ECCC Creep Conference, 12–14 September 2005, London. [email protected].

uk

Creep Damage and Expected Creep Life for

Abstract

The damage mechanisms affecting engineering steels at high temperatures include creep cavitation and
cracking that can form the path for final failure in susceptible locations such as welds. The evolution of
observed damage is widely used in condition monitoring, timing of inspections and support of life
management. However, the damage evolution is material-dependent, and requires confirmation from
inspection data. For most low-alloy steels, compilations of inspection data have been applied to establish
guidelines for this purpose. Useful experience of the in-service damage is less easily available from newer
steels, and infrequently reported from e.g. the 9-11% chromium steels that are used in hot steam lines of
power plants. However, even then the expected damage evolution can be characterized by using high tensile
multi-axiality for damage acceleration. This approach is also useful for ductile steels with relatively slow
development of creep cavitation in conventional creep testing.

The inspection experience shows very modest creep cavitation in the conventional 11% Cr steel
X20CrMoV11-1 even after long term service. This is of particular interest also because early creep failures
have been reported from steam systems made of P91 (X10CrMoVNb9-1). The differences between these
steels appear to be largely related to the extent precipitation hardening is utilized in providing creep strength.
With more efficient precipitation hardening, P91 and other new high chromium steels are more susceptible
than X20 to deviations in e.g. heat treatments.

Keywords: creep life, creep damage, steel

1. Evolution of creep damage: general features

Creep damage as cavitation and cracking can potentially limit component life, as the evolving cracks can
form the final fracture surface. On the other hand, the observed extent of such damage is also used to
indicate the damage status in the in-service inspections, particularly for welds that are weaker than the parent
material in creep. This is useful because the inspections can then be safely limited to a very small fraction of
the whole structure, and because the inspection experience has facilitated guidelines on setting safe
inspection periods [1]. However, such experience is not available for new steels, and different materials can
deviate in unexpected ways from the foreseen damage evolution and creep life.

The objective of the present paper is to explore the occurrence and reasons for differences in the creep and
creep damage behaviour between 9-11% Cr steels.

The high temperature damage processes include parallel effects of other mechanisms such as creep straining
and strain-dependent damage (e.g. softening), thermal degradation as microstructural changes and related
redistribution of constitutive elements and phases, environmental effects like internal and external oxidation
and corrosion. Although creep cavitation and cracking therefore only represent a part of the total damage,
cavitation also to some extent reflects the impact of other mechanisms weakening the material. This works
particularly well in relatively thick-wall components, when surface oxidation, hot corrosion and short-term
temperature excursions have relatively less impact than e.g. in the internal parts of boilers.

Creep cavitation damage and therefore the relevant criteria for in-service inspections are often tedious to
obtain in a laboratory environment, because of required long testing time. This is particularly true for steels

783
 P. Auerkari, S. Holmström, J. Veivo and J. Salonen

with good creep ductility, such as low-alloy 2¼Cr-1Mo (10CrMo9-10, P22) steel or many higher alloyed 9-
11% Cr steels including X20 (X20CrMoV11-1 or X20CrMoV12-1), P91 (X10CrMoVNb9-1), E911
(X11CrMoWVNb9-1-1) and P92 (NF616). Consequently, characteristic signatures of creep damage are
normally obtained from long term service experience in plant. This is inconvenient for relatively new
materials for which no long term testing data or service experience is available.

An example of observed in-plant creep damage in steel X20 is shown in Fig 1. This is a relatively rare case
in this steel, where inspections tend to show little or no indications of cavitation damage.

Fig 1. Example of in-service creep cavitation damage in an X20 steam line, observed from a surface replica
of a branch weld.

