A universal function of creep rate
Li Jing-Tiana), Rong Xi-Mingb), Wang Jian-Luc), Zhang Bang-Qiangc), Ning Xi-Jing†a)
Institute of Modern Physics, Department of Nuclear Science and Technology, Fudan University, Shanghai 200433, China
Department of Optical Science and Engineer, Fudan University, Shanghai 200433, China
Dongfang Turbine Co. Ltd., Deyang 618000, China

Corresponding author. E-mail: xjning@fudan.edu.cn

*Project supported by the National Natural Science Foundation of China (Grant Nos. 11274073 and 51071048), the Shanghai Leading Academic Discipline Project, China (Grant No. B107), and the Key Discipline Innovative Training Program of Fudan University, China.

Abstract

In this paper, we derive a universal function from a model based on statistical mechanics developed recently, and show that the function is well fitted to all the available experimental data which cannot be described by any function previously established. With the function predicting creep rate, it is unnecessary to consider which creep mechanism dominates the process, but only perform several experiments to determine the three constants in the function. It is expected that the new function would be widely used in industry in the future.

PACS: 34.20.Cf; 62.20.Hg; 81.05.Bx
Keyword: creep; statistical mechanics; metal and alloys
1. Introduction

Creep is the progressive time-dependent plastic deformation of metals under load and elevated temperature conditions, [1] and the rate of creep determines the service lifetime of given materials in engineering. However, theoretically predicting the rate of creep without empirical data is impossible[13] and experimental measurement is time consuming because creep resistance is defined by a stress level required to produce a nominal strain, say 0.1%, 0.2%, or 0.5% in periods of about 100000  h.[2, 4, 5] In fact, creep rate depends not only on the applied stress but also on the ambient temperature, while the stress and temperature may differ greatly in different usages. Thus, to evaluate the creep lifetime of a given material, the creep rate test should cover the combination of various stresses and temperatures entirely. Clearly, this is an impossible task and experimental measurements of creep rate have to be limited to constant stress and constant temperature.[612]

If a universal function can be established to describe the dependence of creep rate at stress and temperature, the creep rate of any condition can be obtained by only several measurements performed at limited points of stresses and temperatures. Since the 1920s, much theoretical effort has been made to find such a function, [10, 1316] and several equations are established as follows:

where , σ , Qc, R, and T are the minimum creep rate, the applied stress, the apparent creep activation energy, the gas constant, and the absolute temperature, respectively, while As (s = 1, 2, … , 6), B, and n are constant for a given material.

In 1965, Monkman and Grant found that the creep rupture time trup multiplied by ɛ is a constant for many metals:[17]

Based on Eq.  (7), the Larson– Miller parameter method[18] was developed from Eq.  (4) to predict the creep life and is being used widely to predict long-term creep life from short-term measurement data via

where LMP depends on the applied stress.[19] If the units of T and trup are Kelvin and hour, respectively, the value of CLM was found to be around 20 for many materials and was expected to be the same for different materials.[20] However, recent experimental works showed that the value varies obviously for different materials.[2126] Even for a given material, it is shown via Eq.  (5) that the value of CLM is proportional to Qc with high correlation coefficient rather than a constant.[27] According to Eq.  (6), Wilshire et al.[4, 5, 28] showed that CLM depends on Qc, and the extrapolation always overestimates long-term performance with CLM being a constant. For instance, if only the short-term data for creep measurements within 1000  h are applied to determine the constant of Eq.  (8), then the prediction of the rupture strength for P92 alloys at 600  ° C by extrapolation to 100000  h is 180  MPa. However, if longer measurements data (> 3000  h) are used in Eq.  (8), the prediction under the same conditions changes to 128  MPa, obviously smaller than the former prediction.[29]

If a creep rate equation, such as the creep constitute equation [Eq.  (6)], is universal for different materials, CLM can be reasonably derived via Eq.  (7) and the prediction of rupture time will be reliable. However, previous experiments showed that none of Eqs.  (1)– (6) are universal. Taking the creep constitute equation as an example, recent experiments showed that n and Qc depend obviously on stress and temperature in some cases.[5, 8, 30, 31] For example, the value of n drawn from the experiment for K435[30] increases from 3.8 to 11.0 when the stress varies from 140  MPa to 250  MPa, while for polycrystalline copper[10] (or AZ31[31]) the values increase from 1 to 4.5 (or 3 to 5) when the stress varies, and for IN738LC[30] (or TC11 alloy[8]) the value of Qc increases from 420  kJ/mol to 730  kJ/mol (or 67  kJ/mol to 196  kJ/mol) with the temperature increasing. The experimental data for K435 and IN738LC are shown in Fig.  1, where the values of n and Qc fitted by Eq.  (6) are also presented. Clearly, a universal and reliable function of creep rate upon the stress and temperature is imperative for predicting creep life.

