Anda di halaman 1dari 13

Mechanical behaviour of wood in the orthotropic directions

J. L. Morais1, J. C. Xavier1, N. M. Dourado1, J. L. Lousada2

Departamento de Engenharias, ICETA/UTAD, Portugal
Departamento de Florestal, ICETA/UTAD, Portugal

A numerical and experimental work was performed to examine the suitability of several
specimen configurations (straight-sided, dog-bone and notched) for the determination of
the tensile strength of clear Pinus Pinaster wood, in the orthotropic directions. A 3-D
linear finite element model of the specimens was used to assess the uniformity and
purity of stress and strain fields in their gauge section. It was found that the straight-
sided and dog-bone specimens are reliable geometries for characterizing the initial
linear elastic response. However, the results predicted by the finite element method
using the Tsai-Hill failure criterion and the experimental observations, clearly indicate
that failure is due to stress concentration effects and occurs under a complex stress state.

1 Introduction

A fundamental requirement for effective utilisation of wood as a competitive structural

material is the accurate knowledge of its mechanical properties. However, the
experimental identification and the analytical modelling of mechanical behaviour of
wood remains an open problem, due to its natural variability, inhomogeneity and
Wood as long been recognised as an orthotropic material [1], with an internal structure
which is characterised by the existence of three mutually perpendicular planes of
symmetry. Those planes are defined by the longitudinal direction (L) along the fibres,
the radial direction (R) towards annual rings and tangential direction (T) to the annual
rings. To fully characterize the mechanical behaviour of wood it is necessary to know
the stress-strain relationships referred to the LRT reference frame. The mechanical tests
are the only way to obtain such data, but several difficulties arise in making the right
experimental measurement, particularly those concerning the identification of ultimate
In the framework of synthetic composites an intensive research effort has been devoted
to the experimental identification of stress-strain relationships, including the ultimate
stresses [2-8]. The same does not hold for solid wood, for which there are few studies
about its orthotropic mechanical behaviour [9-14]. The majority of published works
devoted to this subject focused on the initial elastic behaviour. Indeed, the structural
computations are in general built in the hypothesis of a wood linear elastic behaviour
[15-17]. However, it is clear from the results published by several authors that wood
possesses a non-linear mechanical behaviour [18]. Hence, a more realistic
representation of mechanical behaviour of wood is needed for some structural problems,
like those in the mechanics of timber joints [18].
The main problem to overcome concerning the experimental identification of complete
stress-strain relationships of solid wood, in its principal material axes (L, R and T axes),
is the determination of the true ultimate stresses. This is not a trivial task owing to the
orthotropic nature of wood. The current standards are unsatisfactory to determine the
true strengths of wood, namely the shear strengths and the strength perpendicular to the
grain [14-19]. To achieve the most suitable test methods more detailed investigations
must be performed about the influence of specimen geometry, size effects, loading
mode and gripping arrangements.
The objective of this work is to evaluate the most accurate mechanical tests for the
experimental identification of on-axis tensile strength of clear wood. This is a part of a
more broad investigation recently undertaken by the authors, about the non-linear
mechanical behaviour of wood.

2 Experimental work

Logs from Pinus Pinaster wood were live-sawn into boards, from which matched
samples of mature wood were cut. Several specimens were prepared to be loaded in the
L, R and T directions. Three types of specimen geometries were examined: straight-
sided, dog-bone and notched specimens. Figures 1 to 3 shows these specimens
configurations and the corresponding nominal dimensions. The specimen ends were
reinforced with fibreboard tabs, in order to prevent any damage introduced by the
loading fixture. All the specimens were tested in the green state, immediately after






40 30 11 39

Figure 1. Geometry and dimensions of longitudinal specimen



40 3 15 3






40 9.4 41.2 3 15 3





40 15.6 28.8 3 15 3


Figure 2 a, b, c. Configurations of radial specimens

(a) straight-sided specimen
(b) dog-bone specimen
(c) notched specimen



40 3 15 3

Figure 3. Geometry and dimensions of tangential specimen

Quasi-static tensile tests were conducted in an Instron 1125 universal testing machine,
under a crosshead speed of 1 mm / min . Wedge action grips were employed, which
prevent the rotation of the specimen ends. One of the grips remains fixed while the
other is displaced at a constant speed (1 mm / min ) until specimen failure. All the tests
were performed in the laboratory temperature and moisture conditions.

3 Finite element analysis

The numerical modelation of uniaxial tensile tests has been achieved by the finite
element method, using the commercial software ANSYS 5.6.2. A 3-D linear elastic
analysis was performed based on the SOLID 45 element, from the ANSYS finite
element library [20]. This element is defined by eight nodes and orthotropic directions,
with a total 24 degree of freedom.
The specimen geometries considered in the finite element analysis were the dog-bone
and notched geometries (see Figs. 1 to 3). The finite element mesh generated to model
such specimens has a number of elements between 1200 and 648. A more refined mesh
does not significantly alter the results obtained.
Fig. 4 illustrates the boundary conditions applied to the model, which are intended to be
in agreement with the non-rigid and non-rotating gripping arrangement. On one side of
the specimens, the nodes along one of the specimen edge were fixed, while a
longitudinal displacement was imposed on the other side. The prescribed nodal
displacements allow the free contraction in the thickness direction due to Poisson effect.
Although these are idealized boundary conditions, a comparison of numerical and
experimental results can be made. A plot of the deformed model was used to check the
boundary conditions of the elements.
In the finite element analysis described above the material properties are those
corresponding to the Pine Loblolly wood [21], which are presented in Tab. 1.

