Uploaded by José Eberte

Cr35Ni45 Radiant Tube Alloy: Microstructure & Life Assessment

advertisement
Engineering Failure Analysis 88 (2018) 63–72
Contents lists available at ScienceDirect
Engineering Failure Analysis
journal homepage: www.elsevier.com/locate/engfailanal
Microstructure evolution and residual life assessment of service
exposed Cr35Ni45 radiant tube alloy
Ruokang Song, Sujun Wu
T
⁎
School of Materials Science and Engineering, Beihang University (BUAA), Beijing 100191, PR China
AR TI CLE I NF O
AB S T R A CT
Keywords:
Radiant tube
Phase transformation
Carbide
Oxidation
Residual life
The degradation of material microstructure was frequently encountered in radiant tubes during
high temperature service, resulting in their premature failure. Effect of service time on the phase
transformation and oxidation behavior of the Cr35Ni45 radiant tubes was analyzed using scanning electron microscopy (SEM), electronic probe coupled with wavelength dispersive spectrometer, X-Ray diffraction (XRD) and transmission electron microscopy (TEM). The orthorhombic
M7C3 and the face-centered cubic NbC primary carbides in the as-cast tube transformed into the
face-centered cubic M23C6 and the face-centered cubic G phase, respectively, during service at
1000 °C. Metallographic observation showed that a continuous Cr2O3 layer and a discontinuous
SiO2 layer formed on the inner surface of the radiant tubes due to the oxidizing environments,
and a precipitate free region was found beneath the oxide layers. Mechanical properties of the
tubes decreased with the increasing degree of microstructure degradation as service time prolongs. Residual life assessment of the serviced tubes was conducted by nonlinear fitting
Larson–Miller parameter versus rupture testing stress. In spite of lower mechanical properties for
the serviced tubes compared to as-cast tube, residual life assessment indicates that they have
considerable residual life at the service temperature.
1. Introduction
Centrifugally casting of high-temperature austenitic alloys rich in Ni and Cr is often the technique of choice to fabricate radiant
tubes in the petrochemical industries. Typical examples such as HK40 (Cr25Ni20) and HP40 (Cr25Ni35) austenitic alloys have
traditionally been used for decades, owing to their superior strength to creep rupture and good corrosion resistance at operating
temperature [1–3]. However, requirements for higher productivity at increasing temperature (≥1000 °C) and the reduction in fuel
impose harsher operating conditions in many industrial sectors, which have raised the demand for improved performance of the
alloy. Later developed materials are proprietary alloys with higher alloying elements, such as the niobium modified HP steels and the
Cr35Ni45 alloy [4–6]. Although some of these newer grades have not yet consistently standardized, their maximized creep strength
and ductility, together with improved resistance to corrosion in oxidizing furnace environments, make these alloys be widely used for
fabrication of the tubes in cracking operations [7–11].
The high Cr and Ni content for above mentioned alloys promotes the formation of an as-cast microstructure composed of a
network of primary precipitates within an austenitic matrix. These precipitates can improve the resistance to high temperature creep
of the alloy by inhibiting dislocation motion and grain boundary sliding. During service at high temperature, however, precipitates
will evolve which results in changes in the mechanical properties. In cracking facilities, one of the key reliability aspects relates to
⁎
Corresponding author.
E-mail address: wusj@buaa.edu.cn (S. Wu).
https://doi.org/10.1016/j.engfailanal.2018.01.002
Received 1 September 2016; Received in revised form 19 April 2017; Accepted 12 January 2018
Available online 13 January 2018
1350-6307/ © 2018 Elsevier Ltd. All rights reserved.
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
ensuring the integrity and reliable operation of radiant tubes in harsh environments [12–16]. The tubes experience temperatures
about 1000 °C for a considerable length of time, which makes creep become the dominant mode of failure. Although radiant tubes
have a design life of ~15 years, premature failure of these tubes is very often observed as a result of microstructural degradation
accumulated with service time. The consequence of failure of a component in-use can be tragic and expensive.
In the past few decades, many researchers have studied the microstructural damage and described the mechanisms of the premature failures for the life prediction of tubes which are mainly fabricated by HK-grade and HP-grade alloys [17]. However, the data
related to the newly developed Cr35Ni45 alloy are comparatively rare. The aim of this work is to analyze the microstructure evolution and oxidation characteristics induced by service exposure of four centrifugally cast Cr35Ni45 radiant tubes (as-cast tube,
serviced for 13,100 h, 21,900 h and 52,500 h). Residual life assessment model was discussed and the residual life of the three serviced
tube was predicted.
