Correlation between structural changes of M23C6 carbide and mechanical behaviour of P91 steel after thermal aging

Author’s Accepted Manuscript Correlation between structural changes of M23C6 carbide and mechanical behaviour of P91 steel after thermal aging Albertas Grybėnas, Vidas Makarevičius, Arūnas Baltušnikas, Irena Lukošiūtė, Rita Kriūkienė www.elsevier.com/locate/msea

PII: DOI: Reference:

S0921-5093(17)30580-4 http://dx.doi.org/10.1016/j.msea.2017.04.103 MSA35004

To appear in: Materials Science & Engineering A Received date: 29 December 2016 Revised date: 24 April 2017 Accepted date: 25 April 2017 Cite this article as: Albertas Grybėnas, Vidas Makarevičius, Arūnas Baltušnikas, Irena Lukošiūtė and Rita Kriūkienė, Correlation between structural changes of M23C6 carbide and mechanical behaviour of P91 steel after thermal aging, Materials Science & Engineering A, http://dx.doi.org/10.1016/j.msea.2017.04.103 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Correlation between structural changes of M23C6 carbide and mechanical behaviour of P91 steel after thermal aging Albertas Grybėnasa, Vidas Makarevičiusa,*, Arūnas Baltušnikasa, Irena Lukošiūtėa, Rita Kriūkienėa a

Laboratory of Material Research and Testing, Lithuanian Energy Institute, Breslaujos st. 3, LT-44403 Kaunas, Lithuania

*Corresponding author: Vidas Makarevičius

E-mail address: [email protected]

Abstract The results of thermal aging effect on the P91 steel after a long-term exposure at 600, 650 and 700°C are presented. Mechanical behaviour of the aged material was assessed by performing tensile, fracture toughness and short term creep tests along with microstructure examination. Sample aging duration and temperatures were chosen considering the structural changes of M23C6 carbide, which are described by the exponential equation as a function of time and temperature. The time-temperature dependence was previously established from M 23C6 lattice expansion as a result of alloying elements’ diffusion into the carbide. The main focus is to determine whether the same relationship could be used for the assessment of the mechanical properties of P91 steel. Tensile, fracture toughness and creep test data at different aging duration and temperature indicated that predicted data correlate well with experimental data and gives a possibility to forecast changes caused by steel thermal exposure. Experimental results demonstrate that the used method provides a sufficiently good technique to predict the mechanical behaviour of P91 steel during aging.

Keywords: P91steel, aging, M23C6 carbide, microstructure, mechanical characterization 1. Introduction Ferritic -martensitic heat resistant steel P91, introduced in plants in 1988s, is a widely used material for high temperature applications in power industry, such as steam pipes and turbine components [1]. Continuous operation of steel at elevated temperature and pressure result in degradation of microstructure with time. The evolution of changes in the microstructure of steels, such as grain coarsening, precipitation of secondary phases leads to a reduction of mechanical, creep and fatigue properties [2-4]. In order to use these steels safely in service of power plants, development of the methodology of creep rupture strength estimation for metal is very important [5]. Time-temperature parameters (TTP) are used in order to reduce test time and labour costs for obtaining creep rupture data at low temperatures and stresses. Using TTP approach the prediction of long-term creep behaviour was performed by extrapolation of short-term creep-rupture data to longer times than have been

measured. Arrhenius relation between creep rate and temperature has led to a number of TTP to be developed. The Larson–Miller (LM), the Manson–Haferd (MH), the Orr–Sherby–Dorn (OSD) and the Mendelson-Roberts-Manson (MRM) parameter methods (physically-based models) are frequently used in creep rupture data assessment [6-9]. Larson-Miller approach (parametric method) is one of the most widely employed techniques for stress-rupture behaviour prediction in metals. It was observed that other approaches, i.e. MH, OSD and MRM properly fit under certain defined conditions [7]. As imperfection of these parametric methods is a lack of details about the appropriate microstructural degradation of steel during aging and damage due to long-term creep exposures, i.e. dependence between microstructure changes and mechanical properties of aged steels [10]. Some researchers [11] performed a comparison of dimensional distribution of precipitates (MX, M23C6, Laves phase) with LMP of P91 steel. Mechanical properties, such as hardness data, can also be usefully plotted against a time - temperature function, such as the Larson-Miller parameter. Hardness vs Larson-Miller parametric plot was created combining hardness data from thermal and creep exposure tests and analysed for the estimation of average in-service exposure temperatures for this alloy [12-15]. Solid-phase reactions and microstructure changes that take place during steel service, in general, can be classified as a diffusion controlled process [16-19]. Alloying element diffusion at isothermal conditions could be described by the well-known Johnson–Mehl–Avrami (JMA) transformation kinetics equation [20]. During operation of the heat resistant steels, the M23C6 carbide lattice increases due to diffusion of alloying elements from the matrix into the carbide lattice, replacing Fe by Cr and Mo. Expansion of the M23C6 lattice proceeds and depends on the fact that chromium and molybdenum have a slightly larger atomic radius 0.125 nm and 0.135 nm, respectively, than that of iron 0.124 nm [21, 22]. It was determined that Cr atoms tend to dominate in the M23C6 carbide lattice due to the higher diffusion coefficient of Cr than Mo [20, 23]. In our previous research [23], X-ray diffraction studies revealed that in a certain interval of time at high-temperature exposure the M23C6 carbide lattice parameter a value plotted in natural logarithmic scale increases linearly and can be fitted by JMA equation. The equation was derived for calculation of M23C6 lattice parameter value depending on the aging time at a given temperature. The limits of exposure time-temperature region, in which the lattice parameter increases linearly, were

