structural grading of old chestnut elements by bending and

STRUCTURAL GRADING OF OLD CHESTNUT ELEMENTS BY
BENDING AND COMPRESSION TESTS
Beatrice Faggiano1, Maria Rosaria Grippa1, Anna Marzo1, Federico M. Mazzolani1
ABSTRACT: The paper deals with the mechanical characterization of timber elements based on bending and
compression tests, according to UNI and ISO codes. The investigations were performed on both structural elements and
defects-free specimens, made of old chestnut wood, aiming at evaluating the influence of typical defect patterns and
wood anatomy on the elastic characteristics, strength properties as well as failure mechanisms of the material. In
particular, the behaviour in compression of clear wood under uniaxial loading is studied with different orientations as
respect to the longitudinal, radial and tangential directions. The experimental results are examined in relation with the
standard values assumed by Italian codes for the quality classes assignment by visual grading.
KEYWORDS: Chestnut wood, Bending tests, Compressive behaviour, Mechanical properties, Failure mechanisms.
1 INTRODUCTION123
In the field of analysis and restoration of ancient timber
constructions, the assessment of the mechanical
properties of the elements becomes difficult due to the
strength high variability of the material among and
within members. Timber strength grading is usually
determined by means of visual inspection, through the
identification of the critical areas, defects and decay.
Accordingly, the UNI 11035-2 (2003) standard provides
strength visual grading rules and characteristics values
for Italian structural timber population, whereas the UNI
11119 (2004) code establishes procedures and criteria in
order to assess the performances of timber members by
means of in situ inspection, providing the admissible
mechanical properties for three category classes.
However, the estimation of the serviceability properties
of new and/or ancient timber constructions by means of
the visual grading method is not entirely reliable due to
many factors influencing the mechanical properties and,
further, the biased influence of the human factor.
Moreover, the information is mostly qualitative.
Therefore, a careful material characterization is needed
by means of standardized laboratory tests, even only for
few sacrificial specimens.
In this context, aiming at the structural identification of
old chestnut elements (Castanea sativa Mill.), a wide
experimental campaign, including destructive tests in
bending and compression, was carried out at the
Laboratory of the Department of Structural Engineering
1
Beatrice Faggiano, Maria Rosaria Grippa, Anna Marzo,
Federico M. Mazzolani, Department of Structural Engineering
(DIST), University of Naples “Federico II”, Naples, Italy.
Email: [email protected]; [email protected];
[email protected]; [email protected].
(DIST) of the University of Naples “Federico II”. The
research activity was developed in the framework of the
Italian project PRIN 2006 “Diagnosis techniques and
totally removable low invasive strengthening methods
for the structural rehabilitation and the seismic
improvement of historical timber structures”, Prof. M.
Piazza coordinator, Dr. B. Faggiano scientific
responsible of the research unit UNINA (University of
Naples “Federico II”) [1].
The methods specified in UNI EN 408 (2004) and UNI
ISO 3787 and 3132 (1985) devoted codes were used for
identifying the stiffness and strength properties of the
material, together with typical failure mechanisms.
In this paper, the local and global bending behaviour of
full-scale beams is described. The mechanical evaluation
in compression parallel to grain is analyzed by
comparing the overall behaviour of structural elements
with the one of defect-free specimens, pointing out the
influence of defects in timber performance [1-3].
Furthermore, the orthotropic nature of the clear wood is
investigated by means of compression tests
perpendicular to grain in radial and tangential loading
orientations. For each type of test, the experimental
results are examined through test statistics, they being
compared with the standard properties assumed by
Italian codes.
2 MATERIAL AND TEST METHODS
Structural elements in actual dimensions were provided
from chestnut (Castanea sativa Mill.) timber trusses roof
of an ancient masonry building of Naples, which was
dismantled in a recent restoration intervention.
From the trusses struts, 10 elements were selected for
bending tests (specimens type SA-B). They had an
irregular circular shape with mean diameter (D) ranging
between 15 to 16.5 cm and length equal to about 19D.