Figure 4. Boundary conditions for the longitudinal specimen

Table 1. Mechanical properties of Pine Loblolly
[GPa] [GPa] [GPa] [GPa] [GPa] [GPa]
13,530 1,529 1,055 0,382 0,292 0,328 0,176 1,096 1,109

Strength parallel to grain Strength perpendicular to Shear parallel to grain

[kPa] grain [kPa ] [kPa]
80,000 3,200 9,600

4 Results and discussions

In the ideal test specimen, the stress and strain fields are pure and uniform, so that
failure occurs in a single mode and is not influenced by stress concentrations or
complex stress states. The objective of the finite element study previously described is
just to assess the degree of purity and uniformity of strain field in the gauge section of
the specimens, and attempt to determine the failure location.
After examining the u , v and w displacement fields (see Fig. 5 to 7) it was found that
the dog-bone specimen exhibit an extended region on the gauge section where the strain
field is almost uniform. The stress state in that region is also a pure and homogeneous
uniaxial stress state. Consequently, accurate measurements of material response can be
made using strain gages bonded to the specimen surface. The same does not hold for
the notched specimens, used by Stauzl-Tschegg et al. [22], because the strain gages
usually cover a larger region than the region where exists a uniform field of linear
strains and zero shear strains. The displacement fields presented in the Figs. 5 to 7 are
representative of the actual displacement field, but the predicted magnitude is not in
agreement with the actual ones because the displacement distribution depends on the
material properties.

Figure 5 a, b, c (continue). Displacement field in longitudinal dog-bone specimen


Figure 5 a, b, c (continue). Displacement field in longitudinal dog-bone specimen

(a) u displacement component
(b) v displacement component
(c) w displacement component


Figure 6 a, b, c (continue). Displacement field in Radial dog-bone specimen



Figure 6 a, b, c (continue). Displacement field in Radial dog-bone specimen

(a) u displacement component
(b) v displacement component
(c) w displacement component


Figure 7 a, b, c (continue). Displacement field in Radial notched specimen


Figure 7 a, b, c (continue). Displacement field in Radial notched specimen
(a) u displacement component
(b) v displacement component
(c) w displacement component

Since linear elastic behaviour is assumed, the location of initial failure can be calculated
by comparing the element stresses with the clear wood strength via the failure criteria.
In this work the Tsai-Hill criterion [23] was used, which accounts for differences in
strengths in principal material directions. The finite element results depicted in Figs. 7
a, b, c suggest that the specimen geometries examined (see Figs. 1 to 3) are not
appropriate to identity the true strengths of wood parallel to the grain and perpendicular
to the grain.
The macrofractography of the tangential specimens (see Fig. 8) demonstrates that the
straight-sided geometry work very well, because the failure occurs in the gauge section
away from the gripping ends. Additionally, it can be seen that the fracture surface is not
oriented in the loading direction but is always fairly oriented in the tangential direction.
This fact is a clear indication of a mixed mode failure, hence the recorded tensile
strength (see Tab. 2) is not the true tensile strength of wood in the tangential direction.
The types of failure in the Radial specimens are shown in Fig. 9. In all the tested
geometries stress concentration effects influence the failure, except for the dog-bone
configuration. Although the finite element analysis predicts the failure occurrence in the
notch root of dog-bone specimens, most of them failed in the gauge section, probably
due to the spatial distribution of wood defects. However the measured strength in the
radial direction (see Tab. 2) must be used with caution.



Figure 7 a, b, c. Tsai-Hill equivalent stress, normalized by its maximum value

(a) longitudinal specimen
(b) radial dog-bone specimen
(c) radial notched specimen

As illustrated in Fig. 10 the failure in the longitudinal specimens occurs in the notch
root as was predicted by the Tsai-Hill criteria, so that the measured strength parallel to
the grain (see Tab. 2) is below the true strength.
Table 2. Tensile strength of Pinus Pinaster

Specimen Tensile Strength [KPa ]

Tensile straight sided

(see Fig. 8 a)
Tensile straight sided
(see Fig. 8 b)
Radial straight sided
(see Fig. 9 a)
Radial dog-bone
(see Fig. 9 b)
Radial dog-bone
(see Fig. 9c)
Radial notched
(see Fig. 9d)
Longitudinal straight sided
(see Fig. 10)



Figure 8 a, b. General appearance of failure in the tangential specimens

(a) straight sided specimen
(b) straight sided specimen
(a) (b) (c) (d)
Figure 9 a, b, c, d. Typical failure of Radial specimens
(a) straight sided specimen
(b) dog-bone specimen
(c) dog-bone specimen
(d) notched specimen

