Professional Documents
Culture Documents
Download textbook Challenges In Mechanics Of Time Dependent Materials Volume 2 Proceedings Of The 2017 Annual Conference On Experimental And Applied Mechanics 1St Edition Alex Arzoumanidis ebook all chapter pdf
Download textbook Challenges In Mechanics Of Time Dependent Materials Volume 2 Proceedings Of The 2017 Annual Conference On Experimental And Applied Mechanics 1St Edition Alex Arzoumanidis ebook all chapter pdf
https://textbookfull.com/product/experimental-and-applied-
mechanics-volume-4-proceedings-of-the-2016-annual-conference-on-
experimental-and-applied-mechanics-1st-edition-yong-zhu/
https://textbookfull.com/product/dynamic-behavior-of-materials-
volume-1-proceedings-of-the-2015-annual-conference-on-
experimental-and-applied-mechanics-1st-edition-bo-song/
https://textbookfull.com/product/micro-and-nanomechanics-
volume-5-proceedings-of-the-2016-annual-conference-on-
experimental-and-applied-mechanics-1st-edition-la-vern-starman/
Conference Proceedings of the Society for Experimental Mechanics Series
Challenges in
Mechanics of Time
Dependent Materials,
Volume 2
Proceedings of the 2017 Annual Conference on
Experimental and Applied Mechanics
Conference Proceedings of the Society for Experimental Mechanics Series
Series Editor
Kristin B. Zimmerman Ph.D.
Society for Experimental Mechanics, Inc.
Bethel, CT, USA
Alireza Amirkhizi
UMass Lowell North Campus, Perry 332
University of Massachusetts
Lowell, MA, USA
Challenges in Mechanics of Time-Dependent Materials represents one of nine volumes of technical papers presented at the
SEM 2017 Annual Conference and Exposition on Experimental and Applied Mechanics organized by the Society for
Experimental Mechanics and held in Indianapolis, IN, in June 12–15, 2017. The complete proceedings also includes the
following volumes: Dynamic Behavior of Materials; Advancement of Optical Methods in Experimental Mechanics;
Mechanics of Biological Systems, Materials and other topics in Experimental and Applied Mechanics; Micro- and
Nanomechanics; Mechanics of Composite, Hybrid and Multifunctional Materials; Fracture, Fatigue, Failure and Damage
Evolution; Residual Stress, Thermomechanics & Infrared Imaging, Hybrid Techniques and Inverse Problems; and Mechan-
ics of Additive and Advanced Manufacturing.
Each collection presents early findings from experimental and computational investigations on an important area within
experimental mechanics, the mechanics of time-dependent materials being one of these areas.
This track was organized to address constitutive, time (or rate)-dependent constitutive, and fracture/failure behavior of a
broad range of materials systems, including prominent research in both experimental and applied mechanics. Papers
concentrating on both modeling and experimental aspects of time-dependent materials are included.
The track organizers thank the presenters, authors, and session chairs for their participation and contribution to this track.
The support and assistance from the SEM staff is also greatly appreciated.
v
Contents
vii
viii Contents
Abstract The delayed mechanical behaviour of the clay-rock/concrete interface under shear stress has been experimentally
investigated. After a first characterisation of the short-term response of the interface with direct shear tests, long duration
shear tests under constant normal load have been performed with a new experimental apparatus which was developed for this
study. The response of the interface in time is discussed.
1.1 Introduction
In the context of the underground nuclear waste disposal projects in deep argillaceous geological formations in several
countries, Underground Research Laboratories (URL’s) have been developed in order to demonstrate the underground
isolation. The French National Radioactive Waste Management Agency (ANDRA) is studying the behaviour of a deep
geological radioactive repository in Callovo Oxfordian clay-rock (known as COx) in Bure, France. Such projects require
prediction of irreversible deformations over a large time scale, in order to asses the durability of the confinement.
The sealing of the underground repository tunnel, is partly ensured by a mechanical barrier made of concrete, directly in
contact with the rock. Thus, the delayed mechanical behaviour of the clay-rock/concrete interface is of a large interest. The
amplitude of delayed strains depends on the material and test conditions. The state of stress determines the mode of failure of
the sample and the brittle-ductile transition [2]. Gonze [3] showed that for higher deviatoric stresses, the strain rate increases.
Zhang and Rothfuchs [8] showed that clay-rock creeps already under very low stress of 2 MPa, indicating that a lower creep
limit seems not to exist for this clay-rock. Concerning the delayed behaviour of concrete, many studies have shown that the
obtained creep deformation doesn’t increase proportionally with the increase of the applied loading levels [6, 7]. Pons and
Torrenti [5] showed that for applied charge lower than 50% of the compressive strength of concrete, the delayed strain rate
decreases with time.
Although the physical properties of most of the materials involved (rock, concrete and bentonite mixtures) are extendedly
studied, there have been few studies on the behaviour of the contact interfaces between these materials. The scientific goal of
this work involves the understanding of the creep mechanism of the interface of two materials and more specifically whether
the interface itself highlights its own creep mechanism, independently of the creep of both materials in contact. An
experimental campaign on the mechanical characterisation of the interface of the underground concrete gallery with the
host rock in time has been developed and will be presented.
E. Stavropoulou (*)
University of Grenoble Alpes, CNRS, Grenoble INP (Institute of Engineering University of Grenoble Alpes, Grenoble, France),
3SR, F-38000, Grenoble, France
Agence Nationale pour la gestion des Déchets Radioactifs (ANDRA), 92290 Châtenay-Malabry, France
e-mail: eleni.stavropoulou@3sr-grenoble.fr
M. Briffaut (*) • F. Dufour
University of Grenoble Alpes, CNRS, Grenoble INP (Institute of Engineering University of Grenoble Alpes, Grenoble, France),
3SR, F-38000, Grenoble, France
G. Camps
Agence Nationale pour la gestion des Déchets Radioactifs (ANDRA), 92290 Châtenay-Malabry, France
The investigation of the mechanical behaviour of the clay-rock/concrete interface in time, required a study of the short-term
response, performing shear tests with the BCR3D apparatus (a 3D shear box for rocks, [4]). This device allows shearing
under constant normal stress or constant normal volume with the application of symmetrical and opposite displacement on
the joint’s active part. Samples of COx (78mmØ) have been moulded in shear boxes with the studied concrete, as shown in
Fig. 1.1. Short-term shear tests under constant normal volume (with applied normal stress up to 12 MPa) have been
performed, in order to characterise the elastoplastic behaviour of the joints. The results have been analysed using the Mohr-
Coulomb failure criterion, allowing the evaluation of the resistance of the interface in shearing.
The characterisation of the mechanical behaviour of the interface in time, required the design of a new experimental
device. For the purpose of this work, SInC Box (Shear-Interfaces-Creep Box) has been developed, a sophisticated device
which allows the study of an interface in shearing under long duration load. SInC Box is based on the existing creep devices
in traction for intact samples [1] and introduces a second loading axis, orthogonal to the existing one. It is designed in a way,
so that the normal force is centered in the mean plane of the interface and the applied shear force passes the axis of the
interface (Fig. 1.1(1 and 2)). The application of a normal and a shear load, is followed by the measurement of the relative
shearing displacement (Fig. 1.1c), as well as dilatancy/contraction, allowing the demonstration of the failure mode of the
interface.