determined, and the equation for calculation of carbide M23C6 lattice parameter a value shift on a time scale was created. The objective of this study is to determine whether this relationship could be used as an indicator for the assessment of the real temperature exposure time of P91 steel and for evaluation of mechanical properties of P91 steel. 2. Materials and methods 2.1. Materials characterization Test samples used for the investigation were cut from the commercial 30-mm-thick P91 steel plate. The chemical composition of the as-received material determined with metal analyser Q4 Tasman is provided in Table 1 and microstructure is presented in Fig.1.

Table 1. Chemical composition of the as-received P91 steel (wt. %). C

Si

Mn

P

S

N

Al

Cr

Mo

Cu

Nb

Ni

V

0.094 0.38 0.35 0.006 <0.0005 0.035 0.012 8.78 0.96 0.07 0.08 0.16 0.18

Fig. 1. SEM microstructure of the P91 steel in the as-received condition.

Steel sections from which subsequently were produced tensile, creep and fracture toughness test specimens were isothermally annealed for up to 20 000 hours at 600, 650 and 700°C. To ensure even temperature distribution and minimize oxidation, samples were placed in steel boxes filled with calcium carbonate powder. The microstructure of samples was analysed by a ZEISS EVO MA10 scanning electron microscope equipped with Bruker XFlash 6/10 EDX detector. The voltage used for the analysis was 20kV. In

order to highlight the steel microstructure, the polished surface was etched by Vilella’s reagent. Quantitative analysis of precipitates was performed at 20 000x magnification. 2.2. Mechanical testing From the aged steel samples the specimens for tensile and creep tests were machined. Specimen dimensions were 8 mm in diameter and 40 mm in gauge length. Tensile tests were performed according to EN ISO 6892-1:2009 (at room temperature) and EN ISO 6892-2:2011 (at elevated temperature) standard recommendations, using Instron Model 8801 servo-hydraulic machine. Necessary heating for tensile (550°C) and creep (600°C) tests was maintained by three-zone split furnace. The short term constant load creep tests were performed at 600°C and applied stress of 200 MPa. The temperature was measured by two K-type thermocouples mounted on the specimen and shielded by a fixed piece of mineral wool. Testing temperature was continuously monitored and maintained constant to within 1°C of the desired value of 2°C. The creep elongations were measured and continuously recorded. Plane strain fracture toughness was determined in accordance with ASTM E 399-12e3 using compact tension C(T) specimens of width W=50 mm and thickness B=25 mm. Specimens were fatigue pre-cracked at cycling load frequency of 5Hz (R=0.1) according to pre-programmed 4-step load decrease depending on the resulting crack length. The initial cycling force Pmax = 33.0 kN (Kmax = 55.3MPa √m) did not exceed 80% of set value KQ. The end of crack formation was performed at the value of Kmin = 37.9 MPa √m, which did not exceed 60% of the set values of KQ. Samples with minimum 1.5 mm pre-crack were loaded with a simultaneous recording of load P and crack opening displacement (COD). The values were calculated using the graphical method described in ASTM E 399. 2.3 Aging of test samples Sample aging duration and temperatures were chosen considering the structural changes of M23C6 carbide which were described in our earlier study [23]. The study revealed that in a certain interval of time at high-temperature exposure the M23C6 carbide lattice parameter a value, determined by X-ray diffraction measurements, plotted in natural logarithmic scale increases linearly, as shown in Fig. 2. For calculation of equivalent changes of M23C6 lattice parameter at any temperature and aging duration, the following equation was derived [23]: lntx = lntTn – E/R(1/Tn – 1/Tx), where tx – calculated time (s) at a temperature Tx (K); Tn – aging temperature, E is the activation energy, 248 kJ/mol; R is the gas constant, 8.314 J/mol.