2. Experimental procedure
The materials investigated are four centrifugally cast radiant tubes made of Cr35Ni45 austenitic alloy with chemical compositions
(at.%) of 2.25 C, 0.58 Nb, 36.77 Cr, 40.05 Ni, 3.07 Si and balance Fe. The tube without service exposure was named as as-cast tube,
while the other three tubes were named, based on their different service exposed time at high temperature around 1000 °C in
petrochemical plant, as tube-A for 13,100 h, tube-B for 21,900 h and tube-C for 52,500 h, respectively. Samples for metallographic
observation were prepared by mechanical polishing. The microstructural observation was performed on the section perpendicular to
the axis of tube using a SUPERTM 55 SEM and a JEOL8230 electronic probe coupled with wavelength dispersive spectrometer (WDS).
JEM-2100F TEM and Rigaku D/MAX-RB X-ray diffractometer were used to characterize the phase transformations after high temperature aging. The XRD analysis was performed from 20° to 90° at the speed of 6°/min with Cu Kα radiation, and the typical
radiation condition was set as 200 kV and 40 mA. The TEM specimens were prepared by mechanical dimpling and subsequent ion
milling. The tubes with different service time were ground for removing the oxide layers on the surfaces, and then electrolyzed as the
anode in DC stabilized power supply with a steel cathode. The electrolytic extraction of precipitates was carried out in 10%
HCl + 90% CH3OH (volume ratio) solution at 5 V for 12 h to extract enough powders for XRD analysis. Hardness distribution along
thickness of the tube wall was measured using a Vickers hardness tester FM-800 at a load of 300 gf with the duration of 15 s, and the
room-temperature tensile tests (MTS810) were carried out to determine the mechanical properties of four tubes with a strain rate of
1 × 10−3 s−1. Accelerated stress rupture test were conducted for tube-B according to HB 5150-1996 specification, at various combinations of testing temperatures (1000–1125 °C) and testing stresses (10–30 MPa) to obtain stress rupture data used in Larson–Miller
equation. Test specimens were cut along the longitudinal direction of the tube. The stress levels above the operating stress at each
temperature were selected in such a way so as to obtain rupture within a reasonable span of time. The hoop net sectional stress σh
acting on the tubes (D < 10 t) was calculated using the thick walled pressure vessel theory [18]:
2
D
⎧⎡
⎫
⎤ + 1⎪
⎪
D − 2(t − Lmax ) ⎦
P0
σh = ⎣
2
D
⎨
⎬
⎪ ⎡ D − 2(t − Lmax ) ⎤ − 1 ⎪
⎦
⎩⎣
⎭
(1)
where P0 (MPa) is the operating pressure, D is the diameter in mm and t is the thickness of the tube in mm, Lmax is total thickness
of oxide layers on inner and outer wall of tube-B. Life assessment of three serviced tubes was performed based on nonlinear fitting
Larson–Miller parameter versus testing stress.
3. Results
3.1. Phase transformation and microstructure evolution in the bulk material
SEM micrograph of the central area of the as-cast tube is shown in Fig. 1a in which the small frame picture with higher magnification is backscattered electron image. It can be seen that there are two kinds of primary eutectic phases in the austenitic matrix,
“bone-like” dark phase and “Chinese script-like” white phase precipitated in the interdendritic areas forming a fragmented network.
XRD pattern in Fig. 2a and WDS quantitative analysis result in Table 1 reveal that the white phase is face-centered cubic NbC and the
dark phase is orthorhombic M7C3 in which M mostly is Cr.
The service exposure at high temperature has an important influence on the microstructure of the bulk material. Fig. 1b–d demonstrates that the microstructure of the concerned material varied significantly with the service time. Compared with the as-cast
tube alloy, the most obvious morphology change in tube-A (Fig. 1b) is the precipitation of many small dispersive particles near the
dendritic boundaries. XRD analysis of the particles extracted from tube-A showed that the main carbide is M23C6 and the volume
fraction of NbC is too low to detected, as shown in Fig. 2b, which indicates that the orthorhombic dark M7C3 has transformed into the
face-centered cubic dark M23C6 and the white NbC has transformed into a grey phase (Fig. 1b). Diffraction information of the M23C6
carbide (Fig. 3a) shows this phase has a lattice parameter of 1.049 nm. Analysis of the grey phases in Fig. 1b, c and d using WDS
attached to the electronic probe proved that these grey phases are silicon-rich phases, similar to the composition of the G phase.
According to previous research [19], the face-centered cubic nickel–niobium silicide phase (a~1.13 nm) in HK-grade alloys was
named as G phase. The grey phase here is a similar silicon-rich phase which contains Ni, Nb and Cr. Selected area electron diffraction
(Fig. 3b) along [001] zone axis of the silicon-rich phase demonstrates that this phase also has a face-centered cubic structure with a
64
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Fig. 1. Secondary electron micrographs in the central region of the tubes accompanied by backscattered electron images at high magnification (small frames). (a) ascast tube; (b) tube-A; (c) tube-B; (d) tube-C.
