FEASIBILITY STUDY OF NEW ZEALAND RADIATA PINE CROSS-

LAMINATED TIMBER
Adrienne Fortune & Pierre Quenneville
University of Auckland, Auckland, New Zealand

Email: [email protected]

ABSTRACT
Cross-laminated timber (CLT) as a structural system is used in increasing volumes in Europe, however this is not the
case in New Zealand despite the large volume of timber available and CLT panels representing a lighter and greener
alternative to precast concrete for walls and floors. This study aimed to establish the use of CLT in New Zealand using
locally grown Radiata pine, laid crosswise into panels either three or five layers thick, 1000 mm wide x 4000 mm long and
bonded using resorcinol adhesive. Panels were subjected to bending and stiffness testing out of plane to establish yield
strength and modulus of elasticity. CLT panels were generally, but not always, stronger than their constituent boards
with bending strengths ranging from 28 to 51 MPa. MOE results from this study ranged from 7.6 to 12 MPa. Bending
strengths were predicted with various models, most accurately using those accounting for rolling shear.

INTRODUCTION available, proven in other CLT research [13, 14], cost

effective and is commonly used in glue-laminated timber
Cross-laminated timber is a structural timber product
production using Radiata pine in New Zealand (NZS 3606,
gaining popularity in Europe, but it is not yet available in
Standards New Zealand, 1993).
New Zealand.
CLT panels were subjected to four point bending tests to
New Zealand produces a large volume of timber per year
establish MOE and bending strength. Calculation methods
of which Radiata pine is the primary plantation species in
were then used to predict bending strength, based on MOE
New Zealand; around 90 % of production forests are
results from acoustic testing.
Radiata pine. The popularity of the species is largely due
to its short harvest time (25-30 years), and high timber
yields. Good sites can achieve high quality timber at
METHODOLOGY
30 m3/ha/year, or up to 50 m3/ha/year [11]. Despite the
large volume of timber available it is mainly exported in Fabrication of the CLT panels represented a significant
its raw form so CLT is a product which may offer portion of this study given that (to the best of our
significant added value. knowledge) this has not been undertaken in New Zealand
previously.
Wood in its natural state is a very strong material with a
higher strength to weight ratio than steel, concrete, and The timber used for the panels was all New Zealand,
masonry [6]. Timber is also a natural, renewable plantation-grown, Radiata pine which was kiln dried before
resource [6]. A number of countries have introduced delivery. The 90 x 41 mm timber had been machine stress
measures requiring buildings to use a percentage of graded to MSG8 [17]. The timber had been gauged to size in
renewable resources [21] so replacing construction the sawmill (as is common with framing timber). In order to
elements, such as walls and floors, often formed in provide a surface which was sufficiently smooth for
concrete, with timber, is one method of achieving this laminating, the boards were then machined on their wide
goal. faces. This process also reduced the variance in board
thickness, which further aided bonding by ensuring that the
Cross laminated timber (CLT) is a panel product
glue surface has more even. The timber was laminated
manufactured by gluing layers of parallel boards, planks
within 72 hours of the timber being machined [8] to ensure
or laths [2]. Layers of boards are laid orthogonally [7, 9,
that the moisture content at the surface of the timber was
12]. CLT has an odd number of laminates (typically 3, 5
optimal for adhesion.
or 7), so that face layers are parallel [9, 12, 19].
For this study the CLT panels were created using two
Laminated timber removes the natural variations in
thicknesses of laminates; 100 x 21 mm and 90 x 41 mm. A
timber strength, such as knots. By cross-laminating the
combination of five-layer and three-layer CLT was
boards, there is a reduction in the differential between
fabricated. These were selected as timber sizes commonly
the strength perpendicular to the grain and parallel to
available in New Zealand and both the sizes and number of
the grain [10, 16].
laminates are in line with those commonly produced in
Lamination of the panels used resorcinol formaldehyde Europe [10].
resin adhesive. This was chosen because it was readily