2. Characterisation of creep damage in new steels

For accelerated laboratory testing to induce characteristic long-term creep damage, the traditional approach
in uniaxial creep testing to elevate either temperature or stress level (or both) will not help much in terms of
cavitation damage, although creep life will be reduced (Table 1). However, increasing the tensile multi-
axiality will accelerate creep cavitation. For this purpose, an approach using notched specimens (Fig 2) has
been developed in the European project LICON and applied for selected 9% Cr steels [2-5].
Although creep damage evolution can be quite unequal in different materials, it also involves certain
common characteristics. These include the general order and appearance of damage stages from separate
cavities to aligned cavity chains, cavity growth and coalescence into small scale microcracks, and finally to
macroscopic cracks that become detectable in the standard non-destructive inspections. Also, damage for
given time of exposure tends to be systematically more severe in locations of high stress, weak material (e.g.
welds) and in materials of low ductility. In these general terms, none of the 9-11% Cr steels are expected to
be totally different. However, for new steels there is no long term service or inspection experience, and then
the expected creep damage evolution can only be estimated through laboratory testing. This appears possible,
as realistic forms of damage similar to that expected in service can be reproduced even in relatively creep
ductile materials (Fig 3).

784
 Creep Damage and Expected Creep Life for Welded 9-11%Cr Steels

Fig 2. Observed creep cavitation and crack initiation in a notched specimen of steel E911 (HAZ) after creep
testing (9500 h / 625°C). Note that most cavities are larger than in Fig 1.

Table 1. Comparison of methods for accelerated creep testing.

Applied variable Advantages Potential disadvantages

Temperature Simple, strong effect on life Mechanism changes 1)

(Effective) stress Simple, strong effect on life Mechanism changes 1)

Stress multiaxiality Acceleration of creep damage Control of stress level 2)

785
 P. Auerkari, S. Holmström, J. Veivo and J. Salonen

10000
10000

CAVITY DENSITY (1/mm^2)

1000
1000

100
100

10 10
0,00 0,50 1,00 1,50 2,00 2,50 3,00
0 0,5 1 1,5 2 2,5
DISTANCE FROM NOTCH TIP (mm)
DISTANCE FROM NOTCH TIP (mm)

a) b)
Fig 3. Cavity density (HAZ) in notched specimens of a) steel E911, b) P22/P91 welds, P22 side [4].

3. Incidence of creep damage in 9-11% Cr steels

There are some peculiarities in the observed in-service damage of 9-11% Cr steels. While the creep
cavitation damage is easily found particularly in the welds of vintage steam systems made of low alloy
steels, the same is not true when the material of construction is the 11% Cr steel X20CrMoV11-1 (or X20 for
short). In general, the service and maintenance experience suggests that the steel X20 performs extremely
well [6]. In fact, much of the reported creep cavitation damage in this material is most probably due to
misinterpretation due to incorrect etching in replica inspections [7]. Although some plants where X20 is used
in e.g. headers, steam lines and turbine parts have exceeded a service time of 200 000 h, lack of damage
means that it is still somewhat unclear how long would be the actual technical lifetime for these components.
The good performance was good news also in the sense that the earlier material of choice, 0.5CrMoV
(14MoV6-3) was rather sensitive to creep cavitation.

The observed lack of definite cavitation damage in the welds of X20 after long term service appears
somewhat baffling, when compared to some recent reports showing early creep damage in welds of P91 [8].
Both steels belong to the same general class of ferritic steels, first heat treated to martensitic state and then
tempered to the required level of mechanical properties, and with a significant contribution of precipitate
strengthening by carbides and carbonitrides. However, there must be some differences also, if the in-service
behaviour is so very different. From design point of view, the actual creep life should not be very much
materials dependent for a given design life, because the design aims to compensate for the materials
properties. When this is not the case, either some of the materials properties are not as expected or there is
other bias in the features that determine the actual creep life. The steel X20 was introduced nearly half a
century ago, and with generally good results until present [6]. It could be seen a little disappointing if a
comparable relative performance were not provided by newer steels like P91, first introduced some 30 years
later. Since the alloy design is comparable, this could also imply potential for similar disappointments or
early creep failures in steels E911 and P92 (or P122), of which even less in-service experience is available.

4. Differences in creep behaviour between steels

One obvious difference between X20 and P91 (or any of the newer 9% Cr steels like E911 or P92) is in the
creep strength levels as they appear in standards and guidelines (Fig 4). The standard creep strength levels in
e.g. EN 10216-2 are considerably lower for the older generation steel X20 than for P91. Therefore, the wall
thickness of components made of P91 can be clearly lower than for X20, although X20 was previously at
least a similar design improvement to the low alloy steel 14MoV6-3.

