and care must be taken when welds
are present as it will not assess local damage such as Type IV cracking.

7. References

[1] Larson, F.R., Miller, J., A time-temperature relationship for rupture and creep
stress. Trans ASME, v.74, 1952, pp. 765-.
[2] Kohlhfer, W.; Penny, R.K., Hardness testing as a means for creep assessment. Int.
J. of Pres. Ves. and Piping, v. 66,1996, pp.1-3.
[3] Hollomon, J.H., Jaffee, L.D., Time-temperature relations in tempering steel. Trans.
AIME, v. 162, 1943, pp. 223-249.
[4] ASTM A 387-92, 1994, "Pressure vessel plates, alloy steel, chromium-molybdenum".
[5] ASTM E 8M, 1994, Standard test methods for tension testing of metallic materials.
[6] Mans on, S.S.,. Mendelson, A. Optimization of parametric constants for creep-
rupture data by means of least squares. NASA Memorandum 3-10-59E, 1959, pp 1-
34.
[7] Spera, D. A., 1968, A Linear Creep Damage Theory for Thermal Fatigue of
Materials, Ph.D. Thesis, University of Wisconsin.

9
OMMI, Vol. 5, Issue 1, May, 2008. www.ommi.co.uk

[8] Furtado,H.C., de Almeida, L.H., Le May, I, Peace, R., Correlation and extrapolation
of creep rupture data: a case study using 9Cr-1Mo steel data. Trans. Canadian Soc.
for Mech. Eng., v.29, no 1, 2005, pp.1-.21
[9] Manson, S.S., Ensign, C.R., A specialized model for analysis of creep-rupture data
by the minimum-commitment, station-function approach. NASA TM.X-52999,
1971, pp. 1-14.

The authors allow permission to the readers for use with the proviso that they
clearly identify the source of the publication

10