Beruflich Dokumente
Kultur Dokumente
Samarasinghe, S. & Kulasiri, G.D. 2000. Displacement fields of wood in tension based on
image processing: Part 2. Crack-tip displacements in mode-I and mixed-mode fracture.
Silva Fennica 34(3): 261–274.
Near tip displacement fields for tensile loaded cracked rubber and wood with a crack
parallel-, perpendicular-to-grain, and a parallel-to-grain crack inclined 30°, 45°, and 60°
to the load axis were obtained from digital image correlation (DIC). Theoretical
displacements were also obtained for rubber and wood using isotropic and orthotropic
fracture theory, respectively. The results showed that DIC can reveal fine details of the
nature of displacements and the influences of crack tip in both rubber and wood.
Experimental crack tip displacements for wood compare well with theory; particularly,
when load is perpendicular-to-grain. Some anomalies were found in the tip displacements
in the direction of the tracheids due to the unique nature of their behaviour not accounted
for by theory. Mixed-mode crack tip displacement fields for wood clearly showed the
increasing influence of crack angle on the displacements, and the displacements perpen-
dicular to crack compared very well with theory. The displacements parallel to crack
showed some variations owing to the involvement of tracheids.
Keywords digital image correlation, fracture modes, orthotropic fracture, tip displace-
ment, wood
Authors´ address Lincoln University, Appl. Computing, Mathematics and Statistics Group,
P.O. Box 84, Canterbury, New Zealand E-mail kulasird@tui.lincoln.ac.nz
Received 19 October 1999 Accepted 23 August 2000
261
Silva Fennica 34(3) research articles
ter, K, stress intensity factor. Underlying assump- and 60° to the direction of loading. The reason
tion in these derivations is that material is homo- for testing rubber was to study displacement fields
geneous and isotropic. Sih et al. (1965) mathe- for a uniform isotropic material and make com-
matically formulated stresses and displacements parisons with the results obtained for highly vari-
near a crack tip in a remotely loaded orthotropic able and orthotropic wood.
plate. This derivation assumes that the material
is homogeneous, truly orthotropic (material sym-
metry coincides with geometric symmetry), and 2 Crack tip Displacement Fields
material properties in a given direction are con-
stant. In wood however, orthotropy is cylindrical Westergaard (1939) and Williams (1952, 1957)
or conical, material symmetry axes and geomet- derived formulae for horizontal (u) and vertical
ric symmetry axes of structural members do not (v) displacements of a point located a distance r
coincide most of the time, and material proper- from the tip and angle θ from the crack plane
ties are highly variable along the length and width. (Fig. 1) as given in Eqs. (1) and (2).
These theoretical issues and the importance of
the knowledge of fracture in the design of wood
structures prompted us to conduct a detailed study
of the behaviour of cracks in wood.
u=
KI
2µ
r
2π
Cos(θ 2) [κ − 1 + 2 Sin (θ 2)] (1)
2
1
u = KI
r
2π
Re
µ
1 − µ 2
(
µ1 p2 Cosθ + µ2 Sinθ − µ2 p1 Cosθ + µ1Sinθ (3)
)
1
v = KI
r
2π
Re
µ
1 − µ 2
(
µ1q2 Cosθ + µ2 Sinθ − µ2 q1 Cosθ + µ1Sinθ
) (4)
262
Samarasinghe & Kulasiri Displacement Fields of Wood in Tension Based on Image Processing: Part 2. ...
1
u = KI
r
2π
Re
µ
1 − µ 2
(
µ1 p2 Cosθ + µ2 Sinθ − µ2 p1 Cosθ + µ1Sinθ
)
(5)
1
+ K II
r
2π
Re
µ
1 − µ
(
2
p2 Cosθ + µ2 Sinθ − p1 Cosθ + µ1Sinθ
)
1
v = KI
r
2π
Re
µ1 − µ2
(
µ1q2 Cosθ + µ2 Sinθ − µ2 q1 Cosθ + µ1Sinθ
)
(6)
1
+ K II
r
2π
Re (
µ1 − µ2
q2 Cosθ + µ2 Sinθ − q1 Cosθ + µ1Sinθ
)
σ
where, KII is mode-II stress intensity factor and
other parameters are the same as described pre-
viously. Details of these derivations are avoided
for clarity and reader is referred to Sih et al.
(1965) for a complete review.