Several tests have taken place on samples of clay-rock/concrete interface, in temperature controlled environment (20 C) and
under different constant normal stress. The results of three tests are presented. The application of the normal load is followed
by the application of a shear load. Test 1 was performed under normal stress σn ¼ 6 MPa and shear stress τ ¼ 2.7 MPa which
we maintained constant throughout the entire test. In Test 2, under σn ¼ 8 MPa, an initial shear stress of τ ¼ 1.3 MPa was
followed by an increase to 2.5 MPa. Finally, Test 3 was performed under normals stress σn ¼ 10 MPa and shear stress
τ ¼ 3.8 MPa.
For the analysis of the results of a long-duration test, the instantaneous elastoplastic response of the interface is taken into
account. Figure 1.2a shows the ratio τ/ τmax for each test, where τmax is the calculated instantaneous shear resistance for the
Fig. 1.1 Left: Sample illustration; (a) 3D concept of the two half shear boxes, (b) Vertical cut illustration the interface of the two geomaterials, (c)
Illustration of the applied forces (Fn normal force, Ft shear force, to the mean plane of the interface) and the relative shear displacement (dut) of the
two half shear boxes. Right: SInC Box zoomed in, illustrating the position of the sample and the application of the two orthogonal loads; (1) normal
force transducer, (2) shear force transducer, (3) sample moulded in the two half-shear boxes (rotated by 90 )
1 Experimental Characterisation of the Mechanical Properties of the Clay-Rock/Concrete Interfaces and Their. . . 3
(*103 )
1.2 12
83%
1 10
compliance (µ m/m/MPa)
8
0.8
6
0.6 74%
83%
4
σ σ
0.4 σ 2 54% σ
σ 29% σ
0.2 0
0 100 200 300 400 500 600 700 800 0 100 200 300 400 500 600 700 800
time (h) time (h)
Fig. 1.2 (a) Evolution of the ratio of applied shear stress over shear strength, (b) compliance of the shear response
given normal stress. The evolution of the measured compliance in shearing is presented in Fig. 1.2b, where the compliance is
calculated as the delayed shear strain over the applied shear stress (εxy/ τ).
Observing the initial evolution of the compliance between the three tests (t ¼ 0 ! t ¼ 100 h), one can notice that the
measured compliance increases faster for higher shear strength levels. Although this observation is not obvious for Tests
1 and 3, given the similar levels of applied shear stress (83% and 74%), it is more clear for Test 2. Indeed for a low τ/
τmax ¼ 29%, the compliance is very low and gets quickly stabilised. An increase from 29% to 54% leads to an increase of
the rate of compliance, being though lower than the rates of Tests 1 and 3. In all three tests the measured product εxy/τ
increases non linearly in time, with a decreasing rate. This response indicates the existence of delayed deformations within
the primary phase.
1.4 Conclusions/Perspectives
The delayed behaviour in shearing of the COx/concrete interfaces has been experimentally investigated. The development of
a new apparatus (SInC Box) allowed measurement of the deformation of the interface in time, in both normal and tangential
direction. For the applied stress paths and the given duration of the tests, only primary creep was observed.
The study of creep in interfaces is complex, as it involves the evolution in time of two different materials. The obtained
delayed response can be the result of the delayed response of each material or of the interface itself independently. The
results of longer duration tests are going to be discussed.
References
1. Fabre, G., Pellet, F.: Creep and time-dependent damage in argillaceous rocks. Int. J. Rock Mech. Min. Sci. 43(6), 950–960 (2006)
2. Gatelier, N.: ETUDE EXPERIMENTALE ET THEORIQUE DE L’ENDOMMAGEMENT DES ROCHES ANISOTROPES. Diss. (2001)
3. Gonze, P.: Techniques de calcul utilisées en congélation des terrains. La thermomecanique des roches. 16, (1988)
4. Hans, J., Boulon, M.: A new device for investigating the hydro-mechanical properties of rock joints. Int. J. Numer. Anal. Methods Geomech. 27
(6), 513–548 (2003)
5. Pons, G., J. M. Torrenti. “Retrait et fluage.” La durabilité des bétons: bases scientifiques pour la formulation de bétons durables dans leur
environnement, Ollivier JP, Vichot A., Presses de l’École Nationale des Ponts et Chaussées. 167–216 (2008)
6. Roll, F.: Long-time creep-recovery of highly stressed concrete cylinders. Spec. Publ. 9, 95–114 (1964)
7. Sellier, A., Buffo-Lacarriere, L.: Towards a simple and unified modelling of basic creep, shrinkage and drying creep of concrete. Eur. J. Environ.
Civil Eng. 13(10), 1161–1182 (2009)
8. Zhang, C., Rothfuchs, T.: Experimental study of the hydro-mechanical behaviour of the Callovo-Oxfordian argillite. Appl. Clay Sci. 26(1),
325–336 (2004)
Chapter 2
Coupled Anisothermal Chemomechanical Degradation Solutions
in One Dimension
Abstract This paper focuses on thermal oxidation of Silicon Carbide (SiC) – a key process of degradation in aircraft turbine
components. In this work, passive oxidation is considered, which produces amorphous silica that is accumulated on top of
the SiC substrate. The mathematical problem is formulated within the context of mixture theory (Gardiner G (2017)
Aeroengine composites, Part 1: the CMC invasion. Composites World 31 July 2015: n. pag. Web. 06 Mar; Jacobson, J
Am Ceram Soc 76(1):3–28, 1993), which allows to model multi-constituent behavior – fluid and solid in this case – within
the same continuum domain, while retaining interaction terms between constituents. Preliminary isothermal results have
shown that the phenomena of interest are captured: expansion due to chemical reaction, change in solid density from
unreacted to fully-oxidized material, interactive force among constituents, and stress variation across reaction zone. The
method presented considers the anisothermal evolution of the problem.
For numerous applications, modern and forthcoming engineering materials must be designed to perform in harsh thermal
and/or chemical environments and numerical methods capable of simulating those conditions are required. In the case of
aeronautical applications, one such group of materials is ceramic matrix composites (CMCs). Due to their thermo-
mechanical properties [1] that allow them to perform stably at high temperatures, CMCs are making their way from into
more areas of the turbine, replacing metallic alloy materials, which require coolant flow to be kept at working performance
temperatures [2]. Some manufactures expect a tenfold increase on the use of CMCs in their engines over the next decade [2],
as improvements in manufacturing and analysis techniques permit production of components of more complex geometry.
However, CMCs are subjected to chemical degradation through oxidation at high temperatures [2–5]. This paper focuses on
thermal oxidation of silicon carbide (SiC) – a key process of degradation in aircraft turbine components. It is also a process
of interest in the electronics industry as it is one of the techniques involved in the manufacturing of semiconductors, where
thermal oxidation is performed deliberately to create an insulating oxide layer [6–8].
The mechanisms of thermal oxidation of SiC have been categorized as either active or passive [3–5]. In the former, the
products of reaction include volatile silicon oxide species that are lost to the surrounding environment and leave more of the
SiC exposed. In contrast, passive thermal oxidation produces amorphous silica that is accumulated on top of the SiC
substrate. Temperature and pressure determine which regime of thermal oxidation will occur [4]. In this work, the passive
oxidation is considered.