Fig.  1. The experimental curves (a) and curves (b) for In738LC and K435 alloys.
In our previous works, a model based on statistical mechanics was developed successfully to predict the stability of materials, atom diffusion rates, etc.[3238] In the present work, it was developed to predict the creep rates and a universal function was derived. In order to test the function, we measured the creep rates of three kinds of materials 10305DA, 10305BA, 10705BU, and collected available experimental data of other materials (IN738LC, [30] K435, [30] NiAl-25Cr, [39] NiAl-28Cr-5Mo, [6] Ti-45Al-2Mn-2Nb, [7] AS31, [9] TC11, [8] Polycrystalline Cu, [10] FGH95, [12] TMW2, [40] 1.25Cr0.5Mo[41]). It was shown that the experimental data are in good agreement with our new function.

2. Model and method

Creep mechanism can be classified as diffusion and dislocation.[42] In any case, when stress is applied on bulk metals, some atoms in a potential valley must cross over a series of barriers of height E0 to reach another valley along the direction of the applied force [Fig.  2], otherwise creep plastic deformation cannot take place. In the absence of applied stress, thermal motion can also make the atoms cross over the barrier with a small probability, but this motion does not lead to macro-deformation because the motion is spacially isotropic. In what follows, we describe how the isotropic motion is affected by the applied stress to result in creep and give the expression of creep rate.

Fig.  2. Schematic diagram of creep caused by a stress σ applied along the x direction (a) and the potential barriers felt by an atom (b).

According to a statical model of a single atom mentioned in Refs.  [32]–   [38], if we observe the motion of an atom in a potential valley in one time unit [Fig.  2(b)], the total time for the atom with kinetic energy ɛ larger than the static barrier E0 is

where (the partition function) and T is the temperature. Assuming that all the hopping events start when the atom stays in the center of the potential well described by V(x) along the minimum energy path, which should be the most probable, the time taken by the atom to cross over the barrier is

which results from Newton’ s law for the atom moving in the potential well. The averaged period can be obtained via

Hence, the averaged hopping frequency is given by

When a stress σ is applied along the x direction, the atom feels a force F, and work of the force is Fa if the atom moves from x0 to x1 [Fig.  2], leading to the increase of the total energy from ɛ up to ɛ + Fa. If the total energy ɛ > E0Fa, the atom is able to cross over the barrier E0. Thus, in one time unit, the total time for the atom having energy larger than E0 is

Correspondingly, the time taken by the atom to cross over the barrier (from x0 to x1) is

and the average period can be obtained via

so the average hopping frequency is given by

and the creep rate is expressed as

where k is a constant related to microstructures of the specific materials, such as the density of defects (dislocations or grain boundaries). In principle, if the density of defects and the microstructures are exactly obtained in advance, then the diffusion potential function V(x) can be obtained theoretically, and the creep rate can be exactly determined by Eq.  (17). Unfortunately, this method is time-consuming because the potential functions may be very different for different defects with different microstructures.

For applications of the above equations in practice, the creep of a column shaped material of mass M can be regarded as the departure of two mass centers [Fig.  2(a)] with each of mass M/2, and an average potential that the mass center feels is described by

where E0 and a are the potential barrier and the half width of the potential valley, respectively. Based on the statics model, [3238] equations  (13)– (17) are still valid for the mass center. Thus, we can determine the three constants, E0, a, and k by measuring creep rate at several temperatures and stress points, and then apply Eq.  (17) to predict creep life for any temperature and stress.

