POTENTIAL OF CLT PRODUCED FROM NON-STRUCTURAL GRADE AUSTRALIAN PINUS RADIATA

Transcription

1 POTENTIAL OF CLT PRODUCED FROM NON-STRUCTURAL GRADE AUSTRALIAN PINUS RADIATA Christophe Sigrist 1, Martin Lehmann 2 ABSTRACT: In Australia CLT has a big potential but has to be imported from overseas to date for quite high prices. Milling of Pinus Radiata using optimised sawing patterns for yield and consecutive mechanical grading leads to a substantial amount of boards that cannot directly be used for structural purposes. Therefore it should be economically interesting to produce CLT using this resource. The authors performed a considerable amount of mechanical tests using various setups and optimised layups in order to investigate the mechanical properties of Pinus Radiata CLT using non-structural boards. The results showed that depending on the layup of the CLT the used resource leads to a product that performs similarly to the ones on the market in Europe. KEYWORDS: CLT, Pinus Radiata, low grade timber, material testing flat and edge, strength and stiffness properties 1 INTRODUCTION 12 Cross laminated timber (CLT) is quite established in Europe and finds many applications in housing and multistory timber construction transmitting high loads in floors and walls. In Australia CLT is only used scarcely and has to be imported to quite high prices for the time being. The Australian housing construction differs quite a lot from the European construction types and most timber houses are built from small standard sections. As all construction timber in Australia is mechanically graded quite a substantial amount of the timber is down-graded as non-structural and has to be sold at low prices. The applied sawing patterns also result in a considerable amount of wing boards presenting nearly defect free timber, boards including the pith and short boards unsuitable for standard construction purposes. Such boards and certain dimensions cannot be used for structural applications. 2 OBJECTIVES This research is to support investigations into the utilization of Australian grown, low grade radiate pine side products obtained by sawmills in order to produce CLT panels. Presently it is not possible to produce CLT in Australia. 1 Christophe Sigrist, Bern University of Applied Sciences, Architecture, Wood and Civil Engineering, Solothurnstrasse 102, CH Biel-Bienne 6, christophe.sigrist@bfh.ch 2 Martin Lehmann, Bern University of Applied Sciences, Architecture, Wood and Civil Engineering, Solothurnstrasse 102, CH Biel-Bienne 6, martin.lehmann@bfh.ch For this reason the novel product was produced at Schilliger Holz AG, Küssnacht am Rigi (Switzerland) using their facilities and production process. About 200 m 3 of radiata pine was shipped to Switzerland for this purpose. The objective was to determine the mechanical properties of CLT consisting of low grade radiate pine. The number of tests and the layups selected should be such to provide sufficient statistical data in order to get the product approved in Australia and to provide sufficient evidence to support the design of a building under AS [1]. As low end material was used the results were backed up with reference panels using machine stress graded ratiata pine which should compare to the material usually used in Europe (mostly C24 and a percentage of C16 for inner layers) to fabricate CLT. It should be economically interesting to produce CLT using this resource. This however asks for optimised layups using the given, at times extremely low mechanical properties of the available resource trying to match strength and stiffness requirements of the final product for structural applications. The objective of the presented research was to investigate in statistically sufficient quantity the very rough input material, the minimum override grading rules, the properties of used boards and the resulting properties of CLT products carrying load flat and on edge. Furthermore various layups should lead to similar strength and stiffness as standard European CLT using C24 / C16 input material. Tests in bending (4-point loading) to determine MOE and MOR were carried out according to EN 408 [7]. The requirements from EN 408 and AS 4063 [2] are identical with respect to bending. EN 408 doesn't specify a beam