A schematic description of passive oxidation process of SiC is presented in Fig. 2.1: oxygen (O2) in-diffuses through the
solid, reaches substrate material (SiC), and reacts to form solid silicon dioxide (SiO2) and gaseous carbon monoxide (CO).
The chemical reaction that occurs between the O2 and the SiC is described by the following chemical equation:
3
SiCðsÞ þ O2ðgÞ ! SiO2ðsÞ þ COðgÞ ð2:1Þ
2
This process exhibits steep gradients in the properties of the solid (SiC/SiO2) and in oxygen concentration across the
domain. Moreover, substantial volume changes occur, as the amorphous silica produced by reaction occupies ~2.2 the
volume of SiC it replaces. Both of these situations pose numerical challenges in simulations, which are addressed by the
method presented in the sequel.
The mathematical problem is formulated within the context of mixture theory, which allows to model multi-constituent
behavior – fluid and solid in this case – within the same continuum domain, while retaining interaction terms between
constituents. One fundamental assumption of mixture theory is that the different constituents co-occupy the same continuum
domain. The balance equations governing the problem are formulated for each individual constituent and for the mixture,
and the constituent level equations contain source-like terms that describe the interaction between constituents. Addition-
ally, we adopt the constitutive relations derived through the imposition of the maximization of entropy production rate from
the work in [9].
We adopt the assumption made in [10] in which the constituent free energies, entropies, and internal energies can be written
in suitable condensed forms, which in turn imply that the constituent temperatures are equal to each other:
ψ f ¼ ψ s ¼ ψ, ηf ¼ ηs ¼ η, εf ¼ εs ¼ ε, θf ¼ θs ¼ θ ð2:2Þ
In addition, the rapport between the mechanical deformation and extent of reaction is strong, as reaction induces a
doubling in volume; thus, it is desirable to solve the system of equations in a fully coupled fashion, since a staggered
approach can potentially introduce inaccuracies in capturing this strongly coupled behavior. Moreover, algorithmic splits in
staggered approaches impede optimal rates of convergence. An evolving internal variable measures the extent of the reaction
throughout the domain; alternatively, this internal variable can also be understood as the measuring the amount of substrate
material (SiC) currently available for reaction.
Based on existing work [10], a consistently derived Variational Multi-Scale Stabilized (VMS) Finite Element Method is
developed to solve the coupled, nonlinear PDEs that arise from conservation laws and mixture theory constitutive relations.
The solution fields are fluid (O2) density and velocity; solid (SiC/SiO2) density and displacement; and mixture temperature.
The stabilization terms arise from the split of the fluid field into coarse and fine scales. Carbon monoxide (CO) is assumed to
diffuse out instantaneously and its mass is accounted for through stoichiometric relations. The FEM interpolation is done
within an Arbitrary Lagrangian-Eulerian (ALE) mesh, which proves suitable to model the diffusion of the fluid constituent
(O2) through the deforming solid (SiC/SiO2). From the point of view of the fluid, the mesh is an ALE frame, while from the
point of view of the solid, the mesh is Lagrangian, and nodal points are solid material points.
2 Coupled Anisothermal Chemomechanical Degradation Solutions in One Dimension 7
The linearized temperature form of the mixture energy balance is derived using concepts from [11] in the context of the
mixture theory presented in [9]. We begin by considering the balance of energy for the mixture (combination of all
constituents):
dε X2 X2
ρ ¼ div ðρα εα uα Þ þ trðTα Lα Þ divq þ ρτ
dt α¼1 α¼1 ð2:3Þ
1
mf vf vs vf vs I f vf vs
2
Through the assumptions made and neglecting the internal dissipation term, followed by appropriate manipulations,
including the linearization of the temperature field, we arrive at the following expression:
€ þ ρc0 θref μ ρ_ f þ ρ_ s cθ ϑ_
θref μcθ ϑ
ρ ð2:4Þ
divq þ ρτ þ ρmθF : F_ s þ ρmθs ρ_ s þ ρmθf ρ_ f ¼ 0
Where:
2 2 2 2
∂ ψ ∂ ψ ∂ ψ s ∂ ψ f
c0 ¼ θref , mθF ¼ θref , mθs ¼ θref ρ_ , mθf ¼ θref ρ_ ð2:5Þ
∂ϑ2 ∂Fs ∂ϑ ∂ρs ∂ϑ ∂ρf ∂ϑ
Based on the modeling assumptions and within the context of mixture theory, the one-dimensional conservation laws can be
written as follows:
dρα ∂v α
Balance of mass : þ ρα 1 mα ¼ 0 ð2:6Þ
dt ∂x
α
∂T 11 dv α
Balance of Linear Momentum : þ ρα b1 þ I 1α ρα 1 ¼ 0 ð2:7Þ
∂x dt
d2 ϑ μ dρf dρs dϑ
Balance of Energy : θref μcθ 2 þ ρc0 θref þ cθ
dt ρ dt dt dt ð2:8Þ
∂q dρ s
dρf
1 þ ρτ þ ρmθF F_ 11
s
þ ρmθs þ ρmθf ¼0
∂x dt dt
Accordingly, the constitutive relations considered are given by the following expressions:
∂ψ ∂ψ
s
T 11 ¼ ρF11 ρs ρ s ð2:9Þ
∂F11 ∂ρ
f
f ∂ψ L ∂v1
T 11 ¼ ρf ρ þ μA ð2:10Þ
∂ρf ∂x
8 M. Anguiano et al.
I 1f ¼ mf v1f v1s μAv v1f v1s ð2:11Þ
I 1s þ ms v1s ¼ I 1f þ mf v1s ð2:12Þ
∂ϑ
q1 ¼ θref μl11 ð2:13Þ
∂x
mf ¼ Γ_ 011 ¼ R0 ρf ϕ ð2:14Þ
c1s θ 1 f
s s s s 2
ψ ¼ A þ ðB þ c Þðθ θ Þ ðθ θ Þ c2 θ ln s þ
ss
Rθρ þ k2f ρf
2 θ ρT
2 ð2:15Þ
ρs 1 s 1
þ s λ þ μTs þ αs þ 2 μLs μTs þ βs E11 s
αTs θs θref þΛ
ρρT 2 2
Z
μ 0 2 s
Λ¼ Γ11 K 1 E11 αTs θs θref þ 2Γ_ 011 A0Γ dΓ011 ð2:16Þ
2
1 _0
2
1 4 Γ 11Γ011 ρK1 E11 s
μ¼ 2 2 ð2:17Þ
2 ∂v1f 2 2 1 0 0 2
AL ∂x
þ cθ ϑ2 þ Av v1f v1s ρA0Γ Γ_ 011 þ l11 ∂ϑ
∂x
2 Γ_ 11 Γ11 ρK1 E11
s
Preliminary isothermal results have shown that the phenomena of interest are captured: expansion due to chemical
reaction, change in solid density from unreacted to fully-oxidized material, interactive force among constituents, and stress
variation across reaction zone. It may be possible to explore other interesting behaviors as different material properties or
functional form dependencies are considered. For instance, the variation of solid density from unreacted to fully oxidized
material may or may not be monotone in its evolution, depending on whether the chemo-mechanical coupling is through the
cube (or some other power) of the extent of reaction. From a physical standpoint, this is due to two competing processes by
which solid concentration changes: (1) rate of mass incorporated to solid by reaction, and (2) rate of expansion due to
reaction. The nature of this density evolution may be material/process dependent, and should be confirmed by experiments.
Finally, the method presented constitutes a more general approach than the ones commonly used in the literature [6, 7]
first, in that it considers and solves for the transport of oxygen from the boundary to the reaction zone separately from the
incorporation of mass from the fluid to the solid at the reaction zone, and secondly, in that it considers the anisothermal
evolution of the problem. This will allow for the treatment of more complex geometries, a more varied range of material
properties, and a wider class of processes in future work. Consider for instance the same problem of thermal oxidation of
SiC, where the rate of reaction is anisotropic. In particular, when the SiC crystal is oriented such that C-face is exposed to
oxygen, it exhibits faster reaction rates than when the Si-face is [8]. Thus, for multi-dimensional geometries, where different
facets may be simultaneously exposed to oxygen, predicting the growth of the oxide scale requires accounting for this
anisotropy in the chemical reaction term, while the diffusion through the oxide remains an isotropic process.
References
1. Snead, L.L., Nozawa, T., Katoh, Y., Byun, T.S., Kondo, S., Petti, D.A.: Handbook of SiC properties for fuel performance modeling. J. Nucl.
Mater. 371(1–3), 329–377 (2007). https://doi.org/10.1016/j.jnucmat.2007.05.016
2. Gardiner, G.: Aeroengine Composites, Part 1: The CMC Invasion. Composites World 31 July 2015: n. pag. Web. 06 Mar. (2017)
3. Jacobson, N.S.: Corrosion of silicon-based ceramics in combustion environments. J. Am. Ceram. Soc. 76(1), 3–28 (1993). https://doi.org/
10.1111/j.1151-2916.1993.tb03684.x
4. Jacobson, N.S., Myers, D.L.: Active oxidation of SiC. Oxid. Met. 75(1–2), 1–25 (2011). https://doi.org/10.1007/s11085-010-9216-4
2 Coupled Anisothermal Chemomechanical Degradation Solutions in One Dimension 9
5. Smialek, J., Jacobson, N. S.: Oxidation of high-temperature aerospace materials. In: High Temperature Materials and Mechanisms,
pp. 95–162. CRC Press, Boca Raton. https://doi.org/10.1201/b16545-6 (2014)
6. Song, Y., Dhar, S., Feldman, L.C., Chung, G., Williams, J.R.: Modified deal grove model for the thermal oxidation of silicon carbide. J. Appl.
Phys. 95(9), 4953–4957 (2004). http://doi.org/10.1063/1.1690097
7. Hijikata, Y., Yaguchi, H., Yoshida, S.: A kinetic model of silicon carbide oxidation based on the interfacial silicon and carbon emission
phenomenon. Appl. Phys. Express. 2(2), 1–3 (2009). http://doi.org/10.1143/APEX.2.021203
8. Hijikata, Y., Yagi, S., Yaguchi, H., Yoshida, S.: Thermal oxidation mechanism of silicon carbide. In: Hijikata, Y. (ed.) Physics and Technology
of Silicon Carbide Devices, pp. 181–206. doi: 10.5772/50748. Available from: https://www.intechopen.com/books/physics-and-technology-
of-silicon-carbide-devices/thermal-oxidation-mechanism-of-silicon-carbide (2012)
9. Hall, R., Rajagopal, K.R.: Diffusion of a fluid through an anisotropically chemically reacting thermoelastic body within the context of mixture
theory. Math. Mech. Solids. 17(2), 131–164 (2012). https://doi.org/10.1177/1081286511407754
10. Hall, R., Gajendran, H., Masud, A.: Diffusion of chemically reacting fluids through nonlinear elastic solids: mixture model and stabilized
methods. Math. Mech. Solids. 20(2), 204–227 (2015). http://doi.org/10.1177/1081286514544852
11. Truesdell, C., Noll, W., Antman, S.S.: The non-Linear Field Theories of MechanicsBerlin. Springer, Heidelberg/Berlin (2004)
Chapter 3
Time-Temperature Analysis, DMA and Compression in PBXs
Darla Graff Thompson, Jamie Stull, Racci Deluca, and Geoff Brown
Abstract The mechanical response of plastic-bonded explosives (PBXs) is highly complex and depends on a number of
factors including, but not limited to temperature, strain rate and binders. In this work, we have measured and analyzed the
mechanical properties of LX-14, which contains 95.5 wt% HMX (octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine) and
4.5 wt% Estane 5702, with no additional plasticizer. We are interested to compare LX-14 with PBX 9501 (95 wt% HMX,
2.5 wt% Estane 5703 and 2.5 wt% nitroplasticizer), which has a relatively high content of plasticizer. The measurements
span a wide range of strain rates and temperatures. We have applied time-temperature analysis on the LX-14 mechanical
properties, i.e. dynamic mechanical analysis (DMA) measurements and quasi-static uniaxial compression to form satisfac-
tory master curves, and these data are compared with those from PBX 9501. A unique inhomogeneity is observed in the
LX-14 compression data as a function of the specimen location.
3.1 Introduction
The addition of polymeric binders to formulated high explosives was to primarily increase the safety of these materials in
applications. Plasticizers are small molecules that are also sometimes included in PBX formulations. They are dispersed in
the polymer and lower its glass transition temperature, causing the room-temperature properties of the material to be softer
and more pliable. The mechanical properties of plasticized composites tend to be lower in strength and in modulus, and
higher in ductility than their unplasticized counterparts.
PBX 9501 and LX-14 are HMX-based formulations that primarily differ by the presence/absence of plasticizer. PBX
9501 is 95 wt% HMX, 2.5 wt% Estane 5703 and 2.5 wt% nitroplasticizer. The nitroplasticizer is a eutectic mixture of
bis(2,2-dinitropropyl)Acetal/Formal (abbreviated here as BDNPA/F or also simply NP). LX-14- contains 95.5 wt% HMX
and 4.5 wt% Estane 5702, with no plasticizer.
Time-temperature superposition principles, developed for and widely applied in the thermal-mechanical analysis of
polymers [1–4], have also been successfully applied to PBX composites [5, 6] including PBX 9501. We are interested to
compare the master curve response for LX-14 that was obtained previously with PBX 9501, to better understand the role of
the plasticizer.
3.2 Experimental
A uniaxial die was used to compact LX-14 molding powder to a cylinder approximately 6 inches in diameter by 5 inches in
height. The target density was 1.830 g/cm3. Cores were removed down the axis of the part. From each core, three
compression test specimens were machined, 28.6 mm long by 9.5 mm diameter. Compression specimens were marked as
top, center or bottom, relating to which part of the pressing they were taken from. Additionally, rectangular bars were
machined with their long axis aligned with the axis of the pressing, 0.1 inch 0.5 inch 1.77 inch.
Immersion density measurements were performed in water on all specimens prior to testing. The density of a silicon
standard was also measured/verified. Specimens were desiccated 10 days before testing commenced.
1.826
1.824
1.822
1.820
1.818
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Specimen #
LX-14 compression tests were conducted remotely to failure on an Instron 5567 workstation with an MTS-upgraded
controller. A Bemco Environmental Chamber was used to control the temperature. The test conditions are listed in Table 3.1.
For all tests, thermal ramps were conducted at 1 C/min, and specimens soaked at target temperatures 30 min prior to testing.
All tests were crosshead rate controlled, and specimen strain was measured using two oppositely-mounted knife-edge
extensometers.
A TA Instruments Q800 DMA in combination with a duel cantilever clamp was used to measure the LX-14 specimens. At
each 3 C step in temperature between 100 and 100 C, the method applied a 10 μm displacement at five different
frequencies (0.3, 1, 3, 5 and 10 Hz). Rectangular specimens 0.1 inch 0.5 inch 1.77 inch, flat and parallel to within
0.001 inch, were machined from hydrostatically pressed billets. Temperature control was performed using liquid nitrogen
and an electric heating element.
3.3 Results
Immersion densities are plotted in Fig. 3.1 versus arbitrary specimen number. Data have been color-coded to indicate
specimen type/location. Likely the 1.830 g/cm3 target density of the large cylinder was not reached as all but one specimen
shows a density below this value. Furthermore, there is evidence of a significant density gradient, as all of the “bottom”
specimens are on the order of 0.006 g/cm3 lower than the “top” and “center.” Note that the six rectangular DMA specimens
span the density range of the compression specimens.
In Fig. 3.2 are plotted the stress versus average strain curves for all of the LX-14 compression tests. Colors and line styles
are used to indicate temperatures and strain rates, respectively. As is typical with PBX materials, over this range of
temperatures and strain rates, the effect of temperature on the material response is much greater than the effect of strain rate.
DMA measures the in-phase (storage) and out-of-phase (loss) elastic modulus as a function of the displacement of the
oscillating frequency and temperature. Tan delta is the ratio of the storage modulus to the loss modulus. The displacement
amplitude applied should be small enough to avoid exceeding the elastic limit of the material. In Fig. 3.3 we have plotted the
storage modulus, loss modulus and tan delta versus temperature. The peak near 25.04 C in the loss modulus curve arises
from the glass transition temperature (Tg) of the soft segment of the Estane binder. In the tan delta trace, it is slightly lower at
a temperature of 22.03 C. These values are near those previously reported for the Tg of Estane in LX-14 [4].
3 Time-Temperature Analysis, DMA and Compression in PBXs 13
Fig. 3.2 Stress versus strain average for all LX-14 compression tests; legend indicates temperature and strain rate
Strorage Molulus (MPa)
500
400
300 B
0.3
Tan Delta
0.2
0.1 C
0
−100 −50 0 50 100
Temperature (C)
For the purposes of analysis, stress-strain curves are often characterized using three parameters: maximum stress, σm, which
is the highest stress value achieved before failure of the material; strain at maximum stress, εm, which is the strain value
corresponding to σm; modulus, E25, which is the slope of the linear elastic portion of the stress-strain curve at 25% of σm. In
this paper we will use only the first two parameters. In Fig. 3.4a, we have plotted all values of σm and εm as a function of log
(strain rate).
14 D.G. Thompson et al.
50 50
(a) (b)
Maximum Stress (MPa)
30 30
20 50°C 20
23°C 50°C
10 10
10°C 23°C
-15°C 0 10°C
0 3.0
2.8 -15°C
2.8
Strain at Max Stress (%)
50
(c)
Maximum Stress (MPa)
40
30
20
50°C
10
23°C
0 10°C
2.4
-15°C
2.2
Strain at Max Stress (%)
2.0
1.8
1.6
1.4
1.2
1.0
-8 -6 -4 -2 0 2 4
log(Strain Rate)
Fig. 3.4 LX-14 Compression parameters max stress and strain at max stress plotted versus log(strain rate). (a) all parameters as measured; (b) all
parameters after time-temperature shift to 23 C equivalent test temperature using 10 C/decade rate; (c) shifted data, same as (b), but with all
“top” located specimens removed. See legend for test temperatures
The development and application of time-temperature superposition principles to polymers [1–4] and to PBXs [5, 6] are
covered elsewhere. For the sake of brevity, we present the simplified relationship used by Siviour [2], in which the change in
strain rate is proportionally related to differences in test temperature using a single constant, A:
ε_ ref 1
logaT ¼ log ¼ T ref T ð3:1Þ
ε_ A
We apply this equation to the parameters in Fig. 3.4a by selecting 23 C as our reference temperature, i.e. data collected at
all other temperatures will be shifted to a new position on the log(strain rate) axis according to time-temperature
equivalence, to the position where it would have been with testing at 23 C. By analyzing the data collected at different
3 Time-Temperature Analysis, DMA and Compression in PBXs 15
0.056 0.35
173 K
182 K
0.054 191 K
200 K 0.3
0.052 209 K
218 K
227 K 0.25
0.05 236 K
245 K
0.048 254 K 0.2
Tan Delta
Tan Delta
263 K
272 K
0.046
281 K
290 K 0.15
0.044 299 K
308 K
317 K 0.1
0.042
326 K
335 K
0.04 344 K 0.05
353 K
362 K
0.038
371 K
0
−0.5 0 0.5 1 1.5 −20 −10 0 10 20 30
(a) log(Frequency) (b) log(Frequency)
5500
5500
5000
5000
4500
4500
4000 4000
Storage Modulud (MPa)
Storage Modulud (MPa)
3500 3500
3000 3000
2500 2500
2000 2000
1500 1500
1000 1000
−0.5 0 0.5 1 1.5 −20 −10 0 10 20 30
Fig. 3.5 LX-14 DMA results (a) as-measured and (b) time-temperature shifted values of tan delta and (c) as-measured and (d) time-temperature
shifted values of storage modulus. Temperatures are listed in the legend
temperatures, and solving for log(_ε ref ), a “by-eye” best fit of the A parameter was performed. For the data plotted in Fig. 3.4b
and c, we used A ¼ 10 C/(decade rate). Maximum stress parameters, now shifted to an equivalent test temperature of
23 C, form the master curve shown at the top of Fig. 3.4b. The strain values at maximum stress values are also time-
temperature shifted. For many PBXs, the strain master curve shows a contiguous trend, but usually only a small change, as a
function of the shifted strain rate.
In Fig. 3.5a and c, the storage modulus and the tan delta (loss/storage) response have been plotted at different
temperatures as a function of frequency. In Fig. 3.5b and d, the same storage modulus and tan delta values are plotted,
but on a frequency-shifted axis. The data were shifted using Eq. 3.1 with A ¼ 5 C per decade frequency.
16 D.G. Thompson et al.
0.25
0.2
Tan Delta
0.15
0.1
0.05
0
1000 2000 3000 4000 5000 6000
Storage Modulus (MPa)
A Wicket plot of the data is plotted in Fig. 3.6, which is the tan delta versus the storage modulus. This plot is intended to
verify the time-temperature assumption of one relaxation time corresponding to a single transition. This is confirmed if the
data from a curve is in the shape of an arch or a croquette wicket [3]; however, Fig. 3.6 reveals that there is only a very small
temperature range between ~235 K and 253 K where the specimen adheres to the principles of the time-temperature
superposition.
3.5 Discussion
Close analysis of the raw uniaxial compression data in Fig. 3.2 shows a disturbing inconsistency. Most test conditions have
three test specimens, and very often one of the three specimens gives rise to a stress-strain curve that extends much further to
the right than the others, showing failure at slightly lower loads and much higher strains. As specimen locations were
randomized for testing, we predicted that these outliers would prove to be the “bottom” specimens, the low-density outliers
shown in Fig. 3.1. Contrary to this, however the anomalous curves were ALL shown to arise from the “top” specimens.
Based on the mapping of density gradients in large die-pressed cylinders by Olinger et al. [7], we believe that a large density
gradient may have formed at the top of the large as-pressed cylinder. When the small compression cylinders were removed,
those from the top layer had average densities that were nearly identical to those from the center layer, see Fig. 3.1. We
believe, however, that the density distribution of the top-layer specimens must have been asymetric, with a significantly
higher density region seated adjacent to a low density region. These data are novel observations, where mechanical property
differences are observed as a function of location, and yet cannot be correlated with density or underlying texture
differences. We believe that the absence of plasticizer in LX-14 reduces the flow of the binder during pressing, having an
effect on the homogeneity of the density in the final pressed part. In Fig. 3.4c, we have removed all compression parameters
from top-located specimens. Note, in particular, how this removes the high values of the strain at maximum stress and the
trend line is narrowed.
The master curve of the LX-14 compressive stress maximum shows apparent adherence to the principles of time-
temperature superposition which are the basis for Eq. 3.1. The value of A ¼ 10 C/(decade rate) for LX-14 compression is
very similar to the best-fit value of A ¼ 8 C/(decade rate) for PBX 9501 [6]. In Fig. 3.7, for comparison, we overlay the
compression master curves for both of these materials. The maximum stress master curve clearly shows the softening effect
of the plasticizer in PBX 9501. Interestingly, the addition of the plasticizer does not result in an increase of the strain at
maximum stress, an effect we might have predicted. As observed with PBX 9501 [6], the DMA master curves of the LX-14
storage modulus and tan delta, given in Fig. 3.5, show apparent adherence to time-temperature principles. However, the
Wicket plot analysis, Fig. 3.6, cautions that only a very narrow range of temperature shows a time-temperature agreement.
We plan to repeat these quasi-static compression and DMA measurements on LX-14 specimens with more uniform density.
3 Time-Temperature Analysis, DMA and Compression in PBXs 17
20
10
0
2.4
2.2
Strain at
1.6
1.4
1.2
1.0
0.8
0.6
-10 -8 -6 -4 -2 0 2 4
log(Strain Rate)
3.6 Conclusion
We have performed quasi-static compression and DMA testing on specimens obtained from a large die-pressed cylinder of
LX-14. We have performed time-temperature superposition analysis on these data with some success. Direct master curve
comparison has been made between LX-14 and PBX 9501, showing the effect of the plasticizer. However, the LX-14
compression data, coupled with the immersion density values, provide our first-ever observation of mechanical property
variation that correlates with specimen location in a pressing, but not caused directly by the specimen density.
References
1. Williams, M.L., Landel, R.F., Ferry, J.D.: The temperature dependence of relaxation mechanisms in amorphous polymers and other glass
forming liquids. J. Am. Chem. Soc. 77, 3701–3707 (1955)
2. Siviour, C.R., Walley, S.M., Proud, W.G., Field, J.E.: The high strain rate compressive behaviour of polycarbonate and polvinylidine difluoride.
Polymer. 46, 12546–12555 (1984)
3. Menard, K.P.: Dynamic Mechanical Analysis, a Practical Introduction, 2nd edn. CRC Press, Taylor and Francis Group, Boca Raton (2008)
4. Hoffman, D.M.: Dynamic mechanical signatures of a polyester-urethane and plastic bonded explosives based on this polymer. J. Appl. Polym.
Sci. 83, 1009–1024 (2002)
5. Williamson, D.M., Siviour, C.R., Proud, W.G., Palmer, S.J.P., Govier, R., Ellis, K., Blackwell, P., Leppard, C.: Temperature-time response of a
polymer bonded explosive in compression (EDC 37). J. Phys. D. Appl. Phys. 41, 85404–85414 (2008)
6. Thompson, D.G., DeLuca, R., Brown, G.W.: Time-temperature analysis of plastic-bonded explosives PBX 9501 and PBX 9502 in tension and
compression. J. Energ. Mater. 30(4), 299–323 (2012). LA-UR 11-07068
7. Olinger, B.W., Dobratz, B.M., Skidmore, C.B.: Density variations in cylinders and hemispherical shells of pressed PBX 9501 and PBX 9502.
23th aging, compatibility, and stockpile stewardship conference proceedings, Sandia National Laboratories, SAND2001-8038, CA, November
14–16, p. 133 (2000)
Chapter 4
Statistical Life Time Prediction Under Tension Loading
for Unidirectional CFRP with Thermoplastics as Matrices
Abstract We have proposed the life prediction method for the statistical creep failure time under the tension loading along
the longitudinal direction of unidirectional CFRP from the statistical static strengths of unidirectional CFRP measured at
various temperatures. First, a method of predicting the statistical creep failure time of CFRP is explained briefly based on
Christensen’s model of viscoelastic crack kinetics. Second, two types of unidirectional CFRP which consist of carbon fiber
T300–3000 with thermosetting resin and thermoplastics as matrices, respectively. Third, the static strengths of these
unidirectional CFRPs are experimentally and statistically measured at various temperatures. Then the creep failure times
of these unidirectional CFRPs are predicted statistically using the statistical static strengths at various temperatures. Finally,
the creep failure times of these unidirectional CFRPs are measured experimentally using these unidirectional CFRPs for
comparison with the predicted ones.
Keywords Carbon fiber reinforced thermoplastics • Creep failure time • Statistical life time • Prediction • Viscoelasticity
4.1 Introduction
Carbon fiber reinforced plastics (CFRP) have been used for the primary structures of airplanes, ships, automobiles and other
vehicles for which high reliability must be maintained during long-term operation. Therefore, an accelerated testing
methodology is strongly anticipated for the long-term life prediction of CFRP structures exposed to actual environmental
temperatures, water, and other influences.
The mechanical behavior of matrix resin of CFRP exhibits time and temperature dependence, called viscoelastic
behavior, not only above the glass transition temperature Tg, but also below Tg. Consequently, it can be presumed that the
mechanical behavior of CFRP depends strongly on time and temperature [1–5]. Our previous papers have proposed the
formulation of statistical static, creep, and fatigue strengths of CFRP based on the viscoelasticity of matrix resin [6, 7].
The tensile strength along the longitudinal direction of unidirectional CFRP constitutes important and basic data for the
reliable design of CFRP structures. The authors developed a test method for the creep and fatigue strengths as well as the
static strength at elevated temperatures by using the resin-impregnated carbon fiber strands (CFRP strands) combined with
T300–3000 and epoxy resin [8]. Our most recent study undertook the prediction of statistical creep failure time under tension
loading along the longitudinal direction of unidirectional CFRP performed using CFRP strands of T300–3000 and epoxy
resin. The statistical creep failure time of CFRP strands at a constant load and temperature was predicted using statistical
results of static tensile strengths of CFRP strands measured at various temperatures and the viscoelastic behavior of matrix
resin. The predicted results quantitatively agree well with the experimentally obtained results measured using creep tests for
CFRP strands [9].
As described herein, the proposed method of predicting the statistical creep failure time under the tension loading along
the longitudinal direction of unidirectional carbon fiber reinforced thermoplastics (CFRTP) from the statistical static
strengths of unidirectional CFRTP measured at various temperatures is clearly valid quantitatively. First, a method of
predicting the statistical creep failure time of CFRTP from the statistical static strengths of CFRTP measured at various
temperatures is explained briefly based on Christensen’s model of viscoelastic crack kinetics [10]. Second, many CFRTP
strands combined with T300–3000 and thermoplastic epoxy resin as the specimens for the longitudinal direction of
unidirectional CFRTP are prepared using simultaneous molding to elicit stable and uniform mechanical and thermal
properties. Third, the static strengths of unidirectional CFRTP are experimentally and statistically measured at various
temperatures using these CFRTP strands. Then the creep failure time of unidirectional CFRTP is predicted statistically using
the statistical static strengths at various temperatures based on the predicting method. Finally, the creep failure times of
unidirectional CFRTP at a constant load and a temperature are measured experimentally and probabilistically using these
CFRTP strands for comparison with the predicted ones.
We have proposed the formulation for the statistical static strength σ s of CFRTP based on the viscoelasticity of matrix resin,
as shown in the following equation in our previous paper [7] as
∗
1 D ðt; T Þ
logσ s ðPf ; t; T Þ ¼ logσ 0 ðt0 ; T 0 Þ þ log½ ln ð1 Pf Þ nR log , ð4:1Þ
αs Dc ðt0 ; T 0 Þ
where Pf signifies the failure probability, t denotes the failure time, t0 represents the reference time, T is the temperature, T0
stands for the reference temperature, σ 0 and αs respectively denote the scale parameter and the shape parameter on Weibull
distribution of static strength, nR is the viscoelastic parameter, and Dc and D* respectively represent the creep and
viscoelastic compliances of matrix resin. The viscoelastic compliance D* for the static load with a constant strain rate is
shown by the following equation.
The statistical static strength σ s is shown by the following equation by substituting Eq. (4.2) into Eq. (4.1).
1 Dc ðt=2; T Þ
logσ s ðPf ; t; T Þ ¼ logσ 0 ðt0 ; T 0 Þ þ log½ ln ð1 Pf Þ nR log ð4:3Þ
αs Dc ðt0 ; T 0 Þ
The creep strength is obtainable by horizontally shifting the static strength by the amount log A. Therefore, the statistical
creep strength σ c is shown by the following equation.
1 Dc ðAt=2; T Þ
logσ s ðPf ; t; T Þ ¼ logσ 0 ðt0 ; T 0 Þ þ log½ ln ð1 Pf Þ nR log ð4:4Þ
αs Dc ðt0 ; T 0 Þ
The failure probability of unidirectional CFRTP under a constant creep stress σ c0 can be shown by the following equation
from Eq. (4.4).
σ c0 Dc ðAt=2; T 0 Þ
Pf ¼ 1 expðFÞ, logF ¼ αs log þ αs nR log ð4:5Þ
σ0 Dc ðt0 ; T 0 Þ
The shifting amount log A determined by the slope kR of the logarismic static strength against the logarismic failure time
is shown by the following equation.
k R ¼ nR m R ð4:7Þ
The parameter mR is the slope of the logarismic creep compliance of matrix resin against the logarismic loading time.
4 Statistical Life Time Prediction Under Tension Loading for Unidirectional CFRP with Thermoplastics as Matrices 21
Table 4.1 Composition and cure schedule of CFRTP strand T300/thermoplastic epoxy resin
Carbon fiber strand Composition of resin (weight ratio) Cure schedule
T300–3000 Thermoplastic epoxy resin: (100) 100 C 0.5 h
Cure accelerator: (6.5)
A CFRTP strand which consists of high strength type carbon fiber T300–3000 (Toray Industries Inc.) and a thermoplastic
epoxy resin was molded by pultrusion method. The composition of thermoplastic epoxy resin and the cure condition of
CFRTP strand are presented in Table 4.1. The diameter and the gage length of CFRTP strands are approximately 1 mm and
200 mm, respectively. The glass transition temperatures Tg of the thermoplastic epoxy resin in CFRTP strand was 102 C
determined from the peak of loss tangent against temperature at 1 Hz using the DMA.
4.4 Creep Compliance of Matrix Resin and Static Strength of CFRTP Strands
The dimensionless creep compliance Dc/Dc0 for thermoplastic epoxy (TP-EP) measured at various temperatures is shown on
the left of Fig .4.1. The long-term Dc/Dc0 at T ¼ 85 C is obtained by shifting horizontally those at various temperatures, as
shown in the right of Fig. 4.1. In the right of Fig. 4.1, the long-term Dc/Dc0 for thermosetting epoxy (TS-EP) at T ¼ 120 C is
shown by gray symbols [9]. The time dependence of Dc/Dc0 is quite different with each other.
The static tension tests for CFRTP strand were conducted at three temperature levels, 25, 80, and 90 C with cross-head
speed 2 mm/min. The tensile strength of the CFRTP strand σ s is obtained using the following equation.
Pmax
σs ¼ ρ ð4:8Þ
te
Therein, Pmax is the maximum load [N]. ρ and te are the density of the carbon fiber [kg/m3] and the tex of the carbon fiber
strand [g/1000 m].
Figure 4.2 shows the Weibull distributions of the static strength of CFRTP strand (T300/TP-EP). αs is the shape parameter
and βs is the scale parameter of CFRTP strand in this figure. Although the scale parameter decreases according to the
temperature raise, the shape parameter maintains almost a constant value for CFRTP strands. σ 0 and αs in Eqs. (4.3), (4.4)
and (4.5) were determined as shown on Table 4.2. In this figure, the Weibull distributions of the static strength of CFRP
strand (T300/TS-EP) are also shown by gray symbols [9]. The temperature dependence and scatter for static strength of
T300/TS-EP are smaller than those for T300/TP-EP.
22 M. Nakada et al.
Table 4.2 Parameters for statistical creep failure time prediction for CFRTP strand
T300/TP-EP T300/TS-EP
Scale parameter of static strength of CFRP strand at 23 C: σ0 [MPa] 3482 3727
Shape parameter of static strength of CFRP strand: αs 22 38
Viscoelastic parameter of matrix resin: nR 0.30 0.05
Slope of viscoelastic compliance of matrix resin: mR 0.67 0.28
Slope of static strength of CFRP strand against failure time: kR 0.201 0.014
Logarithmic time shifting factor: logA 0.77 1.86
Fig. 4.3 Statistical static strength against viscoelastic compliance of matrix resin. (a) T300/TP-EP. (b) T300/TS-EP [9]
Figure 4.3 presents the dimensionless static strength σ s/σ 0 against the dimensionless viscoelastic compliance of matrix
resin D*/Dc0 for CFRTP strand (T300/TP-ES) and CFRP strand (T300/TS-ES) [9]. The relation of σ s/σ 0 against D*/Dc0 can
be shown by the straight line with the slope of nR which is the viscoelastic parameter in Eqs. (4.3), (4.4) and (4.5). The slope
of nR for T300/TP-EP is larger than that for T300/TS-EP. nR is shown on Table 4.2.
4 Statistical Life Time Prediction Under Tension Loading for Unidirectional CFRP with Thermoplastics as Matrices 23
Fig. 4.4 Failure probability against creep failure time. (a) T300/TP-EP. (b) T300/TS-EP [9]
Creep failure tests of CFRTP strands (T300/TP-EP) were conducted using the specially designed creep failure testing
machine [9]. The applied creep stress σc0 was 2263 MPa (65% of scale parameter of static strength at 23 C). Test
temperature was 85 C. Number of specimens was 15. Results of the creep failure tests are presented in Fig. 4.4a. The
predicted creep failure probability against failure time calculated by substituting the parameters on Table 4.2 in Eqs. (4.5),
(4.6), and (4.7) is also shown in Fig. 4.4a. The predicted statistical creep failure time agrees with the experimental data. The
creep failure test results and predicted creep failure probability for CFRP strands (T300/TS-EP) are shown in Fig. 4.4b
[9]. Although the same carbon fibers are used for both strand specimens, the creep failure time region and time dependence
for failure probability for T300/TP-EP are quite different with those for T300/TS-EP. Our proposed method can predict the
creep failure bahevior for both strand specimens.
4.6 Conclusion
The prediction method for statistical creep failure time under tension loading along the longitudinal direction of unidirec-
tional CFRP using the statistical static tensile strength of unidirectional CFRP and the viscoelasticity of matrix resin based
on Christensen’s model for viscoelastic crack kinetics were applied to the case of unidirectional CFRTP with a thermoplastic
epoxy resin as the matrix. We employed two kinds of resin impregnated carbon fiber strand with same carbon fibers and
different matrices. One is the T300/TS-EP with thermosetting epoxy, the other is the T300/TP-EP with thermoplastic epoxy.
Although the same carbon fibers are used for both specimens, the creep failure time region and time dependence of failure
probability are quite different with each other. Our proposed method can predict the creep failure bahevior for both
materials.
Acknowledgments This research was promoted by COI program “Construction of next-generation infrastructure using innovative materials
~Realization of a safe and secure society that can coexist with the Earth for centuries ~ supported by MEXT and JST.
References
1. Aboudi, J., Cederbaum, G.: Analysis of viscoelastic laminated composite plates. Compos. Struct. 12, 243–256 (1989)
2. Sullivan, J.: Creep and physical aging of composites. Compos. Sci. Technol. 39, 207–232 (1990)
3. Gates, T.: Experimental characterization of nonlinear, rate dependent behavior in advanced polymer matrix composites. Exp. Mech. 32, 68–73
(1992)
4. Miyano, Y., Nakada, M., McMurray, M.K., Muki, R.: Prediction of flexural fatigue strength of CFRP composites under arbitrary frequency,
stress ratio and temperature. J. Compos. Mater. 31, 619–638 (1997)
5. Kawai, M., Yagihashi, Y., Hoshi, H., Iwahori, Y.: Anisomorphic constant fatigue life diagrams for quasi-isotropic woven fabric carbon/epoxy
laminates under different hygro-thermal environments. Adv. Compos. Mater. 22, 79–98 (2013)
24 M. Nakada et al.
6. Miyano, Y., Nakada, M., Cai, H.: Formulation of long-term creep and fatigue strengths of polymer composites based on accelerated testing
methodology. J. Compos. Mater. 42, 1897–1919 (2008)
7. Nakada, M., Miyano, Y.: Advanced accelerated testing methodology for long-term life prediction of CFRP laminates. J. Compos. Mater. 49,
163–175 (2015)
8. Miyano, Y., Nakada, M., Kudoh, H., Muki, R.: Prediction of tensile fatigue life under temperature environment for unidirectional CFRP. Adv.
Compos. Mater. 8, 235–246 (1999)
9. Nakada, M., Miyano, Y.: Statistical creep failure time of unidirectional CFRP. Exp. Mech. 56, 653–658 (2016)
10. Christensen, R., Miyano, Y.: Stress intensity controlled kinetic crack growth and stress history dependent life prediction with statistical
variability. Int. J. Fracture. 137, 77–87 (2006)
Chapter 5
Time and Temperature Dependence on Tensile Strength
of Unidirectional CFRP with Various Carbon Fibers
Abstract The method of predicting the statistical creep failure time under the tension loading along the longitudinal
direction of unidirectional CFRP from the statistical static strengths of this CFRP measured at various temperatures is
applied to various unidirectional CFRP with different kinds of carbon fibers. Two kinds of CFRP strands with two kinds of
PAN based carbon fibers as specimens for unidirectional CFRP are prepared. The static strengths of these CFRP strands are
experimentally and statistically measured at various temperatures. Then the creep failure times of these CFRP strands are
predicted statistically using the statistical static strengths based on the predicting method. Finally, the creep failure times of
these CFRP strands at a constant load and a temperature are measured experimentally and probabilistically using these
CFRP strands for comparison with the predicted ones. Additionally, the predicting method is also applied to two kinds of
CFRP strands with two kinds of pitch based carbon fibers.
Keywords Carbon fiber reinforced plastics • Creep failure time • Statistical life time • Prediction • Viscoelasticity
5.1 Introduction
Carbon fiber reinforced plastics (CFRP) have been used for the primary structures of airplanes, ships, automobiles and other
vehicles for which high reliability must be maintained during long-term operation. Therefore, an accelerated testing
methodology is strongly anticipated for the long-term life prediction of CFRP structures exposed to actual environmental
temperatures, water, and other influences.
The mechanical behavior of matrix resin of CFRP exhibits time and temperature dependence, called viscoelastic
behavior, not only above the glass transition temperature Tg, but also below Tg. Consequently, it can be presumed that the
mechanical behavior of CFRP depends strongly on time and temperature [1–5]. Our earlier reports have proposed the
formulation of statistical static, creep, and fatigue strengths of CFRP based on the viscoelasticity of matrix resin [6–7].
The tensile strength along the longitudinal direction of unidirectional CFRP constitutes important data for the reliable
design of CFRP structures. The authors developed a test method for creep and fatigue strengths as well as the static strength
at elevated temperatures for resin-impregnated carbon fiber strands (CFRP strands) combined with T300–3000 and epoxy
resin [8]. Furthermore, the authors have confirmed that the time and temperature dependent tensile static strength is
controlled by the viscoelastic behavior of matrix resin based on the Rosen’s shear lag model [9]. Additionally, the authors
developed a test method for the CFRP strand of T800–12000 and epoxy resin with highly reliable co-cured tab. The
temperature-dependent tensile strength of this CFRP strand was evaluated successfully [10].
Our most recent study undertook the prediction of statistical creep failure time under tension loading along the
longitudinal direction of unidirectional CFRP performed using CFRP strands of T300–3000 and epoxy resin [11]. The
statistical creep failure time of CFRP strands at a constant load and temperature was predicted using statistical results of
static tensile strengths of CFRP strands measured at various temperatures and the viscoelastic behavior of matrix resin. The
predicted results statistically agreed well with the experimentally obtained results measured using creep tests for CFRP
strands.
In this study, the proposed method of predicting the statistical creep failure time under the tension loading along the
longitudinal direction of unidirectional CFRP from the statistical static strengths of unidirectional CFRP measured at various
temperatures is applied to various unidirectional CFRP with different kinds of carbon fibers. First, the method of predicting
the statistical creep failure time of CFRP from the statistical static strengths of CFRP measured at various temperatures is