3. Result and discussion

There are two methods to validate the accuracy and the universality of our model. The first method is to check whether E0, a, and k maintain as constants and are independent of stress and temperature when equation  (17) is applied to fit the available experimental data of the creep rate. The second method is to check whether equation  (17) with E0, a, and k is unchanged because a given material can reproduce previous creep rate equations, such as Eqs.  (1) and (2), which present different forms for the same material under different loads (or temperature) conditions.

To validate the accuracy of our model sufficiently, we collected a large number of experimental data of creep rates[6, 810, 30, 40, 41] upon stress σ and temperature T, and the curves of and are shown in Fig.  3. In addition, we performed creep tests of 10305DA (42CrMo), 10705BA (2Cr12NiW1Mo1V), 10705BU (1Cr12Mo) in air with different loads and temperatures. The alloy ingot was cut into several pieces of cylinder with a diameter of 10  mm and a gauge length of 100  mm. Each test cost several thousand hours until the failure point. The experimental data are listed in Tables  1– 3, while the and curves are shown in Figs.  3(a) and 3(b), respectively. When fitting the experimental data with Eq.  (17), the force F applied on a single atom within the material was regarded as the average of the total force applied on the vertical cross section divided by all the atoms in the cross section. As shown in Fig.  3, our model can reproduce all the experimental data with E0, a, and k (listed in Table  4) unchanged for given materials. The potential barriers for these materials are ∼ eV, which approach the apparent creep activation energy Qc drawn from previous experiments fitted with Eq.  (4).[6, 810, 30, 40, 41] The half widths of the potential valley obtained from our model are of tens of nanometers which is in agreement with a creep mechanism by Seitz, who proposed that isolated vacancies may condense to form plates or disks[43] in the process of plastic deformation and Dash estimated that the diameter of the disks is about 15  nm according to experimental data.[44] This means that the atoms in the disks would feel a potential valley of tens of nanometers in width when creep occurs.

Fig.  3. The experimental data of ((a), (c), and (e)) and ((b), (d), and (f)) (black dots, squares, triangles, inverted triangles, rhombi, pentagons) and the curves (black line) obtained by fitting Eq.  (17) to 10305DA, 10705BA, 10705BU, IN738LC, K435, NiAl-25Cr, NiAl-28Cr-5Mo, Ti-45Al-2Mn-2Nb, AS31, TC11, Polycrystalline Cu, FGH95, TMW2, 1.25Cr0.5Mo.

Table 1. The experimental creep rates of 10305DA (42CrMoA) alloy measured in the present work.
Table 2. The experimental creep rates of 10705BU (2Cr12NiW1Mo1V) alloy measured in the present work.
Table 3. The experimental creep rates of 10705BA (1Cr12Mo) alloy measured in the present work.
Table 4. The average potential barriers of various kinds of alloys.

It is notable that some of the traditional creep rate equations correspond to different creep mechanisms. For example, the stress power law, described by Eq.  (1), is considered to correspond to the climb-controlled creep process when the load is small, while equation  (2) is applicable to glide-controlled creep under a large load. Obviously, even if the parameters A1 and n of Eq.  (1) are determined by some experiments under certain service conditions, we cannot use the equation to predict the creep rate for other service conditions because the value of n may change obviously. For example, the value of n for polycrystalline copper increases from 1 to 4.5 even if the temperature remains unchanged but the stress increases only from 2  MPa to 20  MPa, which is attributed to the transition of the creep mechanism from diffusion to dislocation.[10] For IN738LC, the value of n increases from 3.3 to 8.3 when the creep mechanism transits from climb-plus-glide to Orawan mechanism.[45]

Using our model, it is unnecessary to distinguish which mechanism dominates the creep and we can reproduce Eqs.  (1)– (6) if specific ranges of stress and temperature are given. For example, the experimental curves for high pure aluminium with stress below 10  MPa can be well described by Eq.  (1), but this power law breaks down when the stress surpasses 10  MPa.[46] Thus, when the stress changes from 5  MPa to 20  MPa, the experimental curves shown in Fig.  4 satisfies neither Eq.  (1) nor Eq.  (2). Nevertheless, the experimental curve at 598  K can be well fitted (Fig.  4) by Eq.  (17), with the values of E0, a, and k determined to be 1.6  eV, 38.7  nm, and 1.12× 107, respectively. Furthermore, using these constants, we can reproduce the experimental σ at 613  K for the same stress range (Fig.  4).

