Professional Documents
Culture Documents
art ic l e i nf o a b s t r a c t
Article history: Experimental and numerical finite element results on the mechanical behavior of timber beech finger-
Accepted 19 March 2014 joint are presented. Numerical simulations are based on the Cohesive Zone Model (CZM) of Abaqus
Available online 4 April 2014 software to allow for accurate description of the progressive damage of the bond-lines within the finger-
Keywords: joint up to failure. To increase the finger-joint resistance, the geometry of the finger-joint has been
Finger-joint optimized using the Response Surface Method (RSM) and the Kriging interpolation. The finger length,
Glued solid timber the pitch and the tip gap have been defined as the design variables to optimize. The objective function
Adhesion has been defined in terms of the maximum bending force, obtained from four-point bending tests.
Adhesive Feasibility constraints on the design variables, conforming to the EN 14080 are also taken into account.
Cohesive behavior
The obtained results demonstrated clearly the potential in increasing the finger-joint resistance by
FEM
optimizing its geometry.
Optimization
& 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijadhadh.2014.03.007
0143-7496/& 2014 Elsevier Ltd. All rights reserved.
V.-D. Tran et al. / International Journal of Adhesion & Adhesives 52 (2014) 40–47 41
F/2 F/2
2. Numerical simulation of the finger-joint Fig. 3. Schematic representations of (a) pure tension and (b) pure shear.
The schematic of the initial finger-jointing configuration stu- The behavior of adhesives can be described by a Cohesive Zone
died is shown in Fig. 2; all dimensions are in millimeter (mm). Two Model (CZM) in which the progressive failure is expressed by a bi-
finger-joint configurations are commonly used in practice [34]: the linear traction–separation law. The traction–separation law is
first one is visible on the face, while the second one is visible on generally defined by three cohesive parameters, namely the
the side. The second type has been chosen for the present study. critical cohesive strength, the initial stiffness and the fracture
The specimens consist of finger-jointed solid lamellas made of toughness [35]. The fracture toughness can be determined by
beech wood with average density of 650 kg/m3 determined experiments, but it is difficult to obtain, experimentally, the critical
immediately prior to testing from the mass of the entire speci- cohesive strength and the initial stiffness and consequently they
mens divided by their volume and moisture content which are generally determined iteratively by adjusting the simulation
fluctuated from 9% to 11%. The adhesive used to bond the speci- results to the experimental ones [35–37].
mens was melamine–urea–formaldehyde (MUF). A total of ten In this study, the strength characteristics of the MUF adhesive,
specimens have been tested experimentally, under four-point in pure tension and pure shear, have been assessed by tests
bending tests according to the EN 408 requirements [1], in an according to the EN 302-3 [38] and the EN 302-1 [39] procedures,
Instron machine at speed rate of 3 mm/min. The deflection was respectively, for assessment of glue-line strength characteristics
measured with displacement transducers (LVDT) placed at the (Fig. 3). A total of ten specimens have been tested for each mode
mid-span of the specimen and recorded versus the total applied loading. The mean experimental values of the normal and tangen-
load. All tests have been conducted under ambient temperature of tial strengths were: sn ¼ 3.1 70.28 MPa and st ¼10.3 70.68 MPa,
20 1C and 65% relative humidity. respectively.
42 V.-D. Tran et al. / International Journal of Adhesion & Adhesives 52 (2014) 40–47
where sn and st are the normal and the shear stresses, respec-
tively. δn and δt are the corresponding separations in the normal
and tangential directions, respectively.