(1)

In such way, predictions of lattice changes can be made for material aged at a higher temperature by recalculating equivalent aging time. As the lattice change in steel is a diffusion controlled process, the equation reflects such changes during steel aging. This study will attempt to use Eq. (1) in predicting mechanical behaviour for specimens aged at different temperatures. Aging duration for each temperature was selected so that at least one sample was in the linear zone, and one after the bend of the curve (see Fig.2). Sample aging conditions are given in Table 2, along with the equivalent aging duration values calculated by Eq. (1). Table 2. P91 steel aging temperature and duration. Aging temperature, °C 600 600 650 650 650 700 700

Aging duration, h 10008 20064 792 4152 10872 480 1608

Equivalent aging duration, calculated by Eq. (1), h 600°C

650°C

700°C

5040 26400 69216 16104 53928

1560 3144 2520 8472

298 598 151 787 2064 -

-5 600 °C

480h

10872h

1608h

650 °C

-6

ln[-ln(a0/at)]

700 °C -7

4152h

eq. (1) 792h

-8

20064h 10008h

-9

-10 7

9

11

13

lnt

15

17

19

21

Fig. 2. The plot of M23C6 carbide relative lattice parameter ln [-ln(a0/at)] versus lnt (t is in s) at the isothermal temperature [23 ]. The arrows show aging time for the samples described in this work. 3. Experimental 3.1 Structural analysis of precipitates The microstructure of P91 steel after thermal treatment is presented in Fig.3. In aged steel M23C6 carbides are located along prior austenite grain boundaries and martensitic lath boundaries, fine

MX-type carbonitrides are within grains. The microstructure is similar, as described in [24-26]. The precipitates occur both at the grain boundaries and inside ferrite grains. In steels containing carbon above 0.05%, the main precipitate located along boundaries is M23C6 carbide [27]. During this process, Fe atoms in M23C6 carbide lattice are partly replaced by Cr and Mo, but as diffusion coefficient of Cr is higher than Mo, consequently Cr atoms tend to dominate in the M23C6 carbide lattice [20, 28]. Those suggestions were confirmed earlier [23] by EDX analysis of the residual extracted from the aged P91 steel. As a result of increasing thermal exposure duration, the microstructure of the tested steels is subject to continuous degradation, which manifests itself in the increased size of precipitates after aging at 600°C for 1·104 and 2·104 h. The formation of Laves phase (Fe,Cr)2Mo at grain and lath boundaries is observed (Fig. 3, a), the size of precipitates increases with aging time (Fig. 3, b). Laves phase was found [25] to improve the creep strength for a short-term exposure, but this positive effect can not last long due to the quick coarsening of Laves phase. The precipitation of Laves phase was observed [29] to effectively decrease the minimum creep rate, but coarsening of this phase would promote the acceleration of the creep rate after reaching the minimum creep rate. As a result, the effect of Laves phase on the extension of creep rupture time is rather small. As confirmed by EDX analysis, the Laves phase is not present at 650 and 700°C. According to [30], the Laves phase in P91 steel is stable up to 620°C. During aging treatment at 650 and 700°C mostly M23C6 carbide precipitates along grain boundaries (Fig. 3 c, e) and coarsening increases with exposure time (Fig. 3. d, f).

(Fe,Cr)2Mo (Fe,Cr)2Mo

a

Element Cr Fe Mo V

Wt% 9,3 73.6 13.3 0.25

b

M23C6

1

M23C6

c

d

M23C6 Element Cr Fe Mo V

M23C6

e

Wt% 35.7 58.3 5.8 0.46

f

Fig. 3. SEM microstructure of the P91steel after aging at 600, 650 and 700°C. Images: a– 600°C/10008 h, b– 600°C/20064 h, c– 650°C/4152 h, d–650°C/10872 h, e–700°C/480 h, f– 700°C/1608h. The histogram in Fig. 4 shows that the area fraction of smaller (0.2-0.3 µm) precipitates decreases while the area fraction of larger precipitates increases. This is more evident for the samples aged at 650 and 700°C which correspond to the samples in the non-linear part of the plot, as shown in Fig. 2. Both of 600°C samples are in the linear part and precipitates’ increase is due to the Laves phase formation.