Fig.  4. The experimental σ curves for high pure aluminium at 598  K (black triangles) and 613  K (black squares), which can be well fitted (solid line) by Eq.  (17) with the same values of E0, a, and k.

4. Conclusion

In summary, a universal function was established to predict creep rate upon any stress and temperature and the validity was confirmed by available experimental data, especially by the fact that the function can reproduce all the previous creep rate equations obtained empirically for specific service conditions. In the application of the function to predict creep rate, we need not consider which creep mechanism dominates the process, but only perform several experiments to determine the three constants in the function. It is expected that the new function would be widely used in industry in the near future.

Reference
1 Babitsky V I and Wittenburg J 2007 Modeling of Creep for Structural Analysis Berlin Springer [Cited within:2]
2 Penny R K and Mariott D L 1995 Design for Creep London Chapman & Hall [Cited within:1]
3 Evans R W and Wilshire B 1985 Creep of Metals and Alloys London Institute of Materials [Cited within:1]
4 Wilshire B and Scharning P J 2007 Scripta Mater. 56 701 DOI:10.1016/j.scriptamat.2006.12.033 [Cited within:2]
5 Wilshire B and Scharning P J 2008 Int. Mater. Rev. 53 91 DOI:10.1179/174328008X254349 [Cited within:3]
6 Guo J T, Zhang S Z, Yuan C, Zhou W L and Li G S 2003 J. Aeronautical Mater. Suppl. 23 34 [Cited within:4]
7 Kral P, Dvorak J, Zherebtsov S, Salishchev G, Kvapilova M and Sklenicka V 2013 J. Mater. Sci. 48 4789 DOI:10.1007/s10853-013-7160-9 [Cited within:1]
8 Gu Y, Zeng F H, Qi Y L, Xia C Q and Xiong X 2013 Mater. Sci. Eng. A 575 74 DOI:10.1016/j.msea.2013.03.038 [Cited within:5]
9 Zhu S M, Easton M A, Gibson M A, Dargusch M S and Nie J F 2013 Mater. Sci. Eng. A 578 377 DOI:10.1016/j.msea.2013.04.100 [Cited within:1]
10 Wilshire B and Battenbough A J 2007 Mater. Sci. Eng. A 443 156 DOI:10.1016/j.msea.2006.08.094 [Cited within:6]
11 Abdallah Z, Whittaker M T and Bache M R 2013 Intermetallics 38 55 DOI:10.1016/j.intermet.2013.02.003 [Cited within:1]
12 Xie J, Tian S G and Zhou X M 2013 JMEPEG 22 2048 DOI:10.1007/s11665-013-0477-3 [Cited within:2]
13 Norton F H 1929 The Creep of Steel at High Temperature London McGraw-Hill [Cited within:1]
14 McVetty P G 1943 Trans. ASME 65 761 [Cited within:1]
15 Garofalo F 1965 Fundamentals of Creep and Creep Rupture in Metals New York MacMillan [Cited within:1]
16 Dorn J E 1955 J. Mech. Phys. Solids 3 85 DOI:10.1016/0022-5096(55)90054-5 [Cited within:1]
17 Grant N J and Mullendore A W 1965 Deformation and Fracture at Elevated Temperatures Boston MIT Press [Cited within:1]
18 Larson F R and Miller J 1952 Trans. ASME 74 765 [Cited within:1]
19 Kassner M E and Perez-Prado M T 2004 Fundamentals of Creep in Metals and Alloys Amsterdam Elsevier [Cited within:1]
20 Maruyama K, Armaki H G and Yoshimi K 2007 J. Pressure Vessels Piping 84 171 DOI:10.1016/j.ijpvp.2006.09.015 [Cited within:1]