Digital image correlation (DIC) is a non-con-
tacting full-field displacement measuring tech- v
nique that has been used in few studies to obtain
displacement profiles for wood in compression
and bending, and in wood joints as discussed in u
y r
Part 1 of this study. Even fewer researchers have
applied the method to study fracture mechanisms.
Durig et al. (1991) applied the method to deter-
θ
mine stress intensity factors for aluminium in
mixed-mode fracture and McNeill et al. (1987) x
used it to obtain mode-I stress intensity factor
for plexiglass. The authors have been using the
method for crack tip displacement analysis and
the determination of stress intensity factors for
wood (Samarasinghe et al. 1996, Samarasinghe
and Kulasiri 1998, 1999).
σ
Fig. 1. Two dimensional near tip displacement compo-
nents of a unit area in a plate loaded in tension.
263
Silva Fennica 34(3) research articles
σ σ
(a) (b)
y (v) )
y (v
)
x (u
α
x (u)
σ σ
Fig. 2. Location of the analyzed area with respect to (a) Mode-I and
(b) inclined cracks.
264
Samarasinghe & Kulasiri Displacement Fields of Wood in Tension Based on Image Processing: Part 2. ...
265
Silva Fennica 34(3) research articles
266
Samarasinghe & Kulasiri Displacement Fields of Wood in Tension Based on Image Processing: Part 2. ...
(a) 6 6
X mm (a) X mm
4 4
2 2
0 0
0 0.005
V mm
U mm 0
–0.01
–0.005
2 2
0 0
Y mm –2 Y mm –2
(b) X mm 15
10 (b) X mm 15
5 10
5
0
0
0.01
0.006
V mm 0 0.004
–0.01 0.002
U mm
5 0
0 5
Y mm 0
–5
Y mm –5
Fig. 5. Vertical displacement fields for wood with Fig. 6. Horizontal displacement fields for wood with
a crack parallel-to-grain. (a) experimental profile a crack parallel-to-grain. (a) experimental profile
(b) theoretical profile. (b) theoretical profile.
due to more flexible lignin, undergoing large Negative u displacement indicates the possible
relative displacements near the tip. The nature of influence of mode II (shear) fracture in the vicin-
displacement at the four corners are very similar ity of tip, which may be due to crack not being
and the overall trend is also the same despite the perfectly collinear with grain. It is apparent that
fluctuations of experimental values. The magni- the resemblance shown here is not as close as
tude of displacement is also very similar in the that between experimental and theoretical values
experimental and theoretical plots. The most strik- for isotropic rubber. The magnitude of the ex-
ing difference between the two 3-D plots is the perimental u values compares well with that of
wavy profile from the tested sample and the theory and the u displacement is much smaller
uniform profile from theory. than v displacement shown in Fig. 5(a). Note
The u displacement measures horizontal dis- that u is measured along the direction of the
placement parallel-to-grain at points in front of tracheids. As already stated in Part 1 of this
the tip as a result of the applied stress perpendic- study, the profiles tend to be much noisier for the
ular to the grain. Displacement profiles obtained measurements made parallel-to-grain and a wav-
from digital image correlation and orthotropic iness propagates along the diagonal of the square
fracture theory are shown in Figs. 6(a) and 6(b), area indicating a possible shear influence.
respectively. The theoretical u values as shown The magnitude of v and u displacements is
in the plot closely resembles that obtained for within realistic ranges and compares well with
rubber (Fig. 4(b)). It is interesting to note that theory. The experimental profiles are not as
although the experimental profile is highly vari- smooth and the effect is enhanced in the direc-
able, a closer examination reveals a trend broad- tion parallel-to-tracheids. The results strongly
ly similar to that depicted by theory. Theory suggest the evidence of the influence of unique
assumes a purely orthotropic material whereas cellular structures on the crack tip displacements,
wood cannot be considered as truly orthotropic. which cannot be overlooked.
267
Silva Fennica 34(3) research articles
268
Samarasinghe & Kulasiri Displacement Fields of Wood in Tension Based on Image Processing: Part 2. ...
(a)
X mm 6
8 Specifically, crack tip is stationary with co-ordi-
2
4 nates (0,0), x axis lies in the crack plane, and y
0 axis is perpendicular to it. All displacements are
0.03 presented relative to the tip.
0.02 The theoretical displacements were also ob-
0.01 tained from orthotropic fracture theory. Here dis-
U mm 0
placement components for mode-I and II were
2
0
combined to obtain total u and v using Equations
Y mm –2 (5) and (6), respectively. Material properties are
–4 the same as the average properties used previ-
ously. KI and KII values were determined from
(b) 8
X mm
( )
6
2
4 KI = Y1σ πa Sin 2α (7)
0
0.04
0.03
KII = Y2σ ( )
πa Sinα Cosα (8)
0.02
0.01 where α is the crack angle with respect to the
U mm 0
2
load axis ((Fig. 2(b)), a is the crack length, σ is
0 the applied remote stress, Y1 is the mode I cor-
Y mm –2 rection factor (1.2978), Y2 is the mode II correc-
–4 tion factor (1.0) (Tada et al. 1985).
Fig. 8. Horizontal displacement fields for wood with
Before proceeding to the discussion of results,
a crack perpendicular-to-grain. (a) experimental
a look ahead at what can be expected will aid the
profile (b) theoretical profile.
discussion. In mixed-mode, the remote applied
stress (σ) can be resolved into normal (σI) and
shear (τII) stress components (i.e. mode I and
mode II stress components). These are:
crack loaded parallel-to-grain as predicted by
theory. It can be said that some general features σ I = σ Sin2α ; τ II = σ Sin α Cosα ;
depicted by the theory are followed by the tested For α = 30°, σ I = 0.25σ , τ II = 0.433σ ;
sample. However, experimental profiles are more For α = 45°, σ I = τ II = 0.5σ ;
complex as a result of the material not being For α = 60°, σ I = 0.75σ , τ II = 0.433σ .
truly orthotropic and the cellular structure that
appears to produce mode-II type influence on u The mode II shear stress is relatively constant
displacement, vertical slippage, and significant for the three angles tested with 45° angle show-
internal shear. Crack tip has a prominent influ- ing the highest shear stress. In contrast, influ-
ence on the horizontal displacement near the tip. ence of mode I fracture increases with crack
angle. Therefore, 60° angle must show the larg-
est combined total v displacement followed by
5 Mixed-mode Fracture 45° and 30° must show the least. As for u dis-
placement, mode I component should increase
In mixed-mode fracture, both opening and shear- with crack angle owing to increased normal stress
ing modes influence crack tip displacement fields; and the mode II influence is relatively constant
therefore v and u displacements are the com- for the angles tested due to similar shear stresses
bined total displacements due to both modes in for the three angles. With the aid of the above
respective directions. Crack was parallel-to-grain description, the following experimental and the-
in all mixed mode specimens and experimental oretical displacement profiles can be studied to
results are presented for a 9 mm × 9 mm area understand the crack tip behaviour under mixed-
near the tip as shown in Fig. 2(b) which also mode loading and the relevance of the theory to
shows the convention of the co-ordinate system. predict the same.
269
Silva Fennica 34(3) research articles
5.1 30° Crack Angle the overall magnitude of displacement also com-
pares reasonably well. Although there is a sharp
Fig. 9(a) shows the experimental 3-D plot for v displacement gradient near the tip as predicted
displacement and Figs. 9(b), (c), and (d) show by theory, the displacement along the border
theoretical mode-I, mode-II, and combined total closer to the tip not as smooth as in theory. This
v displacement respectively. At this angle shear gradient can be attributed to mode-I v displace-
stress is larger than normal stress and therefore, ment (Fig. 9(b)).
a dominant mode II influence can be expected. Due to the larger shear stress component, mode-
Theoretical plots in Fig. 9(b) and (d) reveal how II influence must be more prominent here than
mode-II v displacement component alters the that for v. Fig. 10(a) shows the experimental
already familiar mode I displacement profile to displacement plot and Figs. 10(b), (c) and (d)
produce the combined total displacement for a show theoretical mode-I, mode-II, and combined
truly orthotropic material. Specifically, there is a total displacements, respectively. According to
relatively large negative mode-II v displacement theory, mode-I u displacement is all positive,
away from tip which decreases towards the tip. whereas, mode-II contribution introduces a crack
The basic mode-II influence is a uniformly vary- tip gradient as well as positive and negative dis-
ing downward pull of the displacement profile placements above and below x axis respectively.
resulting in much increased negative total dis- As expected mode-II u displacement is much
placements for negative y values, particularly in greater than mode-I u displacement component.
the bottom right-hand region in the analyzed The theoretical combined total displacement plot
area. A variation of this interplay is expected as (Fig. 10(d)) reveals a very strong influence of
the crack angle increases. In the experimental mode-II displacement in the region below x axis.
plot, the overall pattern depicted by theory is Except for the top left hand corner, the experi-
maintained although the plot is not smooth and mental profile vaguely follows the general trend
(a) 15 15
X mm (c) X mm
10 10
5 5
0 0
0.005 0
vII mm
V mm 0 –0.002
–0.005
–0.004
–0.01 –0.006
–0.015
10
10
7.5
7.5
5
5 Y mm
Y mm 2.5
2.5 0
0
–2.5
–2.5
(b) 15
15 (d) X mm
X mm 10
10
5 5
0 0
0.01
0.01
0.005
0.005
0
vI mm 0 (vI + vII) mm
–0.005
–0.005
10 –0.01
7.5 10
5 7.5
Y mm 2.5 5
0 2.5
Y mm
–2.5 0
–2.5
Fig. 9. Vertical displacement fields for 30° crack angle to load in wood with crack parallel-to-grain.
(a) experimental profile (b) theoretical mode-I component (c) theoretical mode-II component and
(d) combined mode-I and mode-II (total) theoretical profile.
270
Samarasinghe & Kulasiri Displacement Fields of Wood in Tension Based on Image Processing: Part 2. ...
(a) X mm 15 (c) X mm 15
10 10
5 5
0 0
0.04
0.005
0.02 uII mm
0
U mm 0
10 10
7.5 7.5
5 5
2.5
Y mm 2.5 0 Y mm 0
–2.5
–2.5
(b) X mm 15 (d) X mm 15
10 10
5 5
0 0
0.006 0.015
0.004 0.01
0.002 0.005
uI mm 0 (uI + uII) mm 0
10 10
7.5 7.5
5 5
2.5
Y mm 2.5 0 Y mm 0
–2.5 –2.5
Fig. 10. Horizontal displacement fields for 30° crack angle to load in wood with crack parallel-to-grain.
(a) experimental profile (b) theoretical mode-I component (c) theoretical mode-II component and
(d) combined mode-I and mode-II (total) theoretical profile.
depicted by theory; however, it appears that the overall effect of this is smaller compared to that
severe distortion present in the plot makes it for 30° crack angle showing the effect of in-
deviate from the predicted pattern. Note that the creased normal stress. Also, there is a much great-
displacement is parallel to tracheids. In both plots er resemblance between experimental and theor-
all u displacements are positive; however, the etical plots for 45° crack angle compared to 30°
experimental values are larger than predicted, angle. Experimental plot also reveals a steep
which have been accentuated by peaks in the displacement gradient near the tip as predicted
plot. The u values exceed v displacement in both by theory. In this case, the overall pattern and
theoretical and experimental plots. magnitude of displacements are similar in experi-
ments and theory. As expected, v displacement
for 45° crack angle is greater than that for 30°
5.2 45° Crack Angle angle as can be seen from Figs. 11(a) and 9(a).
The experimental u displacement profile is
At this angle normal and shear stress compo- given in Fig. 12(a) and theoretical combined
nents are equal. Fig. 11(a) displays experimental total displacement plot is shown in Fig. 12(b).
displacement profile and Fig. 11(b) shows theor- According to theory, mode-I displacement has
etical combined total v displacement, respec- considerably increased compared to that for 30°
tively. According to theory, both mode-I and II crack angle and mode-II components have also
components (not shown here) are higher than slightly increased (not shown here). The com-
those for 30° angle and the total displacement is bined displacement plot reveals that the overall
predominantly due to mode-I contribution. The effect of mode-II is less than that for 30° angle
effect of mode-II is to pull the displacement owing to the influence of the larger normal stress.
down in the negative direction. However, the Much smoother experimental u displacement
271
Silva Fennica 34(3) research articles
(a) 20 (a) 20
X mm 15 X mm 15
10 10
5 5
0 0
0.01 0.04
V mm 0 0.02
–0.01 U mm 0
–0.02 –0.02
5 5
0 0
Y mm Y mm
–5 –5
20 (b) X mm 20
(b) X mm 15 15
10 10
5 5
0 0
0.01
0.01
0
(vI + vII) mm 0 (vI + vII) mm
–0.01
–0.01
–0.02
–0.02
5 5
0 0
Y mm Y mm
–5 –5
Fig. 11. Vertical displacement fields for 45° crack Fig. 12. Horizontal displacement fields for 45° crack
angle to load in wood with crack parallel-to-grain. angle to load in wood with crack parallel-to-grain.
(a) experimental profile (b) theoretical profile. (a) experimental profile (b) theoretical profile.
is evident from Fig. 12(a) where overall dis- here) and reveals the largest overall v displace-
placement is greater than those for 30° angle, ment for this angle compared to 45° and 30°
that confirms the expected trend. Comparison of crack angles. The mode-II component is similar
the experimental results with the theoretical pro- to that for 45° angle. The increased mode-I con-
file indicates a general agreement in the trend tribution diminishes the effect of mode-II even
above x-axis but below x-axis u is negative for more so than that for 45° angle. Experimental
the most part in experimental plot. The u dis- profile more resembles theory than that for the
placement below the x-axis is positive in theor- other two angles and it is interesting to note that
etical plots. Recall that negative u displacement the degree of disturbance in the experimental
below the x-axis comes from mode II contribu- plots decreases as the crack angle increases. The
tion. Thus, experimental results show a stronger theoretical plot captures the most important fea-
influence of mode II than indicated by theory. tures of the experimental profile including the
As for 30° angle, experimental displacements steep gradient at the tip and the overall magni-
are larger than predicted. tudes in the two plots compare well. One notice-
able feature though is that for 60° angle crack tip
influence diminishes much more slowly towards
5.3 60° Crack Angle the back border in the experimental results com-
pared to the theoretical profile.
At 60° normal stress component is the largest for Owing to high normal stress, u displacement
the angles tested. Here σI = 0.75σ and τII = for 60° is expected to be larger than that for 45°
0.433σ. The experimental profile for v displace- and 30°. Fig. 14(a) shows the experimental plot
ment is displayed in Fig. 13(a) and theoretical which indeed shows that the displacement is al-
total v displacement is shown in Fig. 13(b). As most twice as large as that for 30° and roughly
can be expected, theory indicates a much larger about 50 % greater than that for 45° angle. This
mode-I contribution than mode-II (not shown plot indicates that the u displacement plots also
272
Samarasinghe & Kulasiri Displacement Fields of Wood in Tension Based on Image Processing: Part 2. ...
(a) X mm 10 15 (a) X mm 15
10
5 5
0 0
0.04 0.06
0.02 0.04
0 0.02
V mm
U mm 0
–0.02
–0.02
5
5
Y mm 0
–5 Y mm 0
–5
(b) X mm 15
10 (b) 15
X mm 10
5
0 5
0
0.02
0.02
0 0.015
(vI + vII) mm
0.01
–0.02
0.005
5 (uI + uII) mm 0
5
0
Y mm 0
–5 Y mm –5
Fig. 13. Vertical displacement fields for 60° crack Fig. 14. Horizontal displacement fields for 60° crack
angle to load in wood with crack parallel-to-grain. angle to load in wood with crack parallel-to-grain.
(a) experimental profile (b) theoretical profile. (a) experimental profile (b) theoretical profile.
get smoother as the crack angle increases. Fig. 45°, and 60° to the crack plane using digital
14(b) shows the combined total u displacement. image correlation method. The displacement pro-
According to theory, mode-I component has in- files thus obtained revealed intricate details of
creased but mode-II component has stayed al- the mechanisms of load transfer in isotropic rub-
most at the same level as that for 45° angle (plots ber and anisotropic wood with a crack confirm-
not shown here). The overall effect is an in- ing that the digital image correlation is a useful
creased total displacement compared to the other method for detailed analysis of crack tip dis-
two angles. Fig. 14(b) clearly shows the dimin- placements. For cracked rubber and wood mem-
ished mode-II influence owing to high mode-I bers, theoretical displacements were also com-
contribution. The experimental plot more resem- puted and compared with experimental values.
bles theory than those for the other two angles. The experimental plots for mode-I fracture
As shown earlier, experiments reveal a larger clearly showed the remarkable influence of the
mode-II contribution than predicted leading to tip on both u and v displacements; particularly
negative u values below x axis whereas in theory the major changes in the u displacement profile
u is positive. Once again, experimental displace- compared to that without a crack. Trend and
ments are larger than predicted values. magnitude of displacement in theoretical and ex-
perimental plots for wood compared well. As
found for uncracked wood specimens in Part 1
6 Summary and Conclusions of this study, displacements parallel-to-tracheids
in cracked members produce complex displace-
Near tip displacements were obtained for cracked ment profile subjected to slippage and distorted
rubber and wood- with a crack parallel- and per- by internal shear stress, whereas, displacement
pendicular-to-grain, and an inclined crack locat- perpendicular-to-tracheids produce comparative-
ed along the grain of wood but loaded at 30°, ly smoother profiles.
273
Silva Fennica 34(3) research articles
274