50 40

600°C/ 10008h

P91 (as received)

600°C/20064h

30 20 10 0

a

b

Area fraction, %

50 40

650°C/3168h

650°C/792h

c 650°C/4152h

30 20 10 0

d 50

e 700°C/480h

650°C/10872h

f 700°C/1608h

40 30 20

g

h

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 more

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 more

0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 more

10

i

Equivalent diameter of precipitates, µm

Fig.4. Distribution of area fraction of precipitates in P91 steel measured in as-received (a) state and after aging at 600 (b, c), 650 (d, e, f, g) and 700°C (h, i). 3.2. Results of mechanical tests of aged P91 steel P91 steel in the as-received condition is characterised by high mechanical properties. For the each aging condition tensile, fracture toughness and creep tests were performed. Each test was repeated at least twice. The average values for P91 steel in the as-received condition and changes after thermal aging are shown in Table 3. Tensile tests were performed at room temperature (RT) and 550°C. Mechanical test results show that after aging at the selected temperature - time conditions observed changes are not significant. The largest change in strength properties is observed for specimens which aging time is after the bend of the curve (see Fig. 2), i.e. after aging for 10872 hours at 650°C and 1608 hours aging at 700°C. In the latter case, strength and yield strength values are reduced by about 10% at RT and 5% at 550°C.

Toughness characterization was accomplished by Plane Strain Fracture Toughness Test (KIC). The Average values of KQ, and Kmax for as-received P91 steel and after thermal aging is shown in Table 3. Stress intensity vs crack opening displacement curves calculated from the experimental data points are shown in Fig. 5. As shown by the tests carried out on C(T) specimens, P91 has a high resistance to brittle fracture. In all cases, crack opening up to Kmax has a ductile character without larger crack opening displacement which indicates that there is no brittle fracture on the crack tip. As-received P91 specimens are characterized by a large Pmax/PQ ratio reaching 2, which exceeds the value of 1.1 required by ASTM E399 in order that KQ can be valid as the KIC. The decrease in Pmax caused by aging reduced Pmax/PQ ratio only to 1.8 and does not have a significant influence on determined PQ and calculated KQ values. The average values of creep test data after P91 steel aging are given in Table 3. Fig. 6 show creep curves for uniaxial creep tests conducted at 600°C under applied stress level of 200 MPa. In contrast to mechanical and fracture toughness tests, as shown in Table 3 and Fig. 5, significant changes are observed in the creep behaviour of steel after aging, when compared to the as-received state. Strain vs time curves (Figs. 6 a-c) show that isothermal aging of the steel leads to an increase in the creep plasticity. Depending on the aging time aged steel shows noticeably shorter time to fracture and the shape of creep curves for the as-received and aged specimens differ considerably. Tensile, fracture and creep tests show deterioration of properties after aging at each temperature which is even more obvious in creep tests. Table 3. Properties of the P91 steel after different thermal aging conditions. Tensile properties, MPa Thermal aging conditions

As-received 600°C; 10008h 600°C; 20064h 650°C; 792h 650°C; 4152h 650°C; 10872h 700°C; 480h 700°C; 1608h

Yield stress R0.2

Ultimate strength Rm

RT/550°C 513/351 666/382 504/343 657/371 503/335 655/363 507/351 658/389 501/345 651/379 487/334 637/364 497/342 647/374 468/316 634/348

Fracture toughness, MPa m1/2 KQ

Kmax RT

75.1 72.8 72.4 73.4 73.9 70.6 69.7 69.8

146.4 140.8 140.0 143.5 137.5 135.8 142.3 134.3

Short time creep Time to rapture tR, h

Minimum creep rate, εm 10-3, h-1

600°C, 200 MPa 42.8 1.25 33.0 1.55 26.7 1.88 40.1 1.44 26.6 2.20 16.1 3.54 28.2 2.06 17.3 3.31

150

K, MPa m1/2

100 As-received 600°C/10008h 650°C/ 792h 650°C/ 4152h 650°C/10872h 700°C/ 480h 700°C/1608h KQ

50

0 0

0.5 1 1.5 Crack mouth opening displacement, mm

2

Fig.5. Stress intensity vs displacement curves. 25

(a)

600 OC As-received

20

Strain, %

10008 h 200064 h

15

10

5

0 0

10

20 Time, h

30

40

25

(b) 20

Strain, %

650 OC

Asreceived 792 h 4152 h

15

10872 h 10

5

0 0

10

20 Time, h

30

40

25

Strain, %

(c)

700 OC

20

Asreceived 480 h

15

1608 h

10

5

0 0

10

20 Time, h

30

40

Fig. 6. Creep curves at 600°C under applied stress level of 200 MPa for P91steel after aging at 600 (a), 650 (b) and 700ºC (c).