Fig. 2. X-Ray diffraction results of extracted carbide particles for four samples. Note that austenite in pattern (b) is the result of inadequate electrolysis, and SiO2 in
pattern (d) is the result of internal oxidation of Si. (a) as-cast tube; (b) tube-A; (c) tube-B; (d) tube-C.
65
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Table 1
Composition of the dendrite and the precipitated phases in inter-dendritic regions of the four tubes analyzed by WDS (at.%), as indicated by arrows in Fig. 1.
Samples
Phases
C
Si
Cr
Ni
Fe
Nb
As-cast tube
Dendrite
M7C3
NbC
Dendrite
M23C6
G phase
Dendrite
M23C6
G phase
Dendrite
M23C6
G phase
1.32
34.48
38.01
1.87
21.57
3.67
1.65
24.62
7.05
2.14
24.26
4.17
2.60
0.06
0.79
3.11
0.00
11.18
2.19
0.01
15.67
2.05
0.00
15.07
34.67
62.45
8.17
30.55
71.01
31.86
26.28
67.45
32.99
22.50
62.82
34.59
44.64
0.88
7.32
41.49
3.63
30.51
44.09
3.18
29.55
46.81
4.53
30.62
16.60
1.67
2.24
22.92
3.72
9.70
25.61
4.66
2.39
26.38
8.34
2.64
0.17
0.46
43.46
0.15
0.07
13.08
0.18
0.08
12.35
0.12
0.05
12.91
Tube-A
Tube-B
Tube-C
Fig. 3. Bright field images and corresponding selected area electron diffraction patterns of tube-A showing (a) M23C6 phase along [321] zone axis and (b) G phase
along [001] zone axis, respectively, indicating a face-centered cubic structure of carbide and the face-centered cubic structure of G phase.
lattice parameter of 0.675 nm, which suggests the grey silicon-rich phase in serviced Cr35Ni45 tube alloy is G phase. Although the G
phases in Cr35Ni45 alloy and HK alloy have the same structure, G phases in the two alloys have different Cr/Ni ratios, about 1 for the
Cr35Ni45 and about 0.2 for the HK alloy [19, 20], resulting in different lattice parameters of the G phases. Comparison of the
microstructure in tube-A (Fig. 1b) and tube-B (Fig. 1c) indicates that the main precipitated phases and small dispersive particles have
coarsened over increased service time. The transformation of NbC to G phase for the two tubes has not finished as some of the
remnant white phases still exist. Fig. 1d shows that precipitated phases in tube-C have transformed into blocky shape particles
accompanying with some creep cavities in the matrix, and NbC phase has completely transformed into the grey G phase as no white
phase left.
3.2. Oxidation behavior of the tube's inner surface
The radiant tubes were exposed to the oxidizing environment at high temperature during service. It is very common in practice
that cracks derive from weak regions close to the tube's inner surface. Compared to the smooth inner surface of the as-cast tube
(Fig. 4a), the morphology of the inner surface of the serviced radiant tubes has significantly changed as illustrated in Fig. 4b–d.
Elements distribution close to the inner surface of tube-A was measured using the WDS mapping. The results are illustrated in Fig. 5.
It can be seen that the carbides and G phase particles disappeared and a continuous chromium-rich oxide layer (external oxidation)
along with a discontinuous silicon-rich oxide layer (internal oxidation) formed on the inner surface of tube-A. In the matrix beneath
the inner surface of the tube, elements of Cr and Si are depleted due to the formation of the oxides. XRD results of the oxide layers on
tube-A (in Fig. 6) show that the chromium-rich oxide is Cr2O3 and the silicon-rich oxide is SiO2. Before the continuous Cr2O3 layer
completely formed, oxidizing atmosphere at high temperature had led to inner-surface decarburization of the tube-A. Hence, elemental depletion of Cr, Si and C facilitated the dissolution of precipitates in the subsurface region of tube-A, made evident in this
region (precipitate free region, PFR) in which precipitated particles disappeared.
The morphology of the inner surface region of tube-B and tube-C is generally similar to that of tube-A, as illustrated in Fig. 4c and
d, except that the width of PFR showed an increasing trend with exposure time and some creep cavities have also formed in this
region. Such cavities normally form at the original location of the dissolved carbides and can develop into micro-cracks with creep
deformation, which will cause surface damage and mechanical deterioration. Note that particles inside the cavities in Fig. 4c and d
are contaminants collected in the specimen preparation process, which has confirmed by WDS analysis.
66
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Fig. 4. Secondary electron micrographs in the inner surface region of the tubes. (a) as-cast tube; (b) tube-A; (c) tube-B; (d) tube-C.
Fig. 5. Elemental distribution of tube-A in the inner-surface region along with the wavelength spectrometer crystal used for each of the elements.
