Engineering Fracture Mechanics 177 (2017) 153–166

Contents lists available at ScienceDirect

Engineering Fracture Mechanics

journal homepage: www.elsevier .com/locate /engfracmech

Influence of hygrothermal effects in the fracture processin wood under creep loading

http://dx.doi.org/10.1016/j.engfracmech.2017.04.0090013-7944/� 2017 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail address: nicolas.angellier@unilim.fr (N. Angellier).

Nicolas Angellier a,⇑, Frédéric Dubois a, Rostand Moutou Pitti b,c, Malick Diakhaté d,Raoul Spero Adjovi Loko e

aUniversité de Limoges, Heterogeneous Material Research Group, Civil Engineering Center, F-19300 Egletons, FrancebUniversité Clermont Auvergne, Université Blaise Pascal, Institut Pascal, BP 20206, F-63000 Clermont-Ferrand, FrancecCNRS, UMR 6602, Institut Pascal, F-63171 Aubiere, FrancedUniversité de Bretagne Occidentale, FRE CNRS 3744, IRDL, F-29600 Morlaix, FranceeUniversité Bordeaux 1, I2M/GCE, UMR5295, Talence Cedex, France

a r t i c l e i n f o

Article history:Received 25 November 2016Received in revised form 29 March 2017Accepted 7 April 2017Available online 9 April 2017

Keywords:WoodFractureCreep testEnergy release rateVariable environments

a b s t r a c t

The knowledge of crack driving forces such as energy release rate is very important in theassessment of the reliability of timber structures. This work deals with both static andcreep fracture tests in opening mode crack growth under hygro-thermal and mechanicalloadings. The experimental tests combining creep and hygro-thermal loadings are per-formed in a climatic chamber. The Double Cantilever Beam specimen with variable inertiamachined in Douglas Fir and White Fir species is used to investigate the effects of theseloadings on fracture processes. Two experimental protocols are presented. First, instanta-neous tests are carried out in order to identify the moisture content effects on fractureproperties. R-curves are studied using a finite element approach. Secondly, creep testsare performed by imposing high speed humidity variations in a climatic chamber.During these tests, the evolutions of the crack length are recorded.

� 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Aided by environment aspects and climate changes, timber structures know today a good industrial development in thecivil engineering community. Moreover, due to the advantages provided by its mechanical behavior (particularly underextreme loading conditions such as fire and seismic events), as well as its aesthetic and environmental effects, timber isemployed in building and civil engineering structures [20,21]. However, the development of timber bridges or buildings[8] meets today some brakes induced by the problematic of durability [13]. For example, timber bridges are often placedin extreme climates due to the presence of river. During the in-service life of the structure, and according to the externalloadings (such as permanent loads, service charges, snow, wing or traffic) associated with climatic conditions, severalpathologies can appear such as the acceleration of creep, the development of hydric stresses induced by the shrinkage-swelling process around the join influence zones [22,16]. In addition, cracks can also appear in the surface, due to gradientsof moisture content and the material orthotropy. The aim is to know if these cracks can propagate through the cross section.This problematic also exists in buildings in which heat systems induce dry and hot atmospheres. For example, infra-redsystems can generate high temperatures exceeding 45 null. The blocking of deformations around steel joints causes a

154 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

concentration of tension stresses that can cause cracks [6]. Coupled with external loadings such as permanent and climaticloads, designers must find solution in order to limit, to stop the crack propagations, and to reinforce the joint stiffness [23].To sum up, timber elements can exhibit micro-cracks, which eventually propagate under fatigue [11], overload or creep load-ing, which could cause failure of the structure. In addition, timber is a hygroscopic material whose mechanical behavior isvery sensitive to climatic changes such as temperature and moisture variations. For example, drying process accelerate thecrack growth, while wetting process induce the delay of the crack propagation [10]. This fact is accelerated with the time-dependent behavior of wood during long-term loading and creep tests [2,3]. The coupling between viscoelastic behavior andmoisture content changes is characterized by a coupling that is called today mechano-sorption, and which seems to induce amore complex interaction between mechanical loading and moisture content variations [19].