4. Discussions

Aging time at 650 and 700°C can be recalculated to the equivalent aging time at 600°C using Eq. (1). After that, mechanical properties of the specimens aged at 600, 650 and 700°C were plotted on a single aging time axis, as shown in Fig. 7. As a result, estimated aging can be predicted up to 7·104 h. As can be seen, long-term annealing both at room and elevated temperature has only a slight impact on the reduction in strength properties, i.e. yield point, tensile strength.

700

Rm (RT)

650 600°C

Rm, R0.2, (MPa)

600

650°C

550

R0.2 (RT)

700°C

500 450 Rm (550°C)

400 350

R0.2 (550°C)

300 0

10000

20000

30000 40000 Time, h

50000

60000

70000

Fig. 7. Prediction of yield stress and ultimate strength of P91 steel depending on aging time. Filled data markers are experimental data at 600°C. Transparent markers indicate 650 and 700°C data recalculated to equivalent aging time at 600°C by Eq. (1). The data show a relatively small reduction in strength properties after estimated 7·104 h aging, a decrease in Rm and R0.2 is also similar. The decrease does not exceed 6% at room temperature and 8% at 550°C. Data points of specimens aged at 650 and 700°C which were shifted along the time axis up to 3·104 h shows a good correlation with data points of specimens heat treated at 600°C for up to 2·104 h. Larger deviation from the general trend is observed in a case of specimen aged at 700°C, probably because at this temperature, as shown in Fig. 2, the deviation from linearity according to Eq. (1) at the same lattice parameter a value begins earlier and at higher intensity, if compared to 650 and 600°C. It should be noted that predicted changes of mechanical properties at 600°C for up to 7·104 h correlate well with other studies on P91 steel aging up to 3∙104 h [2] and 5∙104 h [31] and that indicates that long-term annealing at 600°C did not result in significant changes in mechanical properties. As in the case of tensile tests, changes in fracture toughness properties were compared at 600°C when aging durations for 650 and 700°C tests were recalculated by Eq. (1). It appears that KQ values remain almost unchanged, but Kmax values decrease up to 10%. Fracture toughness parameters depending on the equivalent time are shown in Figure 8. The fracture toughness data, in the same way as tensile properties, shows little variation at 600°C under a covered aging period. Experimental data at 600°C also shows good correlation with estimated by Eq. (1) data points at 650 and 700°C. Two aspects should be noted when evaluating test data for aged specimens by applying equivalent time calculated by Eq. (1). On the one hand, tensile and fracture toughness properties depending on the equivalent time at the chosen 600°C temperature predicts their changes sufficiently and are consistent with experimental data at 600°C. However, on the other hand, both the tensile and fracture toughness values do not experience significant change over time. Therefore, it is difficult to evaluate the validity of calculated equivalent time values, considering that the longest experiment duration is up to 2·104 h while prediction based on equivalent time is up to 7·104 h. For such validation short-term creep tests were carried out, because at the same length of time aging has a greater effect on creep rupture strength.

160 Kmax

KQ, Kmax, MPa m1/2

140 120

600°C 650°C

100

700°C

KQ

80 60 40 0

10000

20000

30000

40000

50000

60000

70000

Time, h

Fig. 8. Prediction of stress intensity factors KQ and Kmax for P91 steel depending on aging time. Filled data markers are experimental data at 600°C. Transparent markers indicate 650 and 700°C data recalculated to equivalent aging time at 600°C by the Eq. (1).

The relation between minimum creep rate and time to failure is usually described by the MonkmanGrant relationship [32]: ,

(2)

where m and C are constant. Although the time to failure for the service exposed material depending on temperature appeared to differ from that for the aged materials [33] and over a temperature range from 500 to 700°C [34] for 9Cr–1Mo–steel the Monkman-Grant relation in most cases has a good linear relationship for all aging conditions. Even a better fit for a large data scatter can be accomplished using the modified Monkman-Grant relation proposed by Dobes and Milicka [35]: ,

(3)

where εf is the strain to failure and C' is the modified Monkman-Grant constant. The data plotted in accordance with Monkman-Grant relationship (Fig. 9) shows that there is a small difference between creep tests of aged specimens. In the case of Monkman-Grant relation (Eq. 2) data points for 600°C specimens are located a little below the trend line. This indicates that these specimens have somewhat smaller time to failure or time to failure/strain to failure ratios. This may be caused by the Laves phase formation at this temperature which results in smaller tertiary

creep. This apparently leads to better data coincidence in modified Monkman-Grant relation as it is demonstrated in Fig. 9(b) that shows failure/strain to failure time vs minimum creep rate plot. The parameters of relationship (2) and (3) obtained by regression analysis are indicated in Figure 9 (a) and (b).