3.3. Mechanical properties
Fig. 7a shows the comparison of the Vickers hardness between PFR and matrix. It can be seen that the microhardness in PFR is
lower than that in the matrix for three serviced tubes, resulting from the disappearance of precipitates and decarburization in PFR.
Compared with the as-cast tube, it appears that the microhardness of the matrix did not change significantly with the service
exposure, because the service exposure did not have significant influence on the austenitic matrix of the tubes owing to the protective
continuous Cr2O3 layer on the inner surface of the tubes. However, different hardness levels between PFR and matrix of the serviced
tubes will result in different expansion coefficients which can lead to initiation of cracks during temperature fluctuation.
Tensile properties of the tubes at room temperature are illustrated in Fig. 7b. It can be seen that the yield strength (Rp0.2, 0.2%
67
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Fig. 6. XRD analysis of oxide layers on the inner surface of tube-A. Note some powders of austenite matrix were included in the oxide powders during grinding process.
Fig. 7. Mechanical properties of the tubes. (a) Vickers hardness in PFR and matrix of the tubes, (b) variations of tensile properties with service time.
proof stress), tensile strength (Rm) and ductility (δ) decreased with the service exposure time which is consistent with microstructure
degradation discussed above. The loss in ductility is due to the variations of the type, shape and size of the precipitates, which is an
important parameter to consider during maintenance. If the ductility is too low, the tube cannot be repaired through welding before
an adequate healing heat treatment. It has been claimed that radiant tube alloys are very sensitive to weld cracking and G phase can
act as the nucleation point of weld cracks in serviced tubes [21,22].
4. Discussion
4.1. Phase formation and stability
Equilibrium phase diagram of the investigated as-cast alloy in Fig. 8 indicates that primary precipitated phases are M23C6 and G
phase under equilibrium state of thermodynamics. M23C6 and MC phases precipitate when the temperature decreases to 1300 °C and
1288 °C, respectively. The transformation of MC phase to G phase begins at 1030 °C and finishes at the temperature of ~1000 °C
under equilibrium cooling condition.
It is obvious that the microstructure of the as-cast tube discussed in Section 3.1 is inconsistent with the equilibrium phase diagram
which suggests the M23C6 and G phases as the primary precipitated phases. As shown in Table 1, there is significant composition
difference between dendritic and inter-dendritic regions for the as-cast tube. The inter-dendritic regions, formed at the last stage of
solidification, are enriched with Cr and C as compared to the average alloy composition. Considering the fact that Nb has a greater
affinity for C than Cr, it can replace the latter in carbide formation [19]. For lower Nb content of inter-dendritic regions in the as-cast
tube (Table 1), there is more C available to interact with Cr, resulting in a lower Cr/C ratio. Hence, primary eutectic M7C3 carbides
will precipitate instead of M23C6 in the thermodynamically nonequilibrium state, because centrifugal casting has less time for diffusion in the as-cast alloy. WDS analysis shows that G phase is a silicon-rich precipitate containing several metallic elements, such as
Ni, Cr and Nb, in serviced tube alloys. Thermodynamically, the NbC phase becomes thermally unstable when temperature is below
~1000 °C, and the transformation of NbC to G phase can occur as shown in Fig. 1. During casting of the tubes, however, the
transformation cannot occur because it involves alloying elements incorporation and discharge of C, and the fast cooling casting
68
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Fig. 8. Equilibrium phase diagram of Cr35Ni45 alloy calculated using JMatPro.
process does not give the elements enough time for diffusion. Therefore, the NbC phase is remained and no G phase formed in the ascast tubes. As described above, the primary structure of cast tube alloy is locally out of equilibrium due to the segregation of elements.
The non-equilibrium carbides of M7C3 and NbC will interact with the matrix to form the equilibrium M23C6 and G phases at elevated
temperatures for the three serviced tubes. Similar transformation of carbides has been found in 35Cr-45Ni alloy during laboratoryaging, as reported in Ref. [21].
Microstructure evolution during service exposure can be attributed to the instability of the precipitated phases at high service
temperature. The reactions between Cr and C accompanied by their respective Gibbs free energy changes are expressed as follows
[23–26]:
7Cr + 3C = Cr7C3 , △G Θ (KJ mol−1) = −174 − 0.0259 T (298 K < T < 2171 K)
(2)
23Cr + 6C = Cr23C6 , △G Θ (KJ mol−1) = −309.616 − 0.0774 T (298 K < T < 1773 K)
(3)
Θ
From Eqs. (2) and (3), it can be calculated that the value of Gibbs free energy change △G for Cr23C6 is more negative than that
for Cr7C3 which indicates that Cr23C6 is more stable than Cr7C3 under equilibrium condition. Hence, during high temperature service
aging, the metastable M7C3 that primitively precipitated in matrix due to relatively fast cooling would transform into stable M23C6
carbide, together with precipitation of small M23C6 particles from matrix. Expelled C atoms from the transformations of M7C3 to
M23C6 and NbC to G phase will promote precipitation of the M23C6 phase, which can explain why the majority of small dispersive
M23C6 particles emerged in the vicinity of the prior precipitated phases. The small instable M23C6 particles tend to coarsen into blocky
M23C6 carbides during long time high temperature service (see Fig. 1d).