A damage initiation criterion of maximum stress which obeys
to the following quadratic criterion has been selected:
2 2
sn st
þ ¼ 1: ð2Þ
scn sct
A linear displacement-based softening has been selected to control
the damage evolution. When the criterion of maximum stress is
reached, the cohesive stiffness is degraded, by a scalar damage
variable D, which is related to the effective separation, to simulate
the progressive adhesive failure. Note that the use of a damage
model based on a single damage parameter assumes that the
failure of the adhesive is monotonic [42]. Thus the stress compo-
nents are affected by the damage variable according to the
Fig. 5. Comparison between mean experimental and numerical load–displacement
curves for (a) mode I and (b) mode II. Table 1
Optimal cohesive parameters used for mode I and mode II.
separation law, the cohesive parameters are: the critical cohesive Mode I 4.5 1.6 0.005
strengths ðscn ; sct Þ, the initial stiffness ðK n ; K t Þ and the maximum
K t (N/mm2/mm) sct (N/mm2) δmax (mm)
displacements ðδmax n ; δmax
t Þ. The cohesive parameters in pure mode Mode II 30 9.7
t
0.00005
I have been identified based on a modified DCB test (Fig. 4), similar
to that presented in [40], while the cohesive parameters in pure
mode II have been determined based on the pure shear test shown
in Fig. 3b.
In the identification procedure, the cohesive parameters have
been defined as design variables and identified, in each pure
mode, using an appropriate parametric study, via a user Python
script file. The cohesive parameters were selected on the basis of
the numerical curve that fit well the mean experimental one
(Fig. 5). For further details on the identification procedure, the
references [35,40] are recommended. Finally, the optimal cohesive
parameters obtained in each mode are summarized in Table 1. Fig. 6. Finite element model of the finger-joint specimen.
V.-D. Tran et al. / International Journal of Adhesion & Adhesives 52 (2014) 40–47 43
Fig. 7. Comparison between experimentally and numerically predicted load– Among the existing optimization methods, the Response Sur-
deflection curves. face Method (RSM) is widely and successfully used mathematical
Fig. 8. Comparison between failure modes of the finger-joint. (a) Numerical and (b) experimental.
44 V.-D. Tran et al. / International Journal of Adhesion & Adhesives 52 (2014) 40–47
tool for optimization problems in many fields of structural 3.1. Formulation of the optimization problem
mechanics. It is based generally on statistical considerations in
order to build local approximations of the real unknown function To increase the resistance of finger-joints, the optimization
to be optimized. It uses the Design of Experiments (DoE) method problem is expressed as follows:
to build approximated response surfaces. The DoE method is
max JðxÞ
used to evaluate effectively the response of a system which is
influenced by the variation of the design variables. To properly subjected to xu rx rxv ð5Þ
formulate an optimization problem, it is necessary to understand
u v
the characteristics of the system and to define precisely the design where JðxÞ is the objective function, x and x are the lower and the
variables affecting the response of the system. upper bound vectors of the design variable x, respectively.
The optimization problem consists in determining the optimal
geometry to achieve a higher resistance of finger-joints, which
corresponds to the maximum value of the cost (or objective)
function J. Therefore, the finger length, the pitch and tip gap,
denoted A, B, and C, respectively were defined as the design
variables which affect the failure behavior of the finger-joint. The
cost function J to maximize is related to the maximum bending
force Fmax which can be expressed under its normalized form as
follows:
F max
J¼ ð6Þ
F0
After defining the design variables, the levels and the number
of experiments were determined by using the central composite
Design of Experiments (DoE).
The objective function (J) is implicit compared to the design
variables and its evaluation requires large computation times. To
find the optimal design variables with low cost and with a good
accuracy, the Response Surface Method (RSM) is adopted using the
Kriging interpolation and coupled with a strategy of actualization
Fig. 9. The global flowchart of the automatic optimization procedure [47]. of the search space [47]. Central composite Design of Experiments
dk
k
xopt
2
xopt
d1
1
xopt 1 d2
xopt
x0
Fig. 10. Progression of the optimization procedure for 2D case: (a) global and (b) successive local approximations [47].
V.-D. Tran et al. / International Journal of Adhesion & Adhesives 52 (2014) 40–47 45
(DoE) with five levels for function evaluation has been adopted to A Python Script is used to create the ABAQUS finite element
build the response surfaces. model, to run the computations and to post-process the results,
which are exported in an ASCII file. That file is then read by a
3.2. The Kriging interpolation FORTRAN program, in which the response surface approximations
and SQP algorithm are implemented.
The Kriging [47,48] interpolation method is a powerful inter-
polation technique which is widely used in optimization problems.
It enables to represent complicated functions. This method is 4. Optimization results
applied in this work to represent the response surface in an
explicit form, according to design variables. Only a brief descrip- Optimization results are reported in Table 3, where it can be
tion of the Kriging model is given here, for further details seen that the optimal geometry of the finger-joint leads to an
reference is made to [47–49]. The approximate relationship of increase of the numerical bending force from 10,405 N to 13,803 N.
the objective and constraint function can be expressed as follows: With the optimal geometry the finger length has increased to
its upper level and the pitch of the finger joint has significantly
J n ðxÞ ¼ P T ðxÞβ þ r T ðxÞα ð8Þ
decreased, whereas the tip gap has slightly decreased (Fig. 11),
T
where β ¼ ½β1 ; :::::::::; βm is the coefficient vector and the random leading to an increase of the cohesive area (bonded surface).
fluctuation. r T ðxÞα is an interpolation of the residuals of the Fig. 12 illustrates the comparison between the numerically
regression model, while P T ðxÞβ is used to represent the deviation predicted load–deflection curves with both the initial and optimal
of the global model. finger-joints. In addition to the numerical bending force, which
increased from 10,405 N to 13,803 N, it can be also seen from this [13] Smith MJ, Dai H, Ramani K. Wood thermoplastic adhesive interface – method
figure that the deflection at failure has been increased from about of characterization and results. Int J Adhes Adhes 2002;22:197–204.
[14] Cheng E, Sun X. Effects of wood surface roughness, adhesive viscosity and
20 mm to 29 mm. processing pressure on adhesion strength of protein adhesive. J Adhes Sci
Technol 2006;20:997–1017.
[15] Konnerth J, Gindl W, Müller U. Elastic properties of adhesive polymers. Part I:
polymer films by means of electronic speckle pattern interferometry. J Appl
5. Conclusion Polym Sci 2014 2007; 103(6): 3936–9.
[16] Liu CT, Lii WJ. Effect of physicochemical properties and gluing methods on the
The aim of this paper was to optimize the finger-jointing qualities laminated wood made from fast-growing species. 9. Studies on the
fabrication of end-to-end grain joint laminated wood from Taiwan acacia,
geometry towards the improvement of its mechanical strength. sweet gum and Taiwania. For Prod Ind 1989;8:75–84.
The crack growth across the bond-lines of the finger-joint has been [17] Karastergiou S, Barboutis J, Vassiliou V. Effect of the PVA gluing on bending
successfully modeled using the cohesive surfaces of ABAQUS and strength properties of finger jointed turkey oakwood (Quercus cerris L.). Holz
Roh Werkst 2006;64:339–40.
suitable damage laws. This computational approach was found to [18] Frangi A, Fontana M, Mischler A. Shear behaviour of bond lines in glued
be suitable for cracked bond-lines where the crack propagates laminated timber beams at high temperature. Wood Sci Technol
through a known direction. In this study, finger-joints have been 2004;38:119–26.
[19] Gonzalez G, Moya R, Monge F, Cordoba R, Coto JC. Evaluating the strength
identified as the weakest points of the entire jointed specimens. of finger jointed lumbers of Gmelina Arborea in Costa Rica 2004;28:319–-
Thus, an improvement of the finger-joint resistance is proposed 23New For 2004;28:319–23.
based on the optimization of its geometry. The Response Surface [20] Frangi A, Bertocchi M, Clauß S, Niemz P. Mechanical behaviour of finger joints
at elevated temperatures. Wood Sci Technol 2012;46:793–812.
Method (RSM) has been used for the optimization purpose.
[21] Clauß S, Joscak M, Niemz P. Thermal stability of glued wood joints measured
Response surfaces were built using the Kriging interpolation. The by shear tests. Eur J Wood Prod 2011;69(1):101–11.
objective function has been defined in terms of the maximum [22] Schmidt M, Thö nnißen A, Knorz M, Windeisen E, Wegener G. Relevant wood
bending force, obtained from four-point bending tests. characteristics for gluing beech and ash with regard to discoloration. Eur J
Wood Wood Prod 2012;70:319–25.