100

(a)

600°C 650°C

Time to failure tf, h

700°C as-received

C =0.080; m=0.941

10 0.001

0.01

Minimum creep rate ε̇m , h-1

Time to failure/Strain to failure (tf/εf), h

1000

(b)

600°C 650°C 700°C As-received

100

C' =0.223; m'=1.029

10 0.001

Minimum creep rate ε̇m , h-1

0.01

Fig. 9. Steel P91 creep data after aging at 600, 650 and 700°C plotted according to the MonkmanGrant relationship, Eq.2 (a) and modified Monkman-Grant relationship, Eq. 3(b). Time to failure data from creep tests depending on aging temperature and time are presented in Fig. 10. Using the Eq. (1) and applying calculated equivalent aging time, the data points were recalculated for each temperature. Considering data scatter for individual measurements of creep

test and that a number of test points in each of the selected test temperatures is limited, there is a sufficiently good correlation between experimental data and recalculated data points for each test temperature. The equation describing the dependence of time to failure on the aging temperature and time is given in Fig. 10. Since creep tests were performed only at one stress level, a more detailed study on the aging impact on creep was not performed as it was not the objective of this work. Appropriateness of using Eq. (1) as a time-temperature parametric relationship could be defined more precisely having more experimental data points at a different temperature, time and stress levels. In this case, it can be assumed that the equation (1) reflects the changes in the nature of material properties’ changes which are caused by time – temperature dependent material aging.

100 600°C

Time to failure tf , h

200 MPa 600°C

650°C 700°C

700°C

tf= 38.65e-C(T) tag R² = 0.91

650°C

600°C

10 1

10

100

1000

10000

100000

Time tag , h

Fig. 10. Time to failure vs aging time curves at 600, 650 and 700°C. Filled data markers are experimental data. Open markers represent data recalculated by the time-temperature Eq. (1).

4. Summary and conclusions

Our previous study [23] demonstrated that M23C6 carbide lattice expansion at 600, 650 and 700°C is caused by diffusion of alloying elements into the carbide lattice. The equation for calculation of carbide M23C6 lattice parameter a value depending on a time and temperature was created, and which actually reflects diffusion processes in steel. One of the main objectives of present work was to find out whether a relationship could be found between this equation (1) and changes in mechanical properties, which are influenced by the same temperature - time steel aging factors.

The equation covers a relatively wide temperature range of 600-700°C, which involves also other diffusion processes when not only the M23C6 chemical composition changes but also the carbide size, dislocation density, Laves phase formation and other structural changes take place that affect the condition of steel and its properties. Microstructure comparison of the samples with different aging time and temperature shows that major changes correspond to the ones calculated by Eq. (1) equivalent time, when changes in M23C6 carbide lattice parameter slow down. In this case, at 650°C and especially at 700°C increase in carbide size and decrease in the number of small carbides is observed. At 600°C the Laves phase is formed but after 1·104 h, only slight change in growth of Laves phase particles is observed. Regardless of microstructure differences, which are more obvious at a longer aging duration, mechanical testing data at different aging duration and temperature demonstrated that when recalculating aging time by Eq. (1) for 600°C, the predicted data correlate well with experimental data for specimens with aging duration up to 2·104 h and that gives a possibility to forecast changes caused by material aging. The comparison of mechanical properties suggested that the strength values of P91 steel after long-term aging are comparable with those reported in the literature. This was also confirmed by the creep tests, which revealed a good correlation between predicted and experimental results under different aging conditions. However, in most cases, it was observed that experimental points for specimens aged at 600 and 700°C are often located below the regression line, which apparently is because of the differences in microstructure that were described above and by Monkman-Grant relationship, where 600°C data points were a little below the regression line. The relatively low number of experimental points does not allow more accurate assessment of possible deviations, but mainly experimental results demonstrate that the equation (1) provides a sufficiently good method to predict the mechanical behaviour of P91 steel during aging.

Acknowledgments This research was funded by a grant (No. MIP-023/2014) from the Research Council of Lithuania. References [1] J. Hald, Microstructure and long-term creep properties of 9–12% Cr steels, Int. J. Press. Vessels Piping 85 (1-2) (2008) 30–37. http://dx.doi.org/10.1016/j.ijpvp.2007.06.010. [2] A. Zieliński, J. Dobrzański, H. Purzyńska, G. Golański, Changes in Properties and Microstructure of High-chromium 9-12%Cr Steels due to Long-term Exposure at Elevated Temperature, Arch. Metall. Mater. 61 (2B) (2016) 957–964. http://dx.doi.org/10.1515/amm2016-0163.