A durable timber structure design needs the understanding of the fracture risks. Today, the scientific locks can be clas-sified in three families. (i) The understanding of the crack growth process combining, at the same time, a complex mechan-ical behavior that includes both shrinkage and swelling effects [9], mechano-sorptive [12,7] behavior and viscoelasticproperties. (ii) The fracture analysis that includes the mechanical fields around the crack tip as well as their evolutions dur-ing climatic changes, energetic balance in the crack tip vicinity including the separation energy in terms of viscoelastic andmechano-sorptive dissipations. (iii) The experimental demonstration of the crack growth process caused by humidity vari-ations under creep loadings. This paper deals with this third point. Because it was observed that crack initiations areobserved during climate reheating or drying, this work focuses on desorption phases. The paper’s target is the highlightof moisture content variations under creep loadings configurations on the crack growth process by taking into accountmechanical behavior and fracture characteristic evolutions versus moisture content. Thus, the paper is divided into threesections.

The present work is devoted to the fracture tests performed in a climatic chamber. It deals with the crack growth processin timber elements subjected to both mechanical loadings and climatic variations. Through an experimental approach, itfocuses on the crack growth advance induced by high speed climatic variations applied on a hygroscopic and viscoelasticmaterial characterizing wood. The first part of the paper presents the experimental set-ups with samples preparation fortwo species used in Europe (Douglas Fir [1] and White Fir). The wood sample geometry is a Double Cantilever Beam(DCB) [5] in order to provide the stability of energy release rate during the crack growth process in opening mode. The frac-ture properties through characterization static tests are studied by considering two distinct moisture levels, a low one and ahigh one, respectively: the results are presented in terms of stiffness, R curve and critical energy release rate value. In the lastsection, we present and discuss results in terms of crack propagation driven by humidity cycles through creep tests.

2. Double cantilever beam geometry

2.1. Linear fracture mechanic

In the past, Chazal and Dubois [4] and Dubois et al. [5] studied the effects of viscoelasticity on the crack growth process. Inorder to separate loadings effect and long term evolutions induced by creep properties, a double cantilever beam had beenproposed. Because the crack stability was the main design criterion, the specific geometry includes a variable inertia. Thisstability is driven by the instantaneous evolution of the energy release rate versus the crack length. Let us consider a crackedbody in which Wext , We are the external loading work and the strain energy density, respectively. By neglecting the dynamiceffects during the crack tip advance @a, the energy balance can be written through the following energy release ratedefinition:

G ¼ @

@aðWext �WeÞ ð1Þ

During the crack process, according to an instantaneous fracture test, let us neglect the displacement variations versus thecrack tip advance. In this case, the expression (1) can be simplified by:

G ¼ � @We

@að2Þ

The crack growth initiation is driven by the critical value of the energy release rate called Gc according to the following frac-ture criterion:

G < Gc : crack stationary; G ¼ Gc : crack growth initiation ð3Þ

We can add to the last criterion a crack stability concept based on the rate evolution G versus the crack length a:

@G@a

< 0 and G < Gc : crack stability;@G@a

P 0 and G P Gc : crack instability ð4Þ

The criteria stated in expressions (3) and (4) will be used for the determination of the specimen geometry in an open modeconfiguration.

N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166 155

2.2. Crack growth stability

The future experimental protocols are based on a Double Cantilever Beam (DCB) with variable inertia beam specimen thatallows the crack growth to be stable. Usually, creep tests in fracture mechanic are based on the use of twin samples in whichit is defined fracture properties such as the critical force inducing the crack propagation. Nevertheless, wood material is char-acterized as a heterogeneous material. Moreover, the next experimental protocols need to separate the effects of viscoelasticproperties and moisture content variations on the crack growth process. In this case, the specimen’s design requests to have,at the same time, the known of the critical force inducing the crack growth initiation and the crack growth stability. An over-view of the experimental necessities is shown in Fig. 1. The figure shows us three specific zones.

– The first zone, called instable crack allows an initial loading at the critical force corresponding to a crack growth initiationat the initial crack length ao. According to inequalities (4), the crack can propagate until an arrest process at the cracklength a1. The limit value exceeding is resulting to an additional kinetic energy induced by the crack move.

– Between the crack lengths a1 and a2, the crack tip is in the stable zone in which the external loading cannot induce a prop-agation. In this case, as stated in the literature [5], we can count on viscoelastic effects as crack growth driver. However, inthis present case, we prefer to highlight the climatic variation’s effects on the crack propagation.

– The last zone concerns the final instable crack. With a crack length greater than a2, the crack propagates without controluntil reaching the total collapse of the sample.