NEW ZEALAND TIMBER DESIGN JOURNAL VOL 19· ISSUE 3 3

Selection of the timber boards for laminating is CLT panels were subjected to four-point bending tests to
fundamental to the philosophy of creating CLT; higher failure. This was undertaken in line with the parameters
strength boards are used in the outer layers and lower given for stiffness testing, as part of machine stress
strength boards are used internally. For this study the grading AS/NZS 4063 [15] which specifies four point
boards were selected based on moisture content (any testing, as well as the minimum span and configuration
over 16% were rejected, due to expected poor glue of load points, as shown in Figure 2.
adhesion) and modulus of elasticity (MOE). Grading by
MOE was conducted using a Hitman HM200, which
measures the velocity of sound waves in the timber, then
correlated to MOE using equation 1, as per the
manufacturer’s literature [5]:

velocity2
E= (1)
density

The boards were then ranked by MOE. For each timber

size the boards were sorted into strongest (for the outer,
longitudinal laminates), weakest (for inner longitudinal Figure 2. Testing configuration for bending and MOE testing
laminates – 5 layer only) and mid-strength (for lateral (AS/NZS 4063) [15]
laminates). Boards were then cut to length.
Using this technique the bending strength of the panels
Resorcinol formaldehyde resin was chosen as the
was calculated, based on failure loads. Deflection
adhesive for the panels based on its ease of application,
logging will also allow calculation of the MOE for the
ability to cure at room temperature and widespread use
panel as a whole.
in the timber industry. The adhesive was supplied by
Hexion Adhesives, as a two part adhesive (Silvic R15 A selection of individual boards (those graded, but not
resin and Silvic RP50/RP51 hardener). These were mixed used in fabrication) were also tested for bending
in the prescribed 3:1 ratio (by weight) just prior to use, strength and MOE (using a four point bending
measured into jugs with the correct quantity to provide configuration).
one layer at 250 g/m2. Adhesive was applied evenly
All bending tests were undertaken using a 500 kN
across the surface using rollers, left for 10 minutes and
capacity MTS testing machine. Load data, as well as
then the panels were assembled into the press.
deflections at midspan and above the supports (to
Multiple layout areas were used to allow three panels to measure support compression) were recorded
be fabricated in each pressing, without exceeding the continuously throughout the testing.
45 minute working time of the adhesive. Curing of the
panels occurred within a press formed on the strong
floor of the laboratory. Pressure was applied via RESULTS
threaded rods at 400 mm centres both sides of the
During the fabrication process the boards were pushed
panel, through 105 mm thick LVL crossbeams sitting
firmly against the neighbouring boards by hand, but
above a full length 105 mm thick LVL panel. The
without clamping forces in the horizontal direction. This
threaded rods were hydraulically tensioned to achieve a
allowed the natural bow evident in particularly the 4 m
pressure of 0.7 MPa within the press [8]. This load was
lengths to form a gap to the neighbouring board. The
applied within 15 minutes of the completion of the
significance of these gaps was expected to be minimal,
gluing. Configuration of the press, with three panels
given that the boards were not expected to transfer load
during curing is shown in Figure 1.
to the neighbouring boards through direct contact.
However, there may well be some reduction in strength
owing to the decrease in timber per unit width of the face
due to the gaps. Aesthetically this is likely to be more of
a concern, as the gaps detract from the finish of the
panels, as can been seen in Figure 3.