[3] Ch. Pandey, A. Giri, M.M. Mahapatra, Evolution of phases in P91 steel in various heat treatment conditions and their effect on microstructure stability and mechanical properties, Mater. Sci. Eng. A 664 (2016) 58–74. http://dx.doi.org/10.1016/j.msea.2016.03.132. [4] M. Łomozik, A. Hernas, M.L. Zeman, Effect of welding thermal cycles on the structure and properties of simulated heat-affected zone areas in X10CrMoVNb9-1 (T91) steel at a state after 100,000 h of operation, Mater. Sci. Eng. A 637 (2015) 82–88. http://dx.doi.org/10.1016/j.msea.2015.04.009. [5] F. Masuyama, Creep rupture life and design factors for high-strength ferritic steels, Int. J. Press. Vessels Piping 84 (1-2) (2007) 53–61. http://dx.doi.org/10.1016/j.ijpvp.2006.09.006. [6] W. Bendick, L. Cipolla, J. Gabrel, and J. Hald, New ECCC assessment of creep rupture strength for steel gradeX10CrMoVNb9-1 (Grade 91), Int. J. Press. Vessels Piping 87 (6) (2010) 304–309. http://dx.doi.org/10.1016/j.ijpvp.2010.03.010. [7] Z. Abdallah, V. Gray, M. Whittaker and K. Perkins, A Critical Analysis of the Conventionally Employed Creep Lifting Methods, Materials 7 (5) (2014) 3371–3398. http://dx.doi.org/10.3390/ma7053371. [8] D. Šeruga, M. Fajdiga, M. Nagode, Creep Damage Calculation for Thermo Mechanical Fatigue, J. Mech. Eng. 57 (5) (2011) 371–378. http://dx.doi.org/10.5545/sv-jme.2010.108. [9] D. Seruga, and M. Nagode, Unification of the most commonly used time–temperature creep parameters, Mater. Sci. Eng. A 528 (6) (2011) 2804–2811. http://dx.doi.org/10.1016/j.msea.2010.12.034. [10] B. K. Choudhary, W.-G. Kim, M. D. Mathew, J. Jang, T. Jayakumar and Y.-H. Jeong, On the Reliability Assessment of Creep Life for Grade 91 Steel, Proc. Eng. 86 (2014 ) 335 – 341. http://dx.doi.org/10.1016/j.proeng.2014.11.046. [11] L. Cipolla, A. Di Gianfrancesco, G. Cumino, S. Caminada, Long Term Creep Behaviour and Microstructural Evolution of E911 Steel. ECCC Creep Conference, 12–14 September 2005, London, 288–299. [12] H. V. Atkinson, J. Veverkova, A. Strang, G. Marchant, G. McColvin, Microstructural Evolution in High Temperature Creep and Thermally Aged HA230. Proceedings of the 7th International Charles Parsons Turbine Conference: Power generation in an Era of Climate Change, 2007. [13] S.K. Mukhopadhyay, H. Roy, A. Roy, Development of hardness-based model for remaining life assessment of thermally loaded components, Int. J. Press. Vessels Piping 86 (4) (2009) 246– 251. http://dx.doi.org/10.1016/j.ijpvp.2008.11.014. [14] Sh. Fujibayashi, Life assessment of superheater tubes fabricated from 2.25Cr-1Mo steel. http://www.gruppofrattura.it/ocs/index.php/esis/ECF16/paper/view/7227/3718/, 2016 (accessed 12.05.16). [15] B.B. Jha, B.K. Mishra, B. Satpati, S.N. Ojha, Effect of thermal ageing on the evolution of microstructure and degradation of hardness of 2.25Cr-1Mo steel, Mater. Sci.-Poland 28 (1) (2010) 335–346. http://dx.doi.org/10.4028/www.scientific.net/DDF.303-304.85. [16] M. Hillert, Diffusion and Interface Control of Reactions in Alloys, Metall. Trans. A 6 (1) (1975) 5–19. http://dx.doi.org/10.1007/BF02673664. [17] B.K. Choudhary, K.B.S. Rao, S.L. Mannan, Steady state creep deformation behaviour of 9Cr1Mo ferritic steel forging in quenched and tempered condition, Trans. Indian Inst. Met. 52 (5) (1999) 327–336.