Feasibility constraints on the geometrical parameters are also [23] Aicher A, Reinhardt HW. Delaminierungseigenschaften und scherfestigkeiten
taken into account. In order to improve the RSM accuracy and to von verklebten rotkernigen buchenholz-lamellen. Holz Roh Werkst
deal with the local optimum problem, an efficient optimization 2007;65:125–36.
[24] Aicher S, Radovic B. Untersuchungen zum Einflußder Keilzinkengeometrie auf
strategy allowing the actualization of the search space has been die Zugfestigkeit keilgezinkter Brettschichtholz-Lamellen. Holz Roh Werkst
adopted. 1999;57:1–11.
The obtained numerical results demonstrated clearly the [25] Smardzewski J. Distribution of stresses in finger joints. Wood Sci Technol
1996;30:477–89.
potential of increasing the finger-joint resistance by optimizing [26] Serrano E, Gustafsson PJ, Larsen HJ. Modelling of finger-joint failure in glued-
its geometry. laminated timber beams. J Struct Eng ASCE 2001;127(8):914–21.
[27] Wernersson H. Fracture characterization of wood adhesive joints. Report
TVSM-1006. Lund University, Division of Structural Mechanics: Lund, Sweden;
1974.
Acknowledgment [28] Serrano E, Gustafsson PJ. Influence of bondline brittleness and defects on the
strength of timber finger-joints. Int J Adhes Adhes 1999;19(1):9–17.
[29] Milner HR, Yeoh E. Finite element analysis of glued timber finger joints. J
The authors would like to thank Dr. N. Lebaal for the fruitful Struct Eng 1991;117(3):755–66.
discussions and his help regarding the optimization strategy. Also, [30] Yeh MC, Lin YL. Finger joint performance of structural laminated bamboo
the financial support of the French Ministry of High Education and member. J Wood Sci 2012;58:120–7.
[31] Ratnasingam J, Scholz F. Optimization of finger-jointing in rubber wood
Research (MESR-2011) is gratefully acknowledged "Terres de processing. Eur J Wood Wood Prod 2009;67:241–2.
Hêtre" is gratefully acknowledged for providing beech timber used [32] Walford BG. Effect of finger length on finger-joint strength in radiate pine. In:
in the experiments. Proceedings of the 6th world conference on timber engineering. 2000.
[33] Rao S, Gong M, Chui YH, Mohammad M. Effect of geometric parameters of
finger joint profile on ultimate tensile strength of single finger-jointed boards.
Wood Fiber Sci 2012;4:263–70.
References [34] EN 385. Aboutages à entures multiples dans le bois de construction –
Exigences de performance et exigencies minimales de fabrication. mars 2002.
[1] EN 408. Timber structures – structural timber and glued-laminated timber – [35] Lee MJ, Cho TM, Kim WS, Lee BC, Lee JJ. Determination of cohesive
determination of some physical and mechanical properties. Bruxelles; Novem- parameters for a mixed-mode cohesive zone model. Int J Adhes Adhes
ber 2010. 2010;30:322–8.
[2] EN 14080. Timber structures – glued laminated timber and glued solid timber [36] Li S, Thouless MD, Waas AM, Schroeder JA, Zavattieri PD. Use of mode-I
– requirements. 2012. cohesive-zone models to describe the fracture of an adhesively-bonded
[3] Bourreau D, Aimene Y, Beauchêne J, Thibaut B. Feasibility of glued laminated polymer–matrix composite. Compos Sci Technol 2005;65(2):281–93.
timber beams with tropical hardwoods. Eur J Wood Wood Prod [37] Kafkalidis MS, Thouless MD. The effects of geometry and material properties
2013;71:653–62. on the fracture of single lap-shear joints. Int J Solids Struct 2002;39
[4] Collett BM. A review of surface and interfacial adhesion in wood science and (17):4367–83.
related fields. Wood Sci Technol 1972;6:1–42. [38] EN 302-1. Adhésifs pour structures portantes en bois, Méthode d'essai, partie
[5] Wake WC. Adhesion and formulation of adhesives. London: Applied Science 1: Détermination de la résistance du joint au cisaillement en traction long-
Publishers; 1976. itudinale. Edité et diffusée par l'Association Française de Normalisation
[6] Stehr M, Sltman J, Johansson I. Laser ablation of machined wood surfaces. 1. (ARNOR). 2004.