Figure 1. Press with three CLT panels curing Figure 3. Gaps between outer boards after fabrication

4 NEW ZEALAND TIMBER DESIGN JOURNAL VOL 19· ISSUE

The gaps also resulted in the edges of the panels being graded boards are selected to have their MOE within a
ragged, in spite of the timber being accurately cut to given range. Resulting in CLT slabs with less variation
length beforehand. This was trimmed off to provide well- between the inner and outer layers, therefore
finished products, which is the most likely finish in minimising one of the benefits of CLT fabrication.
commercial fabrication.
CLT panels all demonstrated brittle mode failures, as
In this research the 90 x 41 mm boards were machine expected with timber. Some of the panels experienced
stress graded to AS/NZS 1748 [17] prior to supply and so obvious rolling shear failures, with longitudinal cracks
the strength of the boards was greater and less variable. forming within the cross-wise laminate under tension.
The timber supplied was certified as MSG8. The 100 x These were coupled with tensile failures in the outer
21 mm boards were supplied ungraded resulting in a laminates. Other panels demonstrated tensile failures in
greater variation and included more weaker boards, as the outer laminates which then propagated through the
shown in Table 1. cross-wise layer between the panels, causing failure of
the upper layers in bending.
Table 1. MOE of individual boards

Boards Mean (GPa) Standard Deviation

ANALYSIS
90 x 41 mm 9.60 1.56
Bending strength and moment capacity of the CLT
100 x 21 mm 10.6 2.94 panels was predicted from the MOE values of the boards
obtained by acoustic testing. These values were
The standard deviation of the MOE was roughly twice for correlated to bending strength of the boards using
the ungraded boards compared with the graded boards, methods by Walford [20] and then used to predict the
which means that the CLT panels fabricated from moment capacity for the section.
ungraded boards will have a greater variation in strength
Multiple models were utilised in this study to assist with
between the inner laminates and outer laminates. This
the prediction of the moment capacity for the CLT
variance can be seen in the difference between the MOE
panels. Simple models were initially tested, with the
values shown in Figure 4 and Figure 5.
panel assumed to be one solid isotropic cross-section,
The ungraded panels showed higher peak values as the similar to sawn timber sections. This method grossly

Figure 4. Selection of boards for 100 x 21 mm panels

Figure 5. Selection of boards for 90 x 41 mm panels

NEW ZEALAND TIMBER DESIGN JOURNAL VOL 19· ISSUE 3 5

over-predicted the moment capacity of the section Table 2. MOE from sonic testing data correlated to bending
strength using Walford’s correlation [20]
because it ignored the reduced capacity of the layers
loaded perpendicular to the grain. Transformed section MOE by sonic Fb by
Board Board
models [1], I-joist models [4] and plywood models [18] testing correlation
dimension Grade
were more accurate than prior models. However, a final (GPa) (MPa)
model using the methods by Blass and Fellmoser [3], 100 x 21 Strong 14.4 63.0
given in equations 2-4, combined with the bending
strength prediction correlation from Walford [20], given 100 x 21 Medium 10.0 40.7
in Table 2, was most successful. 100 x 21 Weak 5.58 18.4
Configuration factors for loadings parallel and 90 x 41 Strong 10.9 45.3
perpendicular to the plane of the panel and to the grain
was proposed by Blass and Fellmoser [3] allowing for a 90 x 41 Medium 8.50 33.1
reduction in bending strength of the laminates. The 90 x 41 Weak 7.06 25.8
bending tests on the CLT panels in this study were
loaded perpendicular to the plane of the panel and bent
parallel to the grain of the face laminates. For this DISCUSSION
configuration the composition factor, K1 is given by New Zealand grown Radiata pine was used successfully
equation 2: to fabricate CLT in this study. Boards were oriented
⎛ E ⎞ a 3 − a m3 − 4 + ... ± a13 longitudinally on the outer layers and cross-wise at
k1 = 1 − ⎜⎜1 − 90 ⎟⎟ m − 2 (2) alternate layers. Panels were created with three and five
⎝ E0 ⎠ a m3 layers, using 100 x 21 mm and 90 x 41 mm boards.
Where m is the number of layers in panel Locally sourced resorcinol formaldehyde resin adhesive
am is the thickness of m layers in mm (SYLVIC R15 resin, L5 hardener) was used to bond the
E0 is the MOE parallel to the grain in MPa (by panels, spread at 250 g/m2 and performed well.
sonic testing) The CLT panels in this study were tested to destruction
E90 is the MOE at 90˚ to the grain in MPa using a four-point bending configuration so as to achieve
For this calculation E90 was taken as being E0/20, given pure bending in the central third span. Failures of all six
by Buchanan [4]. In this method the ratio of E0/E90 is panels occurred on the tension face, as is typical for
suggested as being 30. It was considered that the figure timber and large planes of rolling shear failure were
of 20 was most appropriate for Radiata pine, whereas observed.
the 30 relates to research undertaken on spruce. Multiple models were utilised in this study to assist with
Bending strength of the panels was then calculated the prediction of the moment capacity for the CLT
using equation 3. panels. Simple models were initially tested, with the
panel assumed to be one solid isotropic cross-section,
f b , 0 , eff = f b , 0 k 1 (3) similar to sawn timber sections. This method grossly
over-predicted the moment capacity of the section
where fb,0,eff is the effective bending strength of the because it ignored the reduced capacity of the layers
timber parallel to grain in MPa loaded perpendicular to the grain. Transformed section
fb,0 is the bending strength of the timber parallel models [1], I-joist models [4] and plywood models [18]
to the grain in MPa, correlated from the sonic were more accurate than prior models. However, a final
testing results. model using the methods by Blass and Fellmoser [3]
The moment capacity of the section can then be combined with the bending strength prediction
calculated as previous, using equation 4. correlation from Walford [20] was most successful. This
gave mean bending strengths for the 100 x 21 mm five
M = f b , 0 , eff Z eff (4) layer panels of 48.1 MPa, for the 90 x 41 mm three layer
panels of 36.1 MPa and for the 90 x 41 mm five layer
Bending strength of the boards was predicted using the panels of 27.6 MPa. The mean MOE for these panels was
correlation by Walford [20], given in equation 5. 11.3 GPa for the 100 x 21 mm five layer panels, 10.1 GPa
for the 90 x 41 mm three layer panels and 7.63 GPa for
f b = 5 . 0583 E − 9 . 8669 (5)
the 90 x 41 mm five layer panel.
This gave the bending strength results shown in Table 2. These values of bending strength were found to be
This gave mean bending strengths for the 100 x 21 mm similar to those recorded by Vessby et al., [19] who found
five layer panels of 48.1 MPa, for the 90 x 41 mm three bending strengths between 39.5 MPa and 51.1 MPa for
layer panels of 36.1 MPa and for the 90 x 41 mm five spruce CLT. This study also found values in excess of the
layer panels of 27.6 MPa. The mean MOE for these 10 MPa allowable stress given by KLH for their spruce
panels was 11.3 GPa for the 100 x 21 mm five layer CLT, although the sample size was not large enough to
panels, 10.1 GPa for the 90 x 41 mm three layer panels be able to predict the characteristic strength for the
and 7.63 GPa for the 90 x 41 mm five layer panel. panels.

6 NEW ZEALAND TIMBER DESIGN JOURNAL VOL 19· ISSUE

CONCLUSION [9] Jobstl, R.A., Bogensperger, T. and Schickhofer, G.
(2008) In-plane shear strength of Cross Laminated
This study demonstrates that it is possible to construct
Timber. (St. Andrews, Canada), August 2008. pp.
CLT using New Zealand grown Radiata pine.
CIB-W18/41-21-3.
Brittle mode failures were found, as is typical for timber.
[10] KLH (2001) KLH Design Guide. KLH Massivholz
Failures typically occurred on the tension edge of the
GmbH, Austria.
slab. Additionally there were planes of rolling shear
within the cross-wise layers. [11] Maclaren, J.P. (1993) Radiata Pine Growers'
Manual. New Zealand Forest Research Institute,
Prediction methodologies evaluated found that those
Rotorua, New Zealand.
which incorporated rolling shear were more accurate,
particularly that by Blass & Fellmoser [3]. Correlation of [12] Mestek, P., Kreuzinger, H. and Winter, S. (2008)
MOE values obtained by sonic testing were correlated to Design of Cross Laminated Timber (CLT). WCTE
bending strengths following methods by Walford [20]. 2008 - 10th World Conference on Timber
Engineering (Miyazaki, Japan), June 2-5, 2008. pp.