From the king posts, 14 samples were obtained for
compression tests parallel to grain (specimens type SAC). They had a nearly square cross-section, characterized
by large rounded edges, with a mean equivalent diameter
(D) ranging between 14.5 to 16 cm and length of about
6D. The SA specimens had standard dimensions
according to UNI EN 408. The conservation state of the
elements was examined by means of visual inspection,
checking wood features and defects, mainly consisting in
knots, cracks, ring shakes, slope of grain, biological
damage and holes, due to nails and insect attacks.
Therefore, as a result of the visual structural grading,
according to UNI 11119 standard, all specimens were
assigned to the third category class, also providing a
quick means for identifying critical areas.
After the destructive tests, by cutting the undamaged
parts of SA-C samples, 20 structural elements in small
dimensions, 5×5×30 cm sized, were obtained for
longitudinal compression tests (specimens type SS-C).
Furthermore, several defect-free specimens of clear
wood were also extracted, with no macroscopic defects
and alterations (specimens type DF). They had 40 mm
height and 20×20 mm2 cross-sectional area with the
longitudinal axis being oriented along the grain,
according to UNI ISO Italian codes. Therefore, in order
to study the behaviour in compression of the base
material in both parallel and perpendicular to grain
directions, three groups of specimens were prepared
taking into account the orientation of the annual rings
with respect to the direction of the applied load. The first
group consists of 33 specimens for longitudinal tests
(type DF-CL), whereas the second and third groups of 22
specimens for both radial (type DF-CR) and tangential
(type DF-CT) loading orientations.
The geometric features of the specimens are illustrated in
Figure 1, whereas in Table 1 the moisture content (MC)
and density (ρ) values, together with the test types in
bending (B) and compression (C) are summarized.
3.1 TESTING EQUIPMENT AND SET-UP
Bending tests were performed under force control on 10
beams in actual sizes (type SA-B), using the Mohr
Federhaff AG testing machine, a loading cell HBM of
740 kN and displacement transducers (LVDT).
According to the four-points static scheme, the tested
beams were loaded with two concentrated forces, applied
in the third of the beam span by interposing a rigid steel
frame between the actuator and the specimen (Fig. 2).
Figure 2: Bending tests: test arrangement and set-up.
3.2 ELASTIC CYCLES
Three cycles in elastic ranges, equal to 3-6-9 kN, were
carried out using a quasi-static loading procedure and
limiting the maximum force within the elastic
conventional limit (0.4 Fmax). Displacements (w) and
corresponding applied loads (F) have been fitted in F-w
curves (Fig. 3), in order to evaluate both local (Em,l) and
global (Em,g) elastic moduli, according to UNI EN 408.
14
F
[kN]
12
a = 6D
F/2
L1 = 5D
LVDT1
a = 6D
LVDT1
Av. LVDT 2, 3
LVDT 4
F/2
D
LVDT2
F/2
10
LVDT4
LVDT3
F/2
L = 18D
8
6
4
SA-C
SA-B
3 BENDING TESTS
2
w [mm]
L ≈ 6D
0
D
L ≥ 19 D
DF
b
D
10
14
20
33
22
22
MC
[%]
11-12
11-12
10-11
9-11
9-11
9-11
ρ
[kg/m3]
526-638
530-642
483-641
421-720
432-638
396-668
6
8
10
12
14
16
18
3.3 FAILURE CYCLES
Table 1: Specimens physical properties and test types.
n.
4
2b
SS-C
Figure 1: Specimens geometric features.
Specimens
type
SA-B
SA-C
SS-C
DF-CL
DF-CR
DF-CT
2
Figure 3: Bending tests: typical elastic F-w curves.
b
L = 6b
0
B
x
Test types
C // C ⊥
x
x
x
x
x
After the elastic cycles, it was possible to evaluate the
bending strength (fm) and the post-elastic response by
increasing the load up to failure in several loadingreloading cycles, reaching the maximum force within
300±120 sec. (UNI EN 408). For each specimen, the
destructive tests results are provided in terms of applied
force (F) versus loading-actuator displacement (w) (Fig.