2 shear test. The beam shear test from AS 4063 (testing of solid wood) however is different from the test specified in the draft code pren WI 124/128 from CEN/TC 124 (Technical Committee for timber structures) [11] where loading position and spans are different. This draft code will be considered for the shear tests as this is state of the art testing of CLT which incorporates the requirements for ETA and is in line with the CUAP (Common Understanding of Assessment Procedure) dated June 2005 [4]. For the delamination tests the draft standard [8] as well as WI 124/128 were taken as reference. All investigations related to the one component polyurethane for load bearing timber structures were coordinated with Purbond AG. 3 MATERIAL AND METHODS 3.1 INTRODUCTION Timber is a valuable resource in particular if strength graded. In Australia (and other countries) all boards from various species for construction purposes are generally machine stress graded. Board material for timber frame constructions is usually classified into MGP10, MGP12 and MGP15 stress grades with attributed strength and stiffness properties. Grading timber into two or three stress grades leads to a substantial amount of reject timber (merge) that doesn t fulfil the required stiffness criteria. Boards with insufficient length may also be attributed to merge material. In comparison to real merge material such boards may still present good strength and stiffness properties. As shown above merge material present very low stiffness properties in average. Similarly MGP12 and 15 boards have a density which is well above the density of MGP10 and merge material [3]. It is a declared goal for most sawmills to use as much of their products as possible. Engineered wood products (EWP) allow to strategically position required qualities in distinct positions. In glulam applications boards of inferior quality are placed around the neutral axes of the beam. Beam composed of up to three different layers represent the state of the art in most of the European countries. This investigation focuses of using low grade material in the middle part of cross laminated timber (CLT) and to position other non-structural material in the outer layers of the panel. The goal was to first use as much of merge material as possible, to secondly come up with suitable layups in order to fulfil strength and stiffness requirements of the panels and to finally identify possible fields of application for such novel EWP. All panels and layers were produced without edge-gluing. The Pinus Radiata CLT was produced using an industrial production facility in Switzerland. Various grading rules were applied to the non-structural and wing boards with the goal to increase the strength and to ensure a smooth production process. 3.2 TENSION PROPERTIES OF INPUT MATERI- AL Four different types of Australian Pinus Radiata boards were used to produce the CLT: 35x90 mm non-structural boards, 35x90 mm machine grade MGP12, 25x150 mm non-graded wing boards and 40x200 mm non-graded heart-in boards. Only a small quantity of MGP 12 boards was used for best properties and in order to allow a comparison of the results with CLT produced with C24 and C16 Picea Abies boards. No special sampling procedure was applied during the panel production. Figure 1: Typical stiffness distribution of machine graded board material (MGP) Sawing for maximum yield for board material also leads to products that cannot be used in normal timber frame construction. Such boards are sawn after the first chipping process and have a thickness of 20 to 25mm. As these boards are cut from the outer portion of the log they present high density and narrow growth rings which usually are nearly parallel to the width of the board. According to the sawing pattern material including the heart of the log is also obtained. Such boards are difficult to dry, have a low density and are not used for construction purposes. Figure 2: Development of coarse grading rules in order to eliminate large defects to assure minimum strength and smooth production

3 In order to identify which board type / board quality was to be attributed to the different layers first NDE measurements based on frequency and density measurements (Timbergrader, Brookhuis), ultra sonic measurements (Silvatest, CTB) and a visual assessment were carried out. Tension tests were done in order to estimate the strength and the potential of the different board types. The specimens had a length of two metres and the full cross section was tested. Based on these tests some basic grading rules were established and the attribution of the boards to inner or outer layers in the panel section was defined. The results from tension tests on ungraded material were compared to the results of boards where a visual override or a coarse visual regarding was applied. The goal was to simply eliminate very large defects as knots, holes, mechanical destruction and more. Average tensile strength [N/mm 2 ] 3.3 DEVELOPMENT OF LAYUPS Various layups for panels for floor and wall applications were developed. In order to achieve sufficient strength and stiffness 3-, 5- and 7-layer panels were considered. A step by step procedure was adopted optimizing in a first approach the outer layers regarding minimal / optimal regrade and type of board to be used. Three layups consisting of low grade timber and one layup consisting of machine stress graded material (MGP10 and MGP12) in the outer layers were