ACKNOWLEDGEMENTS [13] Park, H., Fushitami, M., Sato, K., Kubo, T. and
Byeon, H. (2006) Bending creep performances of
The authors would like to thank The University of three-ply cross-laminated woods made with five
Auckland, RedStag Timber and Momentive Specialty species. Journal of Wood Science, 52, 220-229.
Chemicals (NZ) for their support in this research.
[14] Park, H., Fushitani, M., Sato, K., Kubo, T. and
Byeon, H. (2003) Static bending strength
REFERENCES performances of cross-laminated woods made
with five species. Journal of Wood Science, 49,
[1] Beer, F.P. and Johnston, J., R. (1992) Mechanics of 411-417.
Materials - Second Edition in SI Units. McGraw-Hill
Book Company, London. [15] Standards New Zealand (1992) AS/NZS 4063:
Timber - Stress graded - in-grade strength and
[2] Bejtka, I. and Lam, F. (2008) Cross laminated stiffness evaluation. Standards New Zealand,
timber as innovative building material. Canadian Wellington.
Society for Civil Engineering 2008 Annual
Conference (Quebec, QC), June 10-13, 2008. pp. [16] Standards New Zealand (1993) NZS 3603: Timber
1090-1099. Structures Standard. Standards New Zealand,
Wellington.
[3] Blass, H. and Fellmoser, P. (2004) Design of solid
wood panels with cross layers. Proceedings of the [17] Standards New Zealand (2006) AS/NZS 1748:
8th World Conference on Timber Engineering Timber - Mecahnically stress-graded for
(Lahti, Finland), pp. 1001-1006. structural purposes. Standards New Zealand,
Wellington.
[4] Buchanan, A. (2007) Timber Design Guide (Third
Edition). New Zealand Timber Industry Federation [18] Standards New Zealand (2008) AS/NZS 2269.1
Inc., Wellington. Plywood—Structural, Part 1: Determination of
structural properties—Test methods. Standards
[5] Carter, P. and Sharplin, N. (2007) HITMAN Sonic New Zealand, Wellington.
Technology for Improved Profitability -
Presentation for North America. Fibre-gen, [19] Vessby, J., Enquist, B., Petersson, H. and
Auckland. http://www.fibre-gen.com/pdf/The% Alsmarker, T. (2009) Experimental study of cross-
20HITMAN%20System%20explained.pdf Accessed laminated timber wall panels. European Journal of
on 12 February 2010. Wood Products, 67, 211-218.

[6] Gagnon, S., Hu, L. and Desjardins, R. (2008) [20] Walford, G.B. (2010) Unpublished data for
Construction solutions for wood-based floors in correlations between strength parameters and
hybrid building systems. (St. Andrew, Canada), sonic testing on Radiata Pine. Scion, Rotorua, New
June 10-13, 2008. pp. 1-10. Zealand.

[7] Guggenberger, W. and Moosbrugger, T. (2006) [21] Yates, M., Linegar, M. and Dujic, B. (2008) Design
Mechanics of Cross-Laminated Timber Plates of an 8 storey Residential Tower from KLH Cross
under Uniaxial Bending. WCTE 2006 - 9th World Laminated Solid Timber Panels. 10th World
Conference on Timber Engineering (Portland, Conference on Timber Engineering (Miyazaki,
Oregon, USA), August 6-10, 2006. pp. Japan), June 2-5. pp. 1-8.

[8] Hexion Specialty Chemicals (2007) SYLVIC R15 /

SYLVIC L5 Technical Data Sheet. Accessed on 13
April 2010.

NEW ZEALAND TIMBER DESIGN JOURNAL VOL 19· ISSUE 3 7