21 Nie M, Zhang J, Huang F, Liu J W, Zhu X K, Chen Z L and Ouyang L Z 2014 J. Alloys Compd. 588 348 DOI:10.1016/j.jallcom.2013.11.080 [Cited within:1]
22 Brunnera M, Huttner R, Bolitz M, Volkl R, Mukherji D, Rosler J, Depka T, Somsen C, Eggeler G and Glatzel U 2010 Mater. Sci. Eng. A 528 650 DOI:10.1016/j.msea.2010.09.035 [Cited within:1]
23 Shrestha T, Basirat M, Charit I, Potirniche G P and Rink K K 2013 Mater. Sci. Eng. A 565 382 DOI:10.1016/j.msea.2012.12.031 [Cited within:1]
24 Terada Y, Murata Y, Sato T and Morinaga M 2011 Mater. Chem. Phys. 128 32 DOI:10.1016/j.matchemphys.2011.03.030 [Cited within:1]
25 Terada Y, Murata Y and Sato T 2013 Mater. Sci. Eng. A 584 63 DOI:10.1016/j.msea.2013.07.012 [Cited within:1]
26 Maruyama K and Yoshimi K 2007ASME 2007 Pressure Vessels and Piping Conference 9 631 [Cited within:1]
27 Tamura M, Abe F, Shiba K, Sakasegawa H and Tanigawa H 2013 Metall. Mater. Trans. A 442013 [Cited within:1]
28 Wilshire B, Scharning P J and Hurst R 2009 Mater. Sci. Eng. A 510–5113 DOI:10.1016/j.msea.2008.04.1252009 [Cited within:1]
29 Ennis P J, Shibli I A, Holdsworth S R and Merckling G 2005 Creep and Fracture in High Temperature Components—Design and Life Assessment Issues Lancaster DEStech Publications [Cited within:1]
30 Guo J T 2008 Materials Science and Engineering for Superalloys Beijing Science press [Cited within:7]
31 Spingarelli S, Ruano O A, Mehtedi M E and del Valle J A 2013 Mater. Sci. Eng. A 570 135 DOI:10.1016/j.msea.2013.01.060 [Cited within:2]
32 Lin Z Z, Yu W F, Wang Y and Ning X J 2011 Europhys. Lett. 94 40002 DOI:10.1209/0295-5075/94/40002 [Cited within:3]
33 Lin Z Z and Ning X J 2011 Europhys. Lett. 95 47012 DOI:10.1209/0295-5075/95/47012 [Cited within:1]
34 Lin Z Z, Zhuang J and Ning X J 2012 Europhys. Lett. 97 27006 DOI:10.1209/0295-5075/97/27006 [Cited within:1]
35 Ming C, Lin Z Z, Cao R G, Yu W F and Ning X J 2012 Carbon 50 2651 DOI:10.1016/j.carbon.2012.02.025 [Cited within:1]
36 Lin Z Z, Li W Y and Ning X J 2014 Chin. Phys. B 23 050501 DOI:10.1088/1674-1056/23/5/050501 [Cited within:1]
37 Yu W F, Lin Z Z and Ning X J 2013 Phys. Rev. E 87 062311 [Cited within:1]
38 Yu W F, Lin Z Z and Ning X J 2013 Chin. Phys. B 22 116802 DOI:10.1088/1674-1056/22/11/116802 [Cited within:3]
39 Zhang G Y, Guo J T and Zhang H 2006 The Chinese Journal of Nonferrous Metals 16 1883 [Cited within:1]
40 Gu Y F, Cui C, Ping D, Harada H, Fukuda T and Fujioka J 2009 Mater. Sci. Eng. A510–511 250 DOI:10.1016/j.msea.2008.04.1282009 [Cited within:3]
41 Vakili-Tahami F, Sajjadpour M and Attari P 2011 Strain 47 414 DOI:10.1111/str.2011.47.issue-5 [Cited within:3]
42 Bressers J 1981 Creep and Fatigue in High Temperature Alloys London Applied Science Publishers LTD [Cited within:1]
43 Seitz F 1950 Phys. Rev. 79 890 DOI:10.1103/PhysRev.79.890 [Cited within:1]
44 Dash W C 1958 J. Appl. Phys. 29 705 DOI:10.1063/1.1723255 [Cited within:1]
45 Guo J T, Ranucci D, Picco E and Strocchi M 1983 Metall. Trans. A 14 2329 DOI:10.1007/BF02663308 [Cited within:1]
46 Tobolová Z and čadek J 1979 Phil. Mag. 26 1419 DOI:10.1080/14786437208220352 [Cited within:1]