Another difference in principle between X20 and P91 is in the maximum allowed service temperatures: X20
has been applied to steam temperatures up to about 560-565°C, and P91 in newer plants slightly higher, up

786
 Creep Damage and Expected Creep Life for Welded 9-11%Cr Steels

to about 570-580°C. However, in most cases the actual service temperatures are lower for both steels,
typically in the range 500-550°C. This is often because of other set limits, for example the limited maximum
gas temperatures in HRSG’s, hot corrosion limits in biomass and process boilers, or other conservatism in
process design. Even when such factors were not specifically considered, the temperature difference between
systems made of X20 and P91 is only partly due to the strength capabilities of the materials. Another reason
is related to design codes and design practices, resulting generally in somewhat lower temperatures in the
geographical regions where X20 has been used.

Further differences can be seen in the chemical compositions: X20 has higher carbon content but less (if any
specified) nitrogen than P91, with resulting differences in the composition of MX precipitates. Higher carbon
content also causes higher martensite hardness, and low toughness before tempering in X20, but this is an
issue for manufacturing and maintenance rather than for in-service properties. However, differences and
deviations in manufacturing also can have a very important impact on the service properties. In particular,
heat treatments and welding can cause significant scatter in the creep strength, and potential sources of
severe problems. In these terms, X20 could be somewhat less sensitive than P91 to deviations in heat
treatments, though less convenient to weld. Additional scatter factors can be expected e.g. from variations in
section thickness or production via rolling, drawing, forging, casting etc.

180

160
200000 h rupture strength, MPa

140

120

100

80

P92
60 X11CrMoWVNb9-1-1
X10CrMoVNb9-1
X20CrMoV11-1
40 14MoV6-3
10CrMo9-10
20
480 500 520 540 560 580 600
Temperature (C)

Fig 4. Comparison of the 200 000 h creep rupture strength of selected 9-11% Cr steels according to EN
10216-2 and ECCC data sheets; also the corresponding strength values of the low-alloy steels 0.5CrMoV
(14MoV6-3) and 10CrMo9-10 are shown.

Comparing the long-term (200 000 h) creep rupture strength of different 9-11% Cr steels (Fig 4) shows that
at the usual levels of service temperatures the relative improvement from X20 to P91 is roughly similar as
was the improvement from the traditional low-alloy steels (10CrMo9-10 and 14MoV6-3) to X20. However,
the relative improvement from P91 to newer steels (E911 and P92) is less impressive. Particularly small
difference seems to now exist between P91 and E911.

As noted above, the plant experience with the X20 steel has been excellent. The reasons are probably at least
partly due to relatively low service temperatures and often also low stresses in comparison with the allowed
maximum levels. However, there are also other factors involved, as can be seen from Figs 5 and 6.

787
 P. Auerkari, S. Holmström, J. Veivo and J. Salonen

Fig 5 shows for the steels of Fig 4 the relative improvement in 200 000 h creep rupture strength from the
corresponding strength of 10CrMo9-10 (2¼Cr-1Mo or P22), as a function of temperature. This creep
strength ratio essentially shows the improvement due to precipitate strengthening, suggesting for example
that the improvement in 14MoV6-3 is practically constant by a factor of about 1.2-1.3, independent of
temperature. In case of X20, the relative improvement factor is between 1.5 and 2.0. However, for X20 the
ratio is less constant and shows an apparent maximum at about 550-560°C, decreasing slightly towards both
lower and higher temperatures. Remarkable features in case of the 9% Cr steels P91, E911 and P92 are not
only the large values of the creep strength ratio but also its increasing trend towards higher temperatures (to
values of about 3 or even higher).

The high values of the creep strength ratios for the newer steels imply that the potential loss in creep strength
is also high, if the intended precipitation hardening is not achieved in manufacturing (heat treatments) or is
lost by in-service degradation. The resulting adverse consequences for materials such as P91 can be therefore
expected to be much more pronounced than for steels with lower strength ratios, like X20. As the difference
in these ratios between P91 and X20 is clearly increasing with temperature, the problems can be exacerbated
in cases where P91 or other newer steels are used at the high end of the service temperature range. Naturally,
as creep is always strongly accelerated by increasing temperature, this is a compounding factor for creep
failures.

This reasoning is roughly in line of the concept of “inherent creep strength”, which implies that the
asymptotic very long term creep strength would be mainly due to solid solution strengthening by elements
such as Mo and W, after all contributing particles have been precipitated and coarsened [10].

4.0
P92
3.5 X11CrMoWVNb9-1-1
X10CrMoVNb9-1
Rm 200kh/T / Rm 200kh(10CrMo)/T

X20CrMoV11-1
3.0
14MoV6-3

2.5

2.0

1.5

1.0
480 500 520 540 560 580 600
Temperature (°C)

Fig 5. The 200 000 h creep strength ratio for the steels of Fig 4, using the rupture strength of 10CrMo9-10 as
a baseline for the comparison.

Another interesting feature arises at temperatures above 570-580°C, where only for P91 and P92 the creep
strength ratios appear to be still increasing. Even for the steel E911, for which there is a recent updated
assessment of recommended creep strength values by ECCC, the ratio appears to be leveling off for
increasing temperatures. There is no direct evidence for this, but it could perhaps be speculated whether
future reassessment of P91 might change this.

788
 Creep Damage and Expected Creep Life for Welded 9-11%Cr Steels

It is an unfortunate but common feature in the development of new steels and materials standards that the
early recommended values of creep strength tend to be somewhat overly optimistic, requiring downward
correction in later amendments. This was the case for nearly all low alloy steels and has been seen also for
E911, for example. The simple reason is that the past or current assessment methods are not able to fully
accommodate the degradation mechanisms that are affecting the long term mechanical properties. Again,
there is no direct evidence for it, but such non-conservative assessment results can be also expected for the
future new steels. It should be also noted that in Fig 4, the 200 000 h creep strength values for P91 are from
extended extrapolation (by a factor > 3 in time), and therefore changes would not be unexpected in later
reassessment of longer term data.

Here the nominal “time safety margin” or design life ratio without additional safety factors is defined as the
following isothermal time factor:

Design life ratio = time to rupture at 0.8 x (200 000 h rupture stress) / 200 000 h
and Fig 6 gives this ratio as a function of temperature for the same steels as Fig 4.

It is apparent that the steels fall into two distinct groups. The low-alloy steels and X20 form here one group,
with a design life ratio ≈ 1.5 -3 at temperatures above some 510°C. The second group includes the 9% Cr
steels P91, E911 and P92, for which this ratio is 2 to 3 times higher and increases clearly with decreasing
temperature (Fig 6).

The higher values of design life ratios for P91, E911 and P92 than for X20 would suggest higher inherent
safety margins for the 9% Cr steels. Therefore, for intact material the latter materials should perform well.
However, this could be more than compensated for, should the creep strength of the 9% Cr steels collapse
due to lacking precipitate strengthening. Such a collapse would have somewhat less drastic effects on X20,
which is apparently more comparable to the low alloy steels.

These features would be parallel for welds of classical steels, which tend to show cross-weld rupture strength
reduced by about 20% from the strength level of the corresponding parent material. The same may not be
quite apply for newer 9% Cr steels, for which higher weld creep strength factors have been proposed.
However, the additional strength reduction could mainly apply to the high temperature end of the application
range, and for many applications the classical rule would still work reasonably well. In such cases at least,
the above reasoning would largely apply to welds as well.

789
 P. Auerkari, S. Holmström, J. Veivo and J. Salonen

9
Grade 92
X11CrMoWVNb9-1-1
8
X10CrMoVNb9-1
X20CrMoV11-1
7 14MoV6-3
10CrMo9-10
6
Time factor

1
480 500 520 540 560 580 600
Temperature (C)

Fig 6. Nominal design life ratio for the same steels as in Fig 2; note approximate similarity of X20 with the
low alloy steels 10CrMo9-10 and 14MoV6-3.

4. Conclusive remarks

Creep damage is unavoidably inflicted to materials at high temperature, and this damage is reflected as
limited creep life and gradual degradation of material properties. Creep damage can also be used as a
monitoring tool, when non-destructive techniques of measurement and experience on damage accumulation
rates are available. Such approach as replica inspections and related guidelines are available and widely
applied for low-alloy steels and their weldments. However, the service history of most 9-11% Cr steels is
much shorter, particularly for relatively new steel types. The inspection experience also suggests that the
only vintage steel in this class, X20CrMoV11-1, shows slower damage accumulation rates than conventional
low alloy steels, and records of creep failures are rare even after long term service. The reasons of good
performance are of interest, not only for the life management of the existing plants with this steel, but also
for newer plants where this steel is largely replaced by P91 (X10CrMoVNb9-1) that has reportedly shown
early creep failures.