produced. The input material was selected randomly, dressed and grooved if required and formed to layers according to the production procedure of Schilliger Holz AG. The boards were not graded mechanically. However extreme defects which were defined in conjunction with the manufacturer / operator in view of process or the institute for strength requirements as well as down size boards were docked or eliminated prior to manufacturing based on Schilliger's experience and their way of producing CLT with spruce. Some of the boards had to be further reduced in thickness as a good gluing quality could not be guaranteed otherwise. Such grading rules will be part of the requirements for a later production in Australia. [N/mm 2 ] Figure 3: Average tensile strength properties of merge (D,E,F) and stress graded MGP12 (C,G,H) input material: no re-grade (D,C), override (E,G) and simple re-grade (F,H) The non-structural (merge) material presented next to very low stiffness values of 7700 N/mm 2 in average for nongraded and re-graded material extremely low strength properties ranging from as little as 3.7 N/mm 2 to maximum 22.5 N/mm 2 with an average of 8.1 N/mm 2 for ungraded material. Good strength and stiffness properties were observed for MGP12 material. MGP12 pine material corresponds roughly to C24 spruce material. The bending and compression strength as well as the MOE are about 15% higher whereas the tension parallel to the grain of pine material is about 15% lower compared to European spruce. In particular for the non-structural 90/35 mm boards a minimum grading action proved to be very efficient. It could be shown that a course visual grading would improve the tensile strength of the boards by more than 40%. A more thorough re-grading of the same boards leads to an increase of the tensile strength of about 115%. The increase of strength for side boards or machine graded material was about 15% to 20%. From this investigation can be concluded that non-structural side boards (25/150 mm) as well as MGP boards may be used for outer layers. Due to low strength and stiffness properties core material and nonstructural pine material may only be used to form the inner layers of the panels. Figure 4: Visualisation of selected panel layups and dimensions of the specimens for the bending tests on flat Series of 5 panels for initial tests were industrially produced by Schilliger Holz AG using an existing vacuum press and the accredited, state of the art production process. The panel presented constant sizes of 6.0 m x 1.5 m. The longitudinal outer layers present the main load carrying direction whereas the transverse layer presented a width of 1.5 m. All boards were finger jointed. In view of the manufacturing process and the simplicity of handling the boards forming the different layers some boards were side glued. Side gluing will however not be a requirement in order to establish the properties of CLT. The appropriate adhesive and application techniques including pressure settings and general support were supplied by Purbond AG, Sempach (Switzerland). All relevant actions as moisture content of the boards, adhesive application, pressure time etc. and manufacturing data were recorded and documented. In a second series the number of tests for the most promising panel layups was increased to 10 repetitions. In order to improve the gluing quality and regarding some

4 unfavourable delamination tests results obtained, some of the wider boards were longitudinally grooved for the second series. influence of the machine operator on the positioning of the specimen the number was always located on the top for the bending tests. All panels were cut according to the same scheme and the specimen after subjected to the same tests. Figure 5: Overview of the developed and investigated panel layups Figure 6: Example of panel layup No. 2 (P02) and grading requirements for the boards The panel No.2 (P02) shown above consisted of the most abundant input material: non-structural (ns) 35mm thick boards which are not suitable for standard construction purposes. In order to achieve minimum tensile strength for boards to be placed in the outer layers some coarse visual grading had to be applied. As the inner layers are hardly loaded a rough visual override and four-side planning to calibrate the boards was sufficient to guaranty a smooth production process. 