As discussed above, the NbC phase will transform into thermodynamically more stable G phase at the temperature below
~1000 °C under equilibrium condition. Under casting condition, however, the transformation cannot occur and the NbC phase
remains, as shown in Fig. 1a. During high temperature service aging, the transformation of NbC phase to G phase would occur.
However, the transformation rate is relatively low since the transformation process involves alloying elements incorporation and
diffusion of C atoms. According to previous investigations [27, 28], the metastable NbC phase would transform into G phase at a
certain range of temperature (700–1000 °C). It may be concluded that a minimum temperature of 700 °C is kinetically necessary for
the elements diffusion during phase transformation.
4.2. Thermodynamical criterion of external oxidation
The formation of external Cr2O3 layer and internal SiO2 layer on the tube's inner surface can be attributed to the selective
oxidation of Cr and Si. According to the oxidation theory proposed by Wagner [29–31], the concentration of an element B in the alloy
needs to be higher than a critical concentration NB∗ to form a continuous external oxide layer on the inner surface, or it will be
oxidized internally. The critical concentration NB∗ can be determined by the following formulas:
1
πg ∗ ⎞ NoS Do⋅Vm ⎤ 2
NB∗ = ⎡ ⎛
⋅
⎢ ⎝ 2ν ⎠ DB ⋅VOX ⎥
⎣
⎦
(4)
V
g ∗ = f⋅ OX
Vm
(5)
Do = 2.567 × 108 exp ⎛−
⎝
where g
⁎
165.32KJ
⎞ m2/ s
RT ⎠
(6)
is critical volume fraction of the oxide, ν equals half of the valence number of element B in the oxide. NS0 (=0.02%) is
69
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Table 2
The parameters used in Eq. (4) for calculation of critical concentration NB∗.
Elements/parameters
ν
f
Vox/Vm
g⁎
DB
NB∗
Cr
Si
1.5
2
0.354
0.016
2.22
3.16
0.786
0.051
9.1 × 10−14 m2/s [34]
2.12 × 10−13 m2/s [35]
18.60 at.%
2.25 at.%
molar fraction of oxygen on the alloy inner surface in the cracking practice, Do and DB is the diffusion coefficient of oxygen and
element B in the alloy at 1000 °C, Vm and Vox is the molar volume of alloy and the oxide of element B, f is molar fraction of oxide of
the element B in the alloy. Note that several parameters of the Cr and Si needed in Wagner theory for Cr35Ni45 tube alloy are not
available in the literature since that this alloy has never been deeply investigated, and therefore the corresponding parameters of Cr
and Si in similar alloys are used for the estimation of the critical content of Cr and Si to form continuous oxidation layer on the inner
surface of the tubes. The diffusion coefficient of oxygen Do is calculated as 4.248 × 10−11 m2/s at 1000 °C using the Eq. (6) for steels
[32, 33]. Table 2 shows the other parameters used in Eq. (4) and the calculated critical concentration of Cr and Si.
As mentioned above, the concentration of Cr in the Cr35Ni45 alloy (35.44 at.%) is much higher than the critical concentration
NCr∗ and the concentration of Si (1.60 at.%) is lower than NSi∗. Therefore, the external continuous Cr2O3 oxide layer is formed on the
inner surface of the tubes. The SiO2 particles, however, are discontinuously distributed beneath the Cr2O3 layer, because small
amount of SiO2 particles cannot link together duo to low concentration of Si. Cr with high concentration was fast oxidized when Si
close to the inner surface of the tubes was depleted. Gibbs free energy of formation of Cr2O3 and SiO2 can be described by the Eqs. (7)
and (8):
Θ
ΔGC r2O3 = ΔGCr
− RT ln PO2
2 O3
(7)
Θ
ΔG SiO2 = ΔGSiO
− RT ln PO2
2
(8)
Θ
As we know, standard Gibbs free energy of formation of SiO2 is lower than Cr2O3 (ΔGSiO2 < ΔGCr2O3Θ), so the Gibbs free energy
variation to form SiO2 (ΔGSiO2Θ) is lower than that of Cr2O3 (ΔGCr2O3Θ) at same oxygen partial pressure. When the external continuous Cr2O3 layer was thick enough, oxygen partial pressure beneath the Cr2O3 layer would reach a critical level which led to
ΔGCr2O3Θ = 0, and Cr2O3 layer would stop thickening. However, ΔGSiO2Θ is still < 0 at this critical oxygen partial pressure, which
accounts for the formation of branched SiO2 oxide particles along the grain boundaries in PFR. In addition, oxidation of Cr and Si
resulted in elemental depletion in PFR, which combined with decarburization, caused precipitates dissolution in this region.
The external continuous Cr2O3 layer on the inner surface has some protective effect for the tube through isolating the base metal
from severe environment condition in the tube during ethylene cracking at high temperature. However, thermal cycles, decoking,
erosion and temperature gradient in inner surface often cause the breakup of oxide Cr2O3 layer. According to above discussion, a
protective Cr2O3 oxide layer can rebuild as long as the Cr content in the alloy matrix adjacent to the inner surface is > 18.60 at.%. In
this work, WDS analysis shows that the atomic percentages of Cr in subsurface regions for three serviced tubes are 25%, 21% and
19.7% in sequence which are larger than the critical concentration NCr∗. Hence, the Cr2O3 oxide layers for the three serviced tubes
were still intact and can protect the tubes from severe microstructural degradation. In addition, high Cr content in the tubes can make
fast formation and self-repair of a dense Cr2O3 oxide layer on the inner surface of the tube that prevents the carburization, initiation
of graphite filament and coke growth, resisting severe metal dusting degradation. Therefore, there is no obvious attack of metal
dusting found in the three tubes after different service time, as shown in Fig. 4b–d. Based on the Petkovic-Luton and Ramanarayanan
model [36], it can be considered that tube-A and tube-B experienced moderate microstructural degradation to approximately Stage II,
whereas tube-C to Stage III. Attempts to correlate the microstructural changes to residual life are of great significance to industry
operations. However, one must bear in mind that the microstructure evolution depends on the tube materials, working condition
including temperature and pressure, and the chemical materials inside the tube. The relationship between microstructure degradation and the residual life for one service condition may not work for other service situations.
4.3. Residual life assessment
As discussed above, it is complicated to correlate the microstructural change to residual life of the serviced tubes. In order to
analyze the degree of mechanical degradation of the serviced tubes and determine their residual service life, the accelerated stress
rupture tests of the tube-B were carried out in the temperature range of 1000–1125 °C at various stress levels (10–30 MPa). These test
parameters were chosen to obtain desired rupture time with high extrapolation capability for the tube. Residual life assessment was
made using Larson–Miller parameter (LMP) with the input of the stress rupture time test results in the LMP equation. The LMP
equation is given by Eq. (9) [37]:
LMP = T (C + log tr ) = a 0 + a1(log σ ) + a2 (log σ )2 + …+ an (log σ )n
(9)
where C is a constant of the material, σ is the applied stress in MPa, tr is the stress rupture time in hours and T is the temperature. In
the present investigation, Cubic Polynomial Model (n = 3) is selected, and the value of C = 16.94 is determined by linear fitting the
experimental data at constant stress. Fig. 9 shows the results of stress rupture test of tube-B given in Logarithmic stress-LMP plots,
along with the polynomial (n = 3) fitting curve. The hoop stress σh acting on the tube-B during service was calculated as 2.538 MPa
70
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
Fig. 9. Applied stress-LMP relationship with fitting curve defined by the Cubic Polynomial Model.
by using the Eq. (1), hence, LMP = 27,879 can be obtained. Such a polynomial fitting curve can be used in conjunction with
operating temperature and operating stress to estimate the residual life of other serviced tubes fabricated by Cr35Ni45 alloy.
The residual life of the investigated three tubes at the service condition can be predicted using the model in Fig. 9. Considering the
working temperature of 1000 °C and stress of 2.538 MPa, the residual life of the serviced tubes are obtained as 11.2 years for tube-A,
10.2 years for tube-B and 6.7 years for tube-C respectively, provided that the main damage mechanism for the tube is creep. It can be
concluded that significant useful life is still available for the three serviced tubes in spite of the reduction in mechanical properties
compared to as-cast tube. The decrease of residual life for the three serviced tubes is related to their microstructure degradation. It is
reported that the interface between the G-phase and the matrix is a preferential site for nucleation of creep cracks [21]. Fig. 1 shows
that precipitated phases in inter-dendritic regions have gradually coarsened and formed a continuous network with increment of
service time which could significantly facilitate the initiation of creep cracks, resulting in a lower value of residual life. Therefore, the
value of residual life for three tubes decreases with prolonging of service time. The comparison between the experimental data and
predicted value, as shown in Fig. 10, reveals that the Cubic Polynomial Model in this investigation presents a good degree of
confidence and high agreement. It should be noted, however, that above theoretical assessment is strongly influenced by the fluctuant
operating parameters (service pressure and temperature) in actual operation. The majority of premature failures caused by premature
creep result from overheating, therefore, further work need to be done to determine an appropriate safety factor.
5. Conclusions
Detailed study on the deterioration of microstructure and mechanical properties, combined with life assessment, was conducted
for service exposed Cr35Ni45 radiant tubes. Based on the results and analysis, several conclusions can be drawn as follows.