To check for the potential for creep cavitation damage, a relatively convenient technique has been developed
using high tensile multi-axiality for damage acceleration. This approach works much better for this purpose
than conventional accelerated creep testing with elevated temperature or effective stress levels. The method
appears particularly useful for relatively ductile new steels, where creep cavitation can develop slowly and
no in-service inspection results can be used to support life assessments.

In case of P91, the predicted long term stress ratio of 200 000 h rupture strength with respect to the
corresponding values for 10CrMo9-10 (2¼Cr-1Mo) steel appears somewhat suspect for the high end of the
service temperatures. The same may be true for P92, but probably not for (X20), which shows a relatively
modest strength enhancement by precipitate strengthening. Combined with also modest design temperatures
in most cases, this largely explains the excellent service records of power plants with X20 as the structural
material of the steam systems. It also points to relatively higher potential of strength loss in 9% Cr steels like
P91, if the intended precipitation strengthening is lost. Recent reports of P91 failures in plant seem to
confirm this. Fortunately, also a significant fraction of P91 applications do not fully utilize the temperature

790
 Creep Damage and Expected Creep Life for Welded 9-11%Cr Steels

and stress capability of the material, and this will reduce the potential impact from e.g. deviations in heat
treatments.

References

1. Auerkari, P., Salonen, J. & Borggreen, K. Guidelines for evaluating in-service creep damage. Nordtest
TR 302, 1995. VTT, Espoo, 24 p.
2. VGB-TW 507. Guidelines for the assessment of microstructure and damage development of creep
exposed materials for pipes and boiler components. VGB, Essen 1992. 83 p.
3. Rantala, H., Hurst, R.C., Auerkari, P., Bendick, W., Holdsworth, S.R. & Nikbin, K. Predicting long term
creep behaviour using the LICON methodology. 9th Int. conf on Creep and Fracture of Engineering
Materials & Structures, Swansea 1-6 April 2001.
4. Auerkari, P., Holdsworth, S.R., Rantala, J.H., Hurst, R., Coussement, C. & Hack, R. Modelling the
development of creep damage: the LICON experience. Proceedings of Baltica V: Int. Conf. on Condition
and Life Management for Power Plants, Porvoo, Finland, 2001. Vol 2, p. 609-620.
5. Bendick, W., Hahn, B. & Schendler, W. Development of creep damage in steel grades X10CrMoVNb9-1
(P/T91) and X20CrMoV12-1. Predicting long term creep behaviour using the LICON methodology. 9th
Int. conf on Creep and Fracture of Engineering Materials & Structures, Swansea 1-6 April 2001.
6. Storesund, J., Borggreen, K., Zang, W., Nilsson, H. & Samuelson, Å. Creep damage in welds of steel
X20 CrMoV 12 1 (in Swedish). Värmeforsk Report 874, Stockholm 2004. 20 p + app.
7. Borggreen, K. Storesund, J. Creep behaviour of welds in X20 CrMoV 12 1 evaluated from replica
inspection results. Proceedings of Baltica VI – Int. Conf. on Life Management for Power Plants. Helsinki-
Stockholm 2004. Vol 2, p. 449-463.
8. Auerkari, P., Holmström, S., Salonen, J., Borggreen, K., Storesund, J. & Rui Wu. Evaluation of creep
damage from replica inspection results. VTT Report VAL B 211, VTT, Espoo 1997. 33 p.
9. Brett, S. J., Bates, J.S. & Thomson, R.C. Aluminium nitride precipitation in low strength Grade 91 power
plant steels. 4th EPRI Int Conf on Advances in Materials Technology for Fossil Power Plants, Oct 26-28,
2004, Hilton Head Island, S.C. 15 p.
10. Kimura, K., Kushima, H. & Abe, F. Degradation and assessment of long-term creep strength of high Cr
ferritic creep resistant steels. Proc. Of the EPRI Int. Conf. on Advances in Life Assessment and
Optimization of Fossil Power Plants. EPRI 2002.

791