4a). F-w envelope curves of all tested beams are depicted
in Figure 4b. It is generally observed that the initial
growing branch presents a linear behaviour up to the
maximum applied load; after the second or third cycle,
an evident reduction of strength occurs, together with
important actuator displacements.
For almost all the specimens the failure modes were
triggered around large knots located at the central zone
and at the tensile edge of the beam cross-section,
manifested by tearing of the more stressed fibres. In any
case an evident buckling in compression was observed,
followed by the propagation of slip phenomena (Fig. 4c).
50
F
[kN]
40
I cycle
II cycle
III cycle
IV cycle
V cycle
VI cycle
VII cycle
VIII cycle
IX cycle
30
20
w [mm]
20
40
60
80
F
[kN]
40
100
120
Failure cycles
fm (I)
fm (II) fm (III)
32.23 14.20 12.30
41.09 35.34 25.61
46.91 46.90 36.00
11.63 26.24 35.58
4.1 STRUCTURAL ELEMENTS
0
0
5-perc.
average
max
CV [%]
Elastic cycles
Em,l
Em,g
11023 10885
12535 12849
15291 15374
10.68 11.02
4 COMPRESSION TESTS
10
50
2
Table 2: Bending tests results [N/mm ].
a)
SA-B; n=10
30
4.1.1 Testing equipment and set-up
The compression tests parallel to grain on the structural
elements in actual (type SA-C) and small (type SS-C)
sizes were carried out under force control using the
Mohr & Federhaff AG machines of 5000 kN and 400 kN
capacity, respectively. In any case, the hydraulic press is
constituted by a top fixed head and a moveable loadingbase plate. Loading cell HBM and displacement
transducers were used during the tests (Fig. 5).
20
SA-C
SS-C
10
w [mm]
0
0
20
40
60
80
100
120
b)
c)
Figure 4: Bending tests: a) Typical F-w failure cycles;
b) Envelope curves; c) Failure modes.
3.4 RESULTS AND DISCUSSION
Table 2 provides the mean statistical parameters of
bending tests results. It can be stated the agreeable
homogeneity of the bending behaviour in terms of both
stiffness (Em,l; Em,g) and maximum strength (fm(I)), as the
pleasant coefficients of variation (CV) confirm.
However, the reduction of the load bearing capacity,
after the peak load reached during the tests, is
emphasized by the strong strength variation in both
second (fm(II)) and third (fm(III) failure cycles.
By comparing the experimental results with the standard
values assumed by the Italian codes, the following
observations can be made:
▪ The experimental average global elasticity modulus
is about 22% and 60% higher than the mean values
provided by UNI 11035-2 (11000 N/mm2) and UNI
11119-III (8000 N/mm2), respectively;
▪ Good agreement is found between the experimental
5-percentile strength and the standard characteristic
value of UNI 11035-2 (28 N/mm2);
▪ The safety coefficient, obtained by the ratio between
the experimental 5-percentile and admissible bending
strength provided by UNI 11119-III (8 N/mm2), is
equal to 4.
Figure 5: Compression tests on structural elements:
test arrangement and set-up.
4.1.2 Destructive tests
The investigations on the behaviour in compression of
the structural elements type SA-C were performed in
several failure cycles. In each step the load was
increased up to failure with a constant velocity gradient,
so that the maximum force was attained within 300±120
sec. (UNI EN 408). The experimental data are fitted in
stress-strain diagrams. The behaviour in compression in
post-peak field was examined by means of envelope
curves (Fig. 6a), which are plotted in Figure 6b for all
tested samples. The global response reveals a initial
linear portion, after that distinct non linearities are
manifested, with softening branches characterized by a
abrupt reduction of the stress carrying capacity. Collapse
modes observed were mainly characterised by the
attainment of the splitting almost parallel to grain,
accompanied by fractures in the radial direction,
cleavages along the annual growth rings or in multidirections, developed due to the presence of cracks and
ring shakes (Fig. 6c).
The stress-strain curves of the elements in small
dimensions (type SS-C) are shown in Figure 7a. After a
linear elastic region, the crushing strength is reached.
Then, a distinct load drop followed by a almost linear
decrease occurs. The complex collapse mechanisms
exhibited were often caused by the propagation of
splitting phenomena in specimens having macroscopic
fissures prior to test. The so-called “wedge split” mode
was developed in some elements, it being characterized
by a not-well defined direction of the split. In any case,
local plasticization shear bands appeared on specimen
surface, due to buckling phenomena (Fig. 7b).
40
I cycle
II cycle
III cycle
IV cycle
V cycle
VI cycle
VII cycle
VIII cycle
IX cycle
envelope
σc,0
2
[N/mm ]
30
20
10
4.2 DEFECT-FREE SPECIMENS
4.2.1 Testing equipment and set-up
The compression tests parallel and perpendicular to grain
on defect-free specimens were conducted under force
control with the universal machine Mohr & Federhaff
AG of 400 kN capacity, equipped by an upper spherical
bearing plate to improve the alignment of the element
and promote uniform stress distribution on cross-section
surfaces (Fig. 8). Deformations were measured by a
displacement transducer placed in the arms of the test
machine. The orientation of the annual growth rings with
respect to the direction of the applied load was taken into
account, so that three groups of samples were tested,
respectively, in longitudinal (L), radial (R) and
tangential (T) directions (Fig. 8).
εc,0 [%]
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
a)
40
SA-C; n=14
σc,0
2
DF-CL
[N/mm ]
30
L
DF-CR
R
DF-CT
T
average curve
20
Figure 8: Compression tests on defect-free specimens:
test arrangement and set-up.
10
εc,0 [%]
0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
b)
c)
Figure 6: Compression tests on structural elements in
actual sizes (SA-C): a) Typical σc,0-εc,0 failure cycles;
b) Envelope curves; c) Failure modes.
60
SS-C; n=20
σc,0
2
50
[N/mm ]
average curve
40
30
20
10
εc,0 [%]
0
0
0.4
0.8
1.2
1.6
2
2.4
2.8
3.2
b)
Figure 7: Compression tests on structural elements in
small sizes (SS-C): a) σc,0-εc,0 curves; b) Failure modes.
a)
4.2.2 Destructive tests
Defect-free specimens were tested up to collapse in order
to evaluate stiffness and strength properties of the clear
material. The load was applied at a constant loadinghead movement so that in longitudinal compression the
maximum force was reached between 90 and 120 sec.
(UNI ISO 3787), while, in transverse compression, it
was attained within 90±30 sec. (UNI ISO 3132).
For longitudinal tests, in Figure 9a all stress-strain
diagrams are represented, together with the average
curve, which is assumed to characterize the mechanical
behaviour in compression parallel to grain of the clear
old chestnut wood. As it can be seen, the material
exhibits moderate ductility beyond the linear response.
Moreover, at deformations larger than 2.0%, a stress
capacity reduction is generally manifested. The failure in
compression along the grain direction was progressive,
usually deemed to be an effect of shear, so-called
shearing mode. In fact, for most of the specimens,
beyond the maximum stress, structural changes began
with the formation of one or two principal gross shear
bands, which consisted in fracture cleavage
approximately perpendicular to the longitudinal axis on
the radial plane and 45° to 70° inclined as respect to the
grain orientation on the tangential plane. Other failure
modes were also manifested, such as crushing, splitting
and combined modes (Fig. 9b).
Typical stress-strain radial curves up to 30% total
deformation are highlighted in Figure 10a. Considering
that the wood microstructure in the radial direction can
be regarded as a sandwich structure, consisting of dense
latewood layers arranged in series between weak
earlywood bands, three deformation levels can be easily
identified: 1) elastic deformation, with a initial
maximum stress value and next sudden load drop due to
buckling of a portion of rays arranged in a growth ring of
the weakest earlywood layer; 2) plastic level, with a
plateau described by an irregular saw-tooth shape,
corresponding to the fracture of individual cell walls; 3)
densification and compaction of the material, with
additional failures in the same or in several other
earlywood layers, corresponding to a stress quick
increment. As a result of these structural changes, the
tested radial specimens assume a deformed shape
crushed in the transversal direction (Fig. 10b).
latewood. In fact, with increasing load a separation
between these layers was often observed (Fig. 11b).
15
DF-CT; n=22
σc,90 T
2
[N/mm ]
12
average curve
9
6
3
εc,90 T [%]
100
0
σc,0
DF-CL; n=33
average curve
0
2
2
4
6
8
10
12
14
16
18
20
[N/mm ]
80
a)
60
40
b)
20
εc,0 [%]
0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
a)
Figure 11: Compression tests on tangential defect-free
specimens (DF-CT): a) σc,90-εc,90 curves;
b) Failure modes.
4.3 RESULTS AND DISCUSSION
b)
Figure 9: Compression tests on longitudinal defect-free
specimens (DF-CL): a) σc,0-εc,0 curves; b) Failure modes.
14
DF-CR; n=22
σc,90 R
12
2
[N/mm ]
10
8
6
4
2
εc,90 R [%]
0
0
3
6
9
12
15
18
21
24
27
30
a)
b)
Figure 10: Compression tests on radial defect-free
specimens (DF-CR): a) Typical σc,90-εc,90 curves;
b) Failure modes.
All experimental curves of tested tangential specimens
and the average one are depicted in Figure 11a. As it
appears, in tangential compression the elastic phase
gradually tends to a plateau zone where the strain
increases rapidly at a constant load.
This behaviour can be explained by the early bond
failure between alternating layers of earlywood and
4.3.1 Compression parallel to grain
The mechanical characteristics of tested old chestnut
timber under compression parallel to grain is analyzed
by comparing the experimental results obtained by
destructive tests performed on the structural elements,
types SA-C and SS-C and defect free specimens, type DFCL. In Table 3 the basic statistics for both modulus of
elasticity (Ec,0) and strength (fc,0) are given, whereas the
stress-strain average curves are shown in Figure 12. It is
evident that the three groups of samples exhibit different
responses in terms of load carrying capacity, being the
stiffness properties nearly similar each other. In
particular, the presence of natural defects, wood
degradations and geometric irregularities, typical of
ancient timber members, seems to reduce of about three
times the compression strength of the base clear
material. Furthermore, while the stress-strain curve of
clear wood shows a plastic branch, highlighting a ductile
behaviour, the same curves of the structural elements
reveal a brittle performance, due to a drastic reduction of
crushing strength after the peak load.
By examining the test results, it is worth noticing that:
▪ The strength values of the specimens in actual
dimensions (SA-C) are affected by higher variability
in terms of coefficient of variation (CV);
▪ The experimental 5-percentile strength value of
specimens type SA-C and the characteristic value
assumed by UNI 11035-2 (22 N/mm2), are very
similar each other;
▪ By comparing the admissible crushing strength of
UNI 11119-III (7 N/mm2) with the 5-percentile
laboratory value, the obtained safety coefficients are
equal to about 3, 5 and 6.5 for actual (SA-C), small
(SS-C) and defect-free (DF-CL) specimens,
respectively.
Table 3: Experimental results of compression tests
2
parallel to grain [N/mm ].
80
σc
70
parallel to grain
2
[N/mm ]
perpendicular to grain (T)
60
Elasticity modulus Ec,0
SA-C
4828
5582
6739
9.84
5-perc.
average
max
CV [%]
SS-C
4619
5713
7358
11.98
DF-CL
4169
6555
9045
18.82
Strength fc,0
50
SS-C
37.67
44.05
56.46
11.98
40
SA-C
20.13
24.40
33.97
17.21
DF-CL
45.63
59.89
75.74
14.27
DF-CL; n=33
30
20
DF-CT; n=22
10
εc [%]
0
0
80
60
actual dimensions
small dimensions
defect-free
50
DF-CL; n=33
σc,0
70
2
[N/mm ]
0.5
1
1.5
2
2.5
3
3.5
4
Figure 13: Comparison between σ-ε average curves of
defect-free specimens tested in compression parallel and
perpendicular to grain.
40
30
5 CONCLUSIONS
20
SS-C; n=20
10
SA-C; n=14
εc,0 [%]
0
0
0.5
1
1.5
2
2.5
3
3.5
4
Figure 12: Comparison between σc,0-εc,0 average curves
of specimens tested in compression parallel to grain.
4.3.2 Compression perpendicular to grain
The laboratory results of transverse compression tests,
performed on both radial (DF-CR) and tangential (DFCT) defect-free specimens, are provided in Table 4 in
terms of modulus of elasticity (Ec,90) and conventional
proportional stress (fc,90).
The following observations can be drawn:
▪ Similar responses are provided by both stiffness and
strength properties in radial and tangential direction,
whereas the tangential tests are affected by higher
coefficients of variation (CV);
▪ The average modulus of elasticity provided by UNI
11035-2 (730 N/mm2) is 1.3-1.6 times the
experimental values;
▪ The 5-percentile strength results are comparable with
UNI 11035-2 value (3.8 N/mm2);
▪ The strength safety coefficient is equal to 1.2-2.0,
being the admissible value equal to 2 N/mm2,
according to UNI 11119-III.
In Figure 13 the stress-strain average curves of
longitudinal and tangential defect-free specimens are
depicted. The tested specimens show very low
compressive elastic and strength properties when loaded
in perpendicular direction, the transverse behaviour
being significantly influenced by the wood anatomy. In
fact, the fc,0/fc,90 and Ec,0/Ec,90 are equal to about 11 and
14, respectively.
Table 4: Experimental results of compression tests
2
perpendicular to grain [N/mm ].
Elasticity modulus Ec,90
5-perc.
average
max
CV [%]
DF-CR
343
545
823
21.21
DF-CT
211
463
850
34.78
Strength fc,90
DF-CR
3.99
5.33
6.69
14.60
DF-CT
3.23
5.62
11.25
41.27
In this paper experimental investigations on old chestnut
wood are illustrated. Firstly, the mechanical
identification in bending and compression parallel to
grain of elements having structural sizes and natural
defects is provided. Furthermore, compression tests on
chestnut clear wood in different orientations, such as
longitudinal, radial and tangential, are described showing
the highly orthotropic material behaviour not only for
elasticity or strength but also for the deformation
patterns. The laboratory results emphasize that the wood
mechanical characterization by means of tests on clear
small specimens can be inadequate to have reliable
information on structural timber, so much depending of
the presence of faults. Furthermore, the obtained safety
coefficients demonstrate that the structural capacity of
the material seems significantly underestimated by
means of the visual grading, according to the UNI 11119
current code.
ACKNOWLEDGEMENTS
Authors acknowledge the PRIN 2006 Italian project
which founded the research activity [1].
REFERENCES
[1] Faggiano B., Grippa M.R., Marzo A., Mazzolani
F.M.: Experimental evaluation of the mechanical
properties of wood by means of non-destructive
compared techniques for the characterization of
existing wooden structures. In Consolidation of
timber structures, ed. M. Piazza, Hevelius Publisher,
pp: 25-78, Italy, 2009 (in Italian).
[2] Faggiano B., Grippa M.R., Marzo A., Mazzolani
F.M.: Mechanical identification by NDT of old
chestnut structural timber. Proc. of the First
International Conference PROHITECH09. Rome,
Italy, 22-24 June, 2009, F.M. Mazzolani Editor,
Vol. 1, pp. 295-300.
[3] Faggiano B., Grippa M.R., Marzo A., Mazzolani
F.M.: Combined non-destructive and destructive
tests for the mechanical characterization of old
structural timber elements. Proc. of 3rd International
Conference on Advances in Experimental Structural
Engineering. San Francisco, United States, 15-16
October, 2009.