(1) The microstructure of as-cast tube is composed of austenitic matrix and a network of carbides, “Chinese script-like” white NbC
and “bone-like” dark M7C3, precipitated in the interdendritic areas. The transformations of M7C3 to M23C6 and NbC to G phase
Fig. 10. Data of stress rupture test results of the tube-B and the prediction curves, indicating high agreement between experimental data and assessed data.
71
Engineering Failure Analysis 88 (2018) 63–72
R. Song, S. Wu
occurred during high temperature service.
(2) A continuous Cr2O3 layer was formed on inner surface of the tubes after long time high temperature service. A discontinuous SiO2
layer was also formed beneath it, penetrating inside along the grain boundaries. The continuous Cr2O3 layer has some protective
effect for the tube through isolating the base metal from severe environment condition.
(3) Mechanical properties of the tubes deteriorated with the increase of the service time. Residual life assessment based on accelerated stress rupture tests of tube-B indicates that considerable service life is still available for the three serviced tubes.
Acknowledgements
This work was performed with the support of University of Science and Technology Beijing, China Special Equipment Inspection
Institute and National High-tech R&D Program (863 Program): 2012AA03A513.
References
[1] S.R. Allahkaram, S. Borjali, H. Khosravi, Investigation of weldability and property changes of high pressure heat-resistant cast stainless steel tubes used in
pyrolysis furnaces after a five-year service, Mater. Des. 33 (2012) 476–484.
[2] S. Ling, T. Ramanarayanan, R. Petkovic-Luton, Computational modeling of mixed oxidation-carburization processes: part 1, Oxid. Met. 40 (1993) 179–196.
[3] S. Borjali, S.R. Allahkaram, H. Khosravi, Effects of working temperature and carbon diffusion on the microstructure of high pressure heat-resistant stainless steel
tubes used in pyrolysis furnaces during service condition, Mater. Des. 34 (2012) 65–73.
[4] M. Whittaker, B. Wilshire, J. Brear, Creep fracture of the centrifugally-cast superaustenitic steels, HK40 and HP40, Mater. Sci. Eng. A 580 (2013) 391–396.
[5] W. Horvath, B. Tabernig, E. Werner, P. Uggowitzer, Microstructures and yield strength of nitrogen alloyed super duplex steels, Acta Mater. 45 (1997)
1645–1654.
[6] S.Y. Kondrat'ev, V.S. Kraposhin, G.P. Anastasiadi, A.L. Talis, Experimental observation and crystallographic description of M7C3 carbide transformation in
Fe–Cr–Ni–C HP type alloy, Acta Mater. 100 (2015) 275–281.
[7] A. Kaya, P. Krauklis, D. Young, Microstructure of HK40 alloy after high temperature service in oxidizing/carburizing environment: I. Oxidation phenomena and
propagation of a crack, Mater. Charact. 49 (2002) 11–21.
[8] A.K. Ray, S. Kumar, G. Krishna, M. Gunjan, B. Goswami, S.C. Bose, Microstructural studies and remnant life assessment of eleven years service exposed reformer
tube, Mater. Sci. Eng. A 529 (2011) 102–112.
[9] W.-G. Kim, J.-Y. Park, S.-J. Kim, J. Jang, Reliability assessment of creep rupture life for Gr. 91 steel, Mater. Des. 51 (2013) 1045–1051.
[10] S. Li, Y. Wang, S. Li, H. Zhang, F. Xue, X. Wang, Microstructures and mechanical properties of cast austenite stainless steels after long-term thermal aging at low
temperature, Mater. Des. 50 (2013) 886–892.
[11] A.K. Ray, Y. Tiwari, P. Roy, S. Chaudhuri, S. Bose, R. Ghosh, J. Whittenberger, Creep rupture analysis and remaining life assessment of 2.25 Cr–1Mo steel tubes
from a thermal power plant, Mater. Sci. Eng. A 454 (2007) 679–684.
[12] S. Sivaprasad, J. Swaminathan, Y. Tiwary, P. Roy, R. Singh, Remaining life assessment of service exposed reactor and distillation column materials of a
petrochemical plant, Eng. Fail. Anal. 10 (2003) 275–289.
[13] C. Liu, Y. Chen, Variations of the microstructure and mechanical properties of HP40Nb hydrogen reformer tube with time at elevated temperature, Mater. Des.
32 (2011) 2507–2512.
[14] N. Parnian, Failure analysis of austenitic stainless steel tubes in a gas fired steam heater, Mater. Des. 36 (2012) 788–795.
[15] A. Ribeiro, L. De Almeida, D. Dos Santos, D. Fruchart, G. Bobrovnitchii, Microstructural modifications induced by hydrogen in a heat resistant steel type HP-45
with Nb and Ti additions, J. Alloys Compd. 356 (2003) 693–696.
[16] R. Voicu, E. Andrieu, D. Poquillon, J. Furtado, J. Lacaze, Microstructure evolution of HP40-Nb alloys during aging under air at 1000°C, Mater. Charact. 60 (2009)
1020–1027.
[17] W. Wang, F. Xuan, Z. Wang, B. Wang, C. Liu, Effect of overheating temperature on the microstructure and creep behavior of HP40Nb alloy, Mater. Des. 32 (2011)
4010–4016.
[18] G. Sinclair, J. Helms, A review of simple formulae for elastic hoop stresses in cylindrical and spherical pressure vessels: what can be used when, Int. J. Press.
Vessel. Pip. 128 (2015) 1–7.
[19] G.D. de Almeida Soares, L.H. de Almeida, T.L. da Silveira, I. Le May, Niobium additions in HP heat-resistant cast stainless steels, Mater. Charact. 29 (1992)
387–396.
[20] E. Kenik, P. Maziasz, R. Swindeman, J. Cervenka, D. May, Structure and phase stability in a cast modified-HP austenite after long-term aging, Scr. Mater. 49
(2003) 117–122.
[21] I.A. Sustaita-Torres, S. Haro-Rodríguez, M.P. Guerrero-Mata, M. de la Garza, E. Valdés, F. Deschaux-Beaume, R. Colás, Aging of a cast 35Cr–45Ni heat resistant
alloy, Mater. Chem. Phys. 133 (2012) 1018–1023.
[22] G.D. Barbabela, L.H.D. Almeida, T.L.D. Silveira, I.L. May, Role of Nb in modifying the microstructure of heat-resistant cast HP steel, Mater. Charact. 26 (1991)
193–197.
[23] Y. Li, Y. Gao, B. Xiao, T. Min, Y. Yang, S. Ma, D. Yi, The electronic, mechanical properties and theoretical hardness of chromium carbides by first-principles
calculations, J. Alloys Compd. 509 (2011) 5242–5249.
[24] Q. Wu, W. Li, N. Zhong, W. Gang, W. Haishan, Microstructure and wear behavior of laser cladding VC–Cr7C3 ceramic coating on steel substrate, Mater. Des. 49
(2013) 10–18.
[25] H. Kleykamp, Thermodynamic studies on chromium carbides by the electromotive force (emf) method, J. Alloys Compd. 321 (2001) 138–145.
[26] F. Cheng, Y. Wang, T. Yang, Microstructure and wear properties of Fe–VC–Cr7C3 composite coating on surface of cast steel, Mater. Charact. 59 (2008) 488–492.
[27] R.A.P. Ibañez, G.D. de Almeida Soares, L.H. de Almeida, I. Le May, Effects of Si content on the microstructure of modified-HP austenitic steels, Mater. Charact. 30
(1993) 243–249.
[28] G.D. Barbabela, L.H. de Almeida, T.L. da Silveira, I. Le May, Phase characterization in two centrifugally cast HK stainless steel tubes, Mater. Charact. 26
(1991) 1–7.
[29] H. Liu, Y. He, L. Li, Application of thermodynamics and Wagner model on two problems in continuous hot-dip galvanizing, Appl. Surf. Sci. 256 (2009)
1399–1403.
[30] A. Madeshia, An amendment to the classical model of internal oxidation: model-inherent transition characteristics, Corros. Sci. 68 (2013) 111–118.
[31] J.B.L.A.M. Pignol, D. Huin, Predicting the Transition From Internal to External Oxidation of Alloys Using an Extended Wagner Model, CR. Mecanique 341 (2013)
314–322.
[32] F. Kang, Y. Geng, S. Chen, Investigation of internal oxidation on Pt-Ir-Zr high-temperature alloy, Rare Metal Mater. Eng. 40 (2011) 585–589.
[33] C.H. Konrad, L. Fuhrmann, R. Völkl, U. Glatzel, Internal oxidation with significant contribution of oxygen diffusion through the oxide phase, Corros. Sci. 63
(2012) 187–192.
[34] H. Daruvala, K. Bube, Tracer diffusion of chromium in 304 stainless steel, Mater. Sci. Eng. 41 (1979) 293–295.
[35] A. Atkinson, A theoretical analysis of the oxidation of Fe-Si alloys, Corros. Sci. 22 (1982) 87–102.
[36] R. Petkovic-Luton, T. Ramanarayanan, Mixed-oxidant attack of high-temperature alloys in carbon-and oxygen-containing environments, Oxid. Met. 34 (1990)
381–400.
[37] A.K. Ray, Y. Tiwari, S. Chaudhuri, Evaluation of mechanical properties and assessment of residual life of a service-exposed water wall tube, Eng. Fail. Anal. 7
(2000) 393–402.
72
Download