Figure 10. Typical failure of longitudinal specimen

5 Conclusions

A numerical and experimental investigation has been carried out to evaluate the
performance of different specimen configurations for the determination of on-axis
mechanical behaviour of wood. Three types of specimen geometries were examined:
straight-sided geometry, dog-bone geometry and notched geometry. The straight-sided
specimen and the dog-bone specimen are suitable for experimental identification of the
initial linear mechanical response, as was confirmed by a 3-D finite element analysis.
However, none of the tested specimens are appropriate to measure the tensile strengths
of wood in its principal material direction. The predicted and observed failure is
influenced by stress concentration effects and occur under combined rather than pure
uniaxial tensile loading.

7 References

1. Schniewird, A. P. and Barret, J. D. (1972): Wood as a linear orthotropic viscoelastic

material. Wood Sci. Techn.,6(1): 43-57
2. Chamis, C. C. And Brinson, J. H. (1977): Ten-deg off-axis test for shear properties in
fiber composites. Exp. Mech., September: 339-346
3. Voloshin, A. and Arcan, M. (1980): Failure of unidirectional fiber-reinforced
materials new methodology and results. Exp. Mech., August: 280-284
4. Walrath, D. E. and Adams, D. F. (1983): The Iosipescu shear test as applied to
composite materials. Exp. Mech., March: 105-110
5. Sims, G. D. (1992): Development in harmonisation of standards for polymer matrix
composites. Composites Testing and Standardisation, Hoggs, P. J. et al. (eds.), ECCM-
CTS, EACM, September: 3-20
6. Hojo, M., Sawada, Y. And Miyairi, H. (1994): Influence of clamping method on
tensile properties of unidirectional CFRP in 0 and 90 directions round robin activity
for international standardisation in Japan. Composites, 25 (8): 786-796
7. Hung, S.-C. and Liechti, K. M. (1999): Finite element analysis of the Arcan specimen
for fiber reinforced composites under pure shear and biaxial loading. J. Comp. Mat.,
33 (14): 1288-1317
8. Odegard, G. and Kumosa, M. (1999): Elasto-plastic analisys of the Iosipescu shear
test. J. Comp. Mat., 33 (21): 1981-2001.
9. Liu, J. Y. (1984): New shear strenght test for solid wood. Wood and Fiber Science,
16 (4): 567-574
10. Sliver, A. and Yu, Y. (1993): Elastic constants for hardwood measured from plate
and tension tests. Wood and Fiber Science, 25 (1): 8-12
11. Sliker, A., Ying, Y. Timothy, W. and Wenjie, Z. (1994): Orthotropic elastic
constants for eastern hardwood species. Wood and Fiber Science, 26 (1): 107-121
12. Oudjehane, A. and Raclin, J. (1995): On the influence of orientation upon the
mechanical behaviour of oak wood in a general state of stress. Wood Sci Techn., 29
(1): 1-10
13. Liu, J. Y., Ross, R. J. And Rammer, S. R. (1996): Improved Arcan shear test for
wood. Proc. of the International Wood Engineering Conference, Gopu and Vijaya, K.
A. (eds.), New Orleans, 2, 85-90
14. Liu, J. Y. (2000): Effects of shear coupling on shear properties of wood. Wood and
Fiber Science, 32 (4): 458-465
15. Cheung, C. K. and Sorensen, H. C. (1983): Effect of axial loads on radial stress in
curved beams. Wood and Fiber Science, 15 (3): 263-275
16. Zalpeh, B. L. and McLain, T. E. (1992): Strength of wood beams with fillet interior
notches: a new model. Wood and Fiber Science, 24 (2): 204-215
17. Pellicane, P. J. and France, N. (1994): Modeling wood pole failure. Part 1: finite
element stress analysis. Wood Sci Techn., 28 (3): 219-228
18. Bouchair, A. and Vergne, A. (1995): An application of Tsai criterion as a plastic
flow law for Timber bolted modelling. Wood Sci Techn., 30 (1): 3-19
19. Booth, L. G. (1991): Strength of timber in tension perpendicular to the grain: UK
procedures past, present and future J. Inst. Wood Sci., 12 (3): 131-142
20. Wood Handbook, (1999) Gen. Tech. Rep. FPL-GTR-113, U.S. Department of
Agriculture, Forest Service, Forest Products Laboratory, 463 p.
21. Stanzl-Tschegg, S. E., Tan, D. M. and Tschegg, E. K., (1995): New splitting method
for wood fracture characterization. Wood Sci Techn., 29 (1): 31-50
23. Azzi, V. D. And Tsai, S. W. (1995): Anisotropic Strength of composites. Exp.
Mech., 5 (9): 283-288


We would like to thanks to the Foundation for Science and Technology from the
Ministry of Science and Technology for the financial support in the execution of this
work, in the ambit of the project POCTI/1999/EME/36270, Non-linear mechanical
behaviour of wood.