Effect on end-grain gluing of pine (Pinus silvestris L.) and spruce (Picea abies [39] EN 302-3. Adhésifs pour structures portantes en bois, Méthode d'essai, partie
Karst.). Holzforschung 1999;53(1):93–103. 3: Détermination de l'influence de l'attaque d'acide des fibres de bois,
[7] Stehr M. Laser ablation of machined wood surfaces. 2. Effect on end-grain résultant de traitement cycliques de température et d'humidité sur la
gluing of pine (Pinus silvestris L.). Holzforschung 1999;53(6):655–61. résistance à la traction transversale. Edité et diffusée par l'Association
[8] Stehr M, Johansson I. Weak boundary layers on wood surfaces. J Adhes Sci Française de Normalisation (ARNOR). 2004.
Technol 2000;14:1211–24. [40] Fortino S, Zagari G, Mendicino AL, Dill-Langer G. A simplified approach for
[9] Frazier CE, Ni J. On the occurence of network interpenetration in the wood- FEM simulation of mode I cohesive crack growth in glued laminated timber
isocyanate adhesive interphase. Int J Adhes Adhes 1998;18:81–7. under short-term loading. Raken Mek (J Struct Mech) 2012;45:1–20.
[10] Sernek M, Resnik J, Kamke FA. Penetration of a liquid urea–formaldehyde [41] Davidson BD, Sundararaman V. A single leg bending test for interfacial fracture
adhesive into beech wood. Wood Fiber Sci 1999;31:41–8. toughness determination. Int J Fract 1996;78(2):193–210.
[11] Buckley CJ, Phanopoulos C, Khaleque N, engelen A, Holwill MEJ, Michette AG. [42] Abaqus theory manual. Dassault Systèmes Simulia Corp. Providence: Rhode
Examination of the penetration of polymeric methylene di-phenyl-di- Island, U.S.A.; 2008.
isocyanate (pMDI) into wood structure using chemical state x-ray microscopy. [43] Oudjene M, Khelifa M. Finite element modelling of wooden structures at
Holzforschung 2002;56:215–22. large deformations and brittle failure prediction. Mater Des 2009;30
[12] Gindl W, Dessipri E, Wimmer R. Using UV microscopy to study diffusion of (10):4081–7.
melamine-urea-formaldehyde resin in cell walls of spruce wood. Holz- [44] Oudjene M, Khelifa M. Elasto-plastic constitutive law for wood behaviour
forschung 2002;56:103–7. under compressive loadings. Constr Build Mater 2009;23(11):3359–66.
V.-D. Tran et al. / International Journal of Adhesion & Adhesives 52 (2014) 40–47 47
[45] O'Loinsigh C, Oudjene M, Shotton E, Pizzi A, Fanning P. Mechanical behaviour [48] Lebaal N, Schmidt F, Puissant S. Design and optimization of three dimensional
and 3D stress analysis of multi-layered wooden beams made with welded- extrusion dies, using constraint optimization algorithm. Finite Elem Anal Des
through wood dowels. Compos Struct 2012;94(2):313–21. 2009;45:333–40.
[46] Guitard D. Mécanique du Matériau Bois et Composites. Collection Nabla. 1987. [49] Lebaal N, Oudjene M, Roth S. The optimal design of sheet metal forming
[47] Oudjene M, Ben Ayed L, Delamézière A, Batoz J-L. Shape optimization of processes: application to the clinching of thin sheets. Int J Comput Appl
clinching tools using the response surface methodology with Moving Least- Technol 2012;43(2):110–6.
Square approximation. J Mater Process Technol 2009;209(1):289–96.