[18] T. M. Whittaker and B. Wilshire, Advanced Procedures for Long-Term Creep Data Prediction for 2.25 Chromium Steels, Metall. Mater. Trans. A 44 (1) (2013) 136–153. http://dx.doi.org/10.1007/s11661-012-1160-2. [19] B. Wilshire and P.J. Scharning, Long-term creep life prediction for high chromium steel, Scripta Mater. 56 (8) (2007) 701–704. http://dx.doi.org/10.1016/j.scriptamat.2006.12.033. [20] G. Sasikala, S.K. Ray, and S.L. Mannan, Kinetics of transformation of delta ferrite during creep in a type 316(N) stainless steel weld metal, Mater. Sci. Eng. A 359 (1-2) (2003) 86–90. http://dx.doi.org/10.1016/S0921-5093(03)00371-X. [21] W. Kraus and G. Nolze, Powder Cell for Windows, Version 2.4. Federal Institute for Materials Research and Testing, Berlin. http://www.ccp14.ac.uk/tutorial/powdcell/index.html/ 2008 (accessed 26.05.08). [22] S.D. Carpenter, D. Carpenter, and J.T.H. Pearce, XRD and electron microscope study of a heat treated 26.6% chromium white iron microstructure, Mater. Chem. Phys. 101 (1) (2007) 49–55. http://dx.doi.org/10.1016/j.matchemphys.2006.02.013. [23] A. Baltušnikas, I. Lukošiūtė, V. Makarevičius, R. Kriūkienė, A. Grybėnas, Influence of thermal exposure on structural changes of M23C6 carbide in P91 steel, J. Mater. Eng. Perform. 25 (5) (2016) 1945–1951. http://dx.doi.org/10.1007/s11665-016-2002-y. [24] T. Shrestha, S. F. Alsagabi, I. Charit, G. P. Potirniche and M.V. Glazoff, Effect of Heat Treatment on Microstructure and Hardness of Grade 91 Steel, Metals 5 (2015) 131–149. http://dx.doi.org/10.3390/met5010131. [25] W. Yan, W. Wang, Y. Shan, and K. Yang, Microstructural stability of 9-12Cr ferrite/martensite heat resistant steels, Front. Mater. Sci. 7(1) (2013) 1–27. http://dx.doi.org/10.1007/s11706-0130189-5. [26] C.G. Panait, W. Bendick, A. Fuchsmann, A.-F. Gourgues-Lorenzon, and J. Besson, Study of the microstructure of the Grade 91 steel after more than 100,000h of creep exposure at 600°C, Int. J. Press. Vessels Piping 87 (6) (2010) 326–335. http://dx.doi.org/10.1016/j.ijpvp.2010.03.017. [27] M. Taneike, K. Sawada, and F. Abe, Effect of Carbon Concentration on Precipitation Behavior of M23C6 Carbides and MX Carbonitrides in Martensitic 9Cr Steel during Heat Treatment, Metall. Mater. Trans. A 35 (4) (2004), 1255–1262. http://dx.doi.org/10.1007/s11661-004-0299x. [28] V. Thomas Paul, S. Saroja, M. Vijayalakshmi, Microstructural stability of modified 9Cr–1Mo steel during long term exposures at elevated temperatures, J. Nucl. Mater. 378 (3) (2008) 273– 281. http://dx.doi.org/10.1016/j.jnucmat.2008.06.033. [29] F. Abe, Creep rates and strengthening mechanisms in tungsten-strengthened 9Cr steels, Mater. Sci. Eng. A, 319 – 321(2001) 770 – 773. http://dx.doi.org/10.1016/S0921-5093(00)02002-5. [30] P. Šohaj, V. Jan, On. Dvořáček, Evaluation of microstructural stability of creep-resistant steels weld joints on the basis of a computational modeling. METAL 2010: 19th International Metallurgical and Materials Conference 18-20.5.2010, Rožnov pod Radhoštěm, Česká Republika. [31] J.-H. Baek, S.-H. Kim, Ch.-B. Lee, and D.-H. Hahn, Mechanical Properties and Microstructural Evolution of Modified 9Cr-1Mo Steel after Long-term Aging for 50,000 h, Met. Mater. Int. 15 (4) (2009) 565–573. http://dx.doi.org/10.1007/s12540-009-0565-y.

[32] F.C. Monkman, N.J. Grant, An empirical relationship between rupture life and minimum creep rate in creep-rupture tests, Proc. Am. Soc. Test Mater. 56 (1956) 593–620. [33] R. W. Swindeman, P. J Maziasz, and C. R. Brinkman, Aging Effects on the Creep-Rupture of 9Cr-1Mo-V Steel, Proceedings of International Joint Power Generation Conference, Miami, FL, July 23-26, 2000, IJPGC2000-15050. [34] K. Kimura, H. Kushima, K. Sawada, Long-term creep deformation property of modified 9Cr– 1Mo steel, Mater. Sci. Eng. A 510–511 (2009) 58–63. http://dx.doi.org/10.1016/j.msea.2008.04.095. [35] F. Dobes, K. Milicka, The relation between minimum creep rate and time to fracture, Metal Sci. 10 (11) (1976) 382–84. http://dx.doi.org/10.1080/03063453.1976.11683560.