Crack length a

Ener

gy re

leas

e rat

e G

cG

oa 1a 2a ta

Fig. 1. Crack stability.

Crown C

1x

2x

10

θ =

00

θ ⎛ ⎞= ⎜ ⎟

⎝ ⎠

⎛ ⎞⎜ ⎟⎝ ⎠

Fig. 2. Integration crown.

156 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

2.3. Modified DCB geometry

In this paper, we’ll use the double cantilever beam geometry modified by Chazal and Dubois [4] in which the variableinertia allows a compatibility of its energy release rate behavior with the one presented in Fig. 1. The DCB geometry isdesigned by calculating the energy release rate evolution for one unitary force using an elastic finite element approach.The calculation of G is performed by using the G-theta method implemented in the finite element software Castem devel-oped by the French atomic energy commission. For a plane configuration, its form is given by:

G ¼ Gh ¼ Crownð�We � hk;k þ rij � ui;k � hk;jÞdC ð5Þ

As shown in Fig. 2, the integration domain is a crown surrounding the crack tip. Its boundaries are virtually defined by a

derivable vector field ~h.

For a pre-crack oriented in the direction x1 the boundary conditions for the ~h field are given in this figure. Because theG-theta method is implemented in a finite element method, this field is defined by using a thermal gradient analogy. Thefinite element mesh is shown in Fig. 3 and an example of integration crown is shown in Fig. 4

According to a horizontal symmetry only the half-DCB specimen is modeled. About boundary conditions, a radial lockingaround mechanical fixations allows taking into account a radial contact without friction. Because the mechanical loading isan imposed displacement, we impose a unitary displacement to the symmetric axe. The calculus of the reaction forces on thisline allows the definition of the equivalent loading force. Because the energy release rate depends on the sample thicknessand elastic properties the G-curve is plotted according to a dimensionless form in which GðaoÞ is assimilated to the criticalvalue corresponding to the crack growth initiation:

~GðaÞ ¼ GðaÞGðaoÞ ð6Þ

According to the geometry definition specified in Fig. 5, the ~G evolution takes the form shown in Fig. 6.

Radial locking

Crack tip Imposed unitary displacement of the crack line

Fig. 3. Finite element mesh and boundary conditions.

2θ

Fig. 4. Integration crown around the crack tip.

R

L

R

T

20mm

3mm

130mm40mm

80mm

30mm

20mm

10mm

oa = 30mm

R=18mm

Fig. 5. Modified DCB geometry.

0.00

2.00

4.00

6.00

8.00

10.00

10 30 50 70 90 110 130 150Crack length a

G

1,00

75mm45mm

∼

Fig. 6. Evolution of ~G versus the crack length.

N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166 157

According to notations posted in Fig. 1, the different crack lengths can be remained below with a crack stability zone of30 mm:

ao ¼ 30 mm; a1 ¼ 45 mm; a2 ¼ 75 mm and at ¼ 150 mm ð7Þ

2.4. Sample’s conditioning

The chosen species are Douglas Fir and White Fir. Several samples of each species have been machined in a Radial-Longitudinal (RL) configuration. Because we would like to put in evidence the effects of humidity changes on the crackgrowth process, samples are conditioned in climatic chambers in two specific climates, Fig. 7. The first one (low moisturelevel climate) is characterized by a temperature of 25 �C and a relative humidity of 40%RH. The second one (high moisturelevel climate) is chosen with a temperature of 25 �C and a relative humidity of 90%RH. These two climates correspond, interms of average moisture content in specimens, to 9% and 18%, respectively. After conditioning, a notch of 50 mm (corre-

Fig. 7. Specimen conditioning in climatic chamber.

158 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

sponding to a pre-crack with an initial length equal to 30 mm), is performed along the grain direction with a band saw andholes with a diameter equal to 10 mm are drilled through the top and bottom cantilevers. In order to limit the time diffusionbut also to limit the risk of instability in torsion mode (Mode III), the sample thickness is equal to 20 mm.

3. Instantaneous behavior

3.1. Experimental test

The understanding of the fracture mechanisms under climatic variations requests the characterization of fracture prop-erties in low and high moisture levels conditions. This first experimental campaign is carried out to characterize both theglobal and the local fracture behavior of samples conditioned at 9% and 18% of moisture content referred to as low and highmoisture level specimens, respectively. We’ll focus on energetic properties through the critical energy release rate, the com-pliance specimen evolution during test and the characterization of the R-curve. A Zwick electromechanical testing machine,with a 50kN load capacity, is controlled in displacement, which allows forcing stable crack growth during the experimentaltest. This load is applied to the specimen by use of shafts pushed into the drilled holes, Fig. 8. Specimens are tested at a con-stant displacement rate of 0.5 mm/min. The Zwick data acquisition system allows the record of displacement and force evo-lutions. In Fig. 8, ko represents the initial experimental stiffness, Fini represents the initial loading that initiate cracking andFmax represents the peak force.

The force-displacement curves are plotted in Figs. 9 and 10 for Douglas Fir and White Fir, respectively: for each specie,two samples of each moisture level (low and high) were tested We can show the effects of species and moisture content

Fig. 8. Experimental setup and definition of force-displacement curve parameters.

0

50

100

150

200

250

300

350

0 1 2 3 4 5 6 7

Forc

e (N

)

Displacement (mm)

: low moisture level samples

: high moisture level samples

Fig. 9. Force-displacement curves for Douglas Fir samples: low and high moisture levels.

: low moisture level samples

: high moisture level samples

Fig. 10. Force-displacement curves for White Fir samples: low and high moisture levels.

N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166 159

conditioning. These figures also show low moisture dependence on the stiffness values. In regards to the crack initiation, theapplied force is higher for high moisture level specimens. For Douglas Fir, low moisture level specimens are characterized bya most important value for the force max, while we can observe the opposite for White Fir: most important value for theforce max for high moisture level specimen.

3.2. Fracture properties: Energetic approach and R-curves

Nevertheless, these curves (Figs. 9 and 10) allow the characterization of the samples in terms of:

– average energy release rate value Gmean calculated along the total crack path;

– stiffness evolution ~k during the fracture test according to the following expression:

~k ¼ Fd

ð8Þ

~k is characterized by its initial value ko and its evolution induced by local damage such as the fracture process zone devel-opment and the crack growth process. Fig. 11(a) shows an example of stiffness evolution during the fracture test.

This curve is used for the determination of the equivalent crack length corresponding to the accumulation of damage con-centrated in the process zone, the crack tip advance and the crack bridging. This equivalent crack length is calculated by per-forming a finite element analysis in the elastic domain based on the mesh visualized in Fig. 3. First, the elastic properties arefitted in order to find the initial experimental stiffness ko. In a next step the equivalent crack length is calculated in the target

to find ~k versus displacement according to Fig. 11(a). In a last step, for each displacement step, the energy release rate is cal-culated using the G-theta method defined by the expression (5).

This calculus allows drawing the R-curve (Fig. 11(b)) usually defined in the quasi brittle fracture approaches [14,15].In this graph, we can note four specific values to complete the sample’s average energy release rate value Gmean:

– the initial critical value of the energy release rate Gcini correspond to the first damage appearance,

– the plateau value GRc corresponding to the maximum value of the energy release rate,– the critical crack length ac corresponding to GRc . Its value is correlated with the size of the fracture process zone aroundthe crack tip and the quasi-brittle character (ductility),

– the unstable crack length au corresponding to the last value before propagation accelerates until collapse.

Figs. 12and 13 present R-curves for Douglas Fir and White Fir samples, respectively. For Douglas Fir, we can note that Gcini

is clearly lower for low moisture level samples while this moisture dependence is lower for the White Fir samples. Let’s notethat for both cases, the plateau is short, due to the size of samples creating a stop for process zone, and more pronounced forhigh moisture level samples. GRc is higher for high moisture level state for White Fir samples and for low moisture level state

0

20

40

60

80

100

120

140

160

0 0,5 1 1,5 2 2,5 3 3,5 4 4,5

Stiff

ness

(N/m

m)

Displacement (mm)

ok

Sample collapse

a)

Fig. 11. (a) Evolution of ~k versus displacement – (b) R-curve and characteristic values of Gh.

0

100

200

300

400

500

600

700

800

20 25 30 35 40 45 50 55 60 65 70

Ener

gy re

leas

e ra

te (J

/m²)

Equivalent crack length (mm)

: low moisture level samples

: high moisture level samples

Fig. 12. R-curves for Douglas Fir samples: low and high moisture levels.

160 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

Table 1Fracture properties.

ko (N/mm) Fini (N) Fmax (N) Gcini (J/m

2) GRc (J/m2) ac (mm) au (mm) Gmean (J/m2)

Low moisture level Douglas Fir 152–175 178–198 311–339 286–307 784–732 40.6–48.6 53–57 333–380Low moisture level White Fir 146–165 188–183 253–281 335–279 550–570 39–41.8 48–51 213–233High moisture level Douglas Fir 149–149 230–235 300–305 493–489 659–624 46.5–50 66–66 433–591High moisture level White Fir 159–163 190–191 297–300 308–307 589–626 47.2–51.3 60–66 346–367

0

100

200

300

400

500

600

700

800

20 25 30 35 40 45 50 55 60 65 70

Ener

gy re

leas

e ra

te (J

/m²)

Equivalent crack length (mm)

: low moisture level samples

: high moisture level samples

Fig. 13. R-curves for White Fir samples: low and high moisture levels.

N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166 161

for Douglas Fir samples. At last, we can observe that the maximal value for au under 70 mm doesn’t exceed the numerical endof stable zone a2 of 75 mm. Once again, this value is lower for lowmoisture level samples, reflecting a more brittle rupture ora reduced process zone.

Finally, all fracture properties are regrouped in Table 1 and will be references for the forthcoming creep tests in variableenvironment.

4. Creep behavior

4.1. Creep tests in variable humidity environment

The experimental device is a creep bench that consists of four poles [5] and allows testing four DCB specimens for eachspecie together with an initial load (determined by the previous static tests) applied through lever arm enabling to initiatecracking without exceeding the stability zone, Fig. 14. In this case, the force can be adjusted by adding small aggregates onthe loading supports.

The creep benches are into a 23 m3 climatic chamber that provides hydric cycles between 40% and 90% RH at a constanttemperature of 25 �C (±1 �C), Fig. 15, in accordance with studied states during instantaneous tests. In fact, two series of testswere performed: the first from a lowmoisture level state (noted A) and the second from a high moisture level state (noted B)for a total of 14 cracked samples (numbered 1–4 for each specie and each conditioning), for which humidity cycles are rel-atively similar (except for singular points) both in terms of amplitude and duration.

4.2. Crack tip evolution vs. Average moisture content

For each test, two reference specimens are weighed regularly for calculating their moisture content variations. Fig. 16shows evolution of moisture content versus time: we can note that moisture content varies effectively between approxi-mately 9% and 18% for both species. Equilibrium is reached in approximately 200 h. We can consider these values as repre-sentatives in regard to the small thickness of samples allowing the hypothesis of homogenous variation of moisture content.

Fig. 15. Temperature and relative humidity monitoring (example of the first test).

Fig. 14. Creep bench.

162 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

The crack tip position is also regularly measured on both sides of the tested specimens, Fig. 17. As we note a little dis-symmetry between front and rear faces, we keep an average value for the crack tip position. In this example of a DouglasFir sample, we see the evolution of the average crack tip position until the sudden crack occurs after 1200 h. We remark alsothat the ranges of crack stability during the test are the same at around 70 h before each intermediary crack evolutions. Thetotal crack length until the collapse of specimen is around 60 mm.

By superposing moisture content evolution and crack tip position evolution, we can observe, as expected from the resultsof instantaneous tests, that crack propagation principally occurs at the beginning of each drying phase, Fig. 18: the sample isweakened when Gc

ini and au decrease and the propagation stops thanks to the stability obtained by the geometry of specimen.The main part of the propagation occurs during the fast-drying phase of the sample from 18 to 11% in less than 24 h. We canquestion about this propagation: does it occur homogenously in the thickness or is it conducted by a moisture gradientbetween surface and depth?

Fig. 17. Example of result: crack growth monitoring.

Fig. 16. Moisture content monitoring for a complete test.

N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166 163

For the first test with a low moisture level initial state, Fig. 19, 2 of the 4 Douglas Fir specimens and 1 of the 3 White Firspecimens remained blocked and didn’t reach the complete rupture after 4 hydric cycles (that to say 52 days). We can notethat in this configuration the drilled holes in the top and bottom cantilevers may cause friction with the swelling of samples.

For the second test with a high moisture level initial state, Fig. 20, 2 of the 3 Douglas Fir specimens and 1 of the 3 WhiteFir specimens reached the complete rupture at first drying and one more White Fir specimen broke at the second drying, theothers remaining blocked. In this configuration, the variation of diameter of drilled holes during drying doesn’t cause sup-plementary friction.

Hence, Table 2 summarizes the applied load, the average crack growth advance, the rupture crack growth position, therupture time and the number of hydric cycles to rupture deducted from Figs. 19 and 20. In average, the crack advance stepis between 5 and 25 mm at each phase. The complete rupture of the test specimens occurs from 1 to 3 or more cycles: whenthe sample doesn’t collapse at the first drying, we observe that the crack tip position is between 70 mm (White Fir) and100 mm (Douglas Fir), corresponding to a crack length of 40 mm and 70 mm respectively, meaning the stability zone hasbeen exceeded, in accordance with the expected values from simulations and instantaneous tests.

Fig. 19. Comparison of crack growth evolutions (White Fir and Douglas Fir) with a low moisture level initial state.

Fig. 18. Example of result: superposition of average moisture content and crack growth evolutions.

Fig. 20. Comparison of crack growth evolutions (White Fir and Douglas Fir) with a high moisture level initial state.

164 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

Table 2Samples crack advance’s parameters.

Sample Appliedload (N)

Average crackadvance step (mm)

Rupture cracktip position (mm)

Rupturetime (h)

Number of hydriccycles to rupture

Douglas Fir A1 215 9 >68 – >4Douglas Fir A2 215 5 >55 – >4Douglas Fir A3 215 25 97 843.9 3Douglas Fir A4 215 13 86 1210.9 4White Fir A1 190 18 76 846 3White Fir A2 190 17 67 857.6 3White Fir A3 190 6 – – ?Douglas Fir B1 210 15 >50 23.6 1Douglas Fir B2 210 29 >70 – ?Douglas Fir B3 210 24 >60 9 1White Fir B1 210 21 >60 – ?White Fir B2 210 30 >60 23.6 1White Fir B3 210 8 >40 23.6 1White Fir B4 210 20 82 368.6 2

N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166 165

5. Conclusion and outlook

The static and creep fracture tests in opening mode crack growth under hygro-thermal and mechanical loadings havebeen investigated in this work. The Double Cantilever Beam specimen with variable inertia machined in Douglas Fir andWhite Fir species has been used in order to know the effect of these loadings on fracture processes. The analysis of theforce-displacement curves has shown that in low moisture level phase resistance of Douglas Fir is more important thanthe Withe Fir; so for this last specie, the fracture toughness in high moisture level phase is important comparatively tothe low moisture level phase. Also, the energy release rate during the low moisture level phase is important in the caseof Douglas Fir and the crack occurs around 100 mm and 70 mm for the White Fir.

Forthcoming outlooks on the experimental protocol to answer the issues raised are to: study samples with several thick-nesses (and so the shift to 3D simulations); use a modified geometry with a longer tail (to limit effects of edges on the pro-cess zone); avoid friction (caused by the swelling of samples during wetting); perform the crack propagation monitoringwith a non-destructive tool (as acoustic emission). These experimental results constitute a first database in order to validatenumerical tools given by a finite element model based on a new analytical formulation of the A-integral [18,17]. This inde-pendent integral allows considering the effect of thermo-hygro-mechanical loads in the cracking process. Coming crackgrowth tests will be conducted with the miniaturized Mixed Mode Crack Growth (MMCG) specimen with non-destructivemeasurements such as optical methods coupling with acoustic emission. In fact, these results will allow a better understand-ing of in-service behavior of timber structures subjected to mechanical loads under climatic variations. The final aim is toexpand the use of wood material (characterized by low environmental impact), and in particular the studied Massif Central’s(France) species in civil engineering structures.

Acknowledgements

The authors wish to strongly acknowledge the National Agency of Research (ANR) for its financial support of this workthrough the project CLIMBOIS N� ANR-13-JS09-0003-01 labeled by ViaMeca.

References

[1] Barret JD. Effect of crack-front width on fracture toughness of Douglas-fir. Eng Fract Mech 1976;8(4):711–7.[2] Becker P, Rautenstrauch K. Time-dependent material behavior applied to timber columns under combined loading. Part I: Creep deformation. Holz als

Roh-und Werkstoff 2001;59(5):380–6.[3] Becker P, Rautenstrauch K. Time-dependent material behavior applied to timber columns under combined loading. Part II: Creep-buckling. Holz als

Roh-und Werkstoff 2001;59(6):491–5.[4] Chazal C, Dubois F. A new incremental formulation in the time domain of crack initiation in an orthotropic linearly viscoelastic solid. Mech Time-

Depend Mater 2001;5:229–53.[5] Dubois F, Chazal C, Petit C. Viscoelastic crack growth process in wood timbers: an approach by finite element method for mode I fracture. Int J Fract

2002;113:367–88.[6] Dubois F, Moutou Pitti R, Pop O, Fournely E. Numerical model about fracture analysis and mechano-sorptive behavior in wood. Mechanics of Time-

dependent Materials (MTDM), Montreal, Canada, 27–30 May, 2014.[7] Dubois F, Husson JM, Sauvat N, Manfoumbi N. Modeling of the viscoelastic mechano-sorptive behavior in wood. Mech Time-Dependent Mater 2012;16

(4):439–60.[8] Franke B, Quenneville P. Analysis of the fracture behavior of Radiata Pine timber and Laminated Veneer Lumber. Eng Fract Mech 2014;116(4):1–12.[9] Hunt DG, Shelton CF. Longitudinal moisture-shrinkage coefficients of softwood at the mechano-sorptive creep limit. Wood Sci Technol 1988;22

(3):199–210.[10] Kang W, Ho Lee N. Mathematical modeling to predict drying deformation and stress due to the differential shrinkage within a tree disk. Wood Sci

Technol 2002;36(6):463–76.

166 N. Angellier et al. / Engineering Fracture Mechanics 177 (2017) 153–166

[11] Singh Konjengbam Darunkumar, Xu Y, Sinclair I. Strip yield modeling of fatigue crack under variable amplitude loading. J Mech Sci Technol 2011;25(12):3025–36.

[12] Leicester RH. A rheological model for mechano-sorptive deflections of beams. Wood Sci Technol 1971;5(3):211–20.[13] Mohareb A, Sirmah P, Pétrisons M, Gérardin P. Effect of heat treatment intensity on wood chemical composition and decay durability of Pinus patula.

Eur J Wood Wood Prod 2012;70(4):519–24.[14] Morel S, Bouchaud E, Schmittbuhl J, Valentin G. R-curve behavior and roughness development of fracture surfaces. Int J Fract 2002;114:307–25.[15] Morel S. R-curve and size effect in quasi-brittle fractures: case of notched structures. Int J Solids Struct 2007;44(13):4272–90.[16] Moutou Pitti R, Dubois F, Sauvat N, Fournely E. Strain analysis in dried green wood: experimentation and modeling approaches. Eng Fract Mech

2013;105:182–99.[17] Moutou Pitti R, Hamdi SE, Dubois F, Riahi H, Angellier N. Numerical fracture analysis under temperature variation by energetic method, vol.

8. Springer; 2016.[18] Moutou Pitti R, Dubois F, Petit C. Generalization of T and A integrals to time dependent materials: analytical formulations. Int J Fract 2010;161

(2):187–98.[19] Nguyen SL, Omar Saifouni O, Destrebecq JF, Moutou Pitti R. An incremental model for wood behaviour including hydro-lock effect. Final cost action

FP0904 conference, recent advances in the field of TH and THM wood treatment, May 19–21, Skellefteå, Sweden; 2014.[20] Riahi H, Moutou Pitti R, Chateauneuf A, Dubois F. Stochastic analysis of mixed mode fracture in timber material using polynomial chaos expansion.

Final cost action FP0904 conference, recent advances in the field of TH and THM wood treatment, May 19–21, Skellefteå, Sweden; 2014.[21] Riahi H, Moutou Pitti R, Dubois F, Chateauneuf A. Mixed-mode fracture analysis combining mechanical, thermal and hydrological effects in an isotropic

and orthotropic material by means of invariant integrals. Theoret Appl Fract Mech 2016. doi: http://dx.doi.org/10.1016/j.tafmec.2016.06.002.[22] Saft S, Kaliske M. A hybrid interface-element for the simulation of moisture-induced cracks in wood. Eng Fract Mech 2013;102:32–50.[23] Tlustochowicz G, Serrano E, Steiger R. State-of-the-art review on timber connections with glued-in steel rods. Mater Struct 2011;44(5):997–1020.