3.4 TESTING OF CLT All in all 70 specifically produced panels comprising 11 different layups were tested. The geometry of the panels was such to accommodate all necessary tests required for an eventual accreditation of the product. Matching bending specimens on flat and on edge as well as specimens for delamination and shear tests could be cut from one and the same panel. All specimens had outer boards with the grain direction running longitudinally to the main panel dimension. Specimens with outer layers running across the board length were not tested. Figure 7 shows the numbering of the specimen. xx stands for the layup number und zz for the panel number, the last two digits indicate the test procedure (01 bending edge; 02 bending-shear edge; 03 bending-shear flat; 04 bending flat; 07 delamination; 05, 06, 08 and 09 were not subjected to any tests). The number was always written on the top left corner of the specimen, therefore the exact position of the specimen in the panels was known. In order to avoid any Figure 7: Organisation and cutting scheme of test specimens within one panel The CLT specimens were tested in four point bending and shear-bending according to the draft version of pren Timber structures Cross laminated timber - Requirements. The four point bending tests proposed in this standard correspond to the EN 408 (Figure 8) requirements. The CUAP in order to possibly achieve an ETA for solid wood slab elements was also considered. The local and global bending stiffness, the bending strength and the shear strength were determined. The CLT was tested for plate and panel action as proposed in pren The specimens for plate action presented a minimum width of 300 mm or three times the board width. The specimens for panel action had a height of 300mm. F Figure 8: Test setup as used for the four point bending tests The shear-bending test proposed in pren is a four point bending with reduced distances (2.5h) between the support and the loading point. To simulate the case that the layers are produced without edge-gluing the outermost layers present a horizontal cut at the neutral axes to provoke shear failure for tests in panel action. The tests on the CLT panels were done in two series. The aim of the first series was to investigate several layups and grading regimes. At this stage eight different layups were produced and tested. In a second series the three most promising layups were produced and tested. For each layup 5 to 10 panels with the dimensions of 1.5 to 6 meters were produced. From each panel all test specimens were cut. 70 panels with eleven different layups and/or qualities of CLT were tested that adds up to 280 bending tests. The moisture content of the specimens was within the range

5 specified by pren therefore no acclimatisation was needed. All specimens were cut using the CNC facility of the panel producer. Prior to testing the specimens were stored in the lab of the BFH. (MOE of the layers perpendicular to the loading direction were set to 0 N/mm 2 ) τ n = maximal nominal shear stress calculated regarding CLT as a homogeneous material and the full cross section ΜΟΕ * = nominal modulus of elasticity calculated regarding CLT as a homogeneous material and the full cross section. Figure 9: Test setup as used for shear-bending for specimens tested in plate action and obtained failure mode pren allows two specimen geometries for the delamination tests. The specimen tested during the described work had a rectangular cross section of 75 mm by 150 mm, the thickness depended on the panel layup. The testing procedure corresponds to the method B described in the EN 391 [6] (glulam quality control). As no edge-gluing was done during production no edge-glue tests were performed. 4 RESULTS 4.1 CALCULATION OF STRESSES Various calculations methods for the stress distribution are already developed. For bending flat the methods yield similar stress distributions over the height of the section. The method applied for the evaluation of the tests is based on an elastic stress distribution over the complete section and assumes rigid bond lines. This corresponds to the method applied by Schickhofer [8] and Kreuzinger. The MOE perpendicular to the fibres is assumed to be zero which is the usual assumption for design purposes. Therefore the bending stress in these layers is also zero and the shear stress is constant over the height of the layers perpendicular to the loading direction. The MOE of the individual layers parallel to the loading directions is assumed to be equal. The following stresses have been calculated: σ m = maximal bending stress calculated with respect to the stress distribution in the CLT specimen (MOE of the layers perpendicular to the loading direction were set to 0 N/mm 2 ) σ n = maximal nominal bending stress calculated regarding CLT as a homogeneous material and the full cross section τ V = maximal shear stress calculated with respect to the stress distribution in the CLT specimen (MOE of the layers perpendicular to the loading direction were set to 0 N/mm 2 ) τ r = maximal rolling shear stress calculated with respect to the stress distribution in the CLT specimen Figure 10: Stress distribution in CLT panels (bending flat) 4.2 RESULTS FROM TESTING FLAT The tension tests on non-structural boards showed that in view of the production a minimum grading has to be applied. Such a visual override and the docking of the largest defects leads were applied to all ungraded boards. In order to reach high strength surface layers a more thorough grading was developed, indicated by ns+, 150/20+ for half of the layups investigated. However the yield in particular in the case of non-structural 90/35 boards is quite low. Figure 11: Numerical results from panel testing flat Figure 11 shows that panels consisting of ungraded, nonstructural 90/35 boards (P01) lead to very unfavourable mechanical properties of the panels. Re-grading these boards increases substantially (~40%) the mechanical properties of these panels (P02). However average bending values of 35.4 N/mm 2 and an average bending stiffness of about 8000 N/mm 2 combined with coefficients of variation of about 20% are considered too low to be structurally used for plate action. The wing boards showed very promising properties if re-graded and can be used as surface layer for all applications. The strength and stiffness proper-

6 ties of panels using such boards in the outer layers (P03) very much compare to panels where high strength stress graded MGP material (P04) is placed in the outer layers. Both strength and stiffness properties are convincing. panels consisting of MGP 12 material and re-graded nonstructural and side boards good values are achieved. Some specimens (P03 and P04) designed for bending failed in shear. Panels failing in shear however present high bending strength situated well above 40 to 50 N/mm 2 and shear strength higher than the characteristic rolling shear stress ( 1.25 N/mm 2 ). All layups failed due to rolling shear if the shear-bending setup was used. Figure 13: Series 1 rolling shear strength for testing CLT panels flat Figure 12: Maximum bending strength and MOE* (section regarded as homogeneous material) of selected specimens tested for plate action The heart-in boards present low density and poor mechanical properties and may only be used as core layers for CLT used in plate action (P08). Amongst the lowest strength and stiffness properties are obtained when using such material in the outer layers of a panel. The usage of MGP12 boards substantially increases CLT properties that meet the expectations. The highest possible results are achieved by such panels. Using strength graded material leads to very constant results: the coefficient of variation is around 6% for stress and 3% for stiffness respectively. Very similar conclusions may be drawn regarding the bending stiffness on flat. MOE * values of 9000 N/mm 2 and more are promising for panels loaded perpendicular to the surface. For the layups showing potential very small COV are obtained from testing. The stiffness of the CLT especially for three layer boards is promising. Generally the panels show good shear properties when considering plate action for tests in bending and shear the rolling shear capacity has to be considered. The three layer boards showed a clear tendency to fail due to rolling shear in the core layer even in the bending test. Here again for 4.3 RESULTS FROM TESTING ON EDGE All panel layups have also been tested on edge. The results achieved for this testing configuration show the same tendencies as indicated in Figure 12 as when testing the panels on flat. Here again the utilisation of non-structural material and heart-in material in the outer layers yield low strength, low stiffness and high coefficients of variation. The layups of the series P03, P04 and P07 indicate promising results. The importance of re-grading non-structural material and side products is again demonstrated in Figure 14. Surprisingly high strength values are obtained from 3 layer panels using side boards in the outer layers. However the COV is considerably larger than when using MGP 12 boards instead of side boards. Figure 14: Numerical results from panel testing on edge For 5-layer panels tested on edge (P07) the strength and stiffness properties are quite low due to the fact that the three interior layers are formed by non-structural 90/35 boards which haven t be re-graded. It must also be considered that the side boards only represent 32% of the total cross section. Docking some of the mayor defects to improve the strength of the inner layers would most probably increase the panel properties on one hand, on the other hand additional grading would lead to a more complex and

7 costly production processes that was to be avoided and reduce the yield. For the delamination tests according to EN four small CLT and four small glulam beams were produced. The standard requests 30 mm thick side boards. As only 25 mm thick raw sawn boards were available the specimens were produced using 20 mm thick side boards. EN requests that all layers have the same grain and year ring orientation. In addition specimens with CLT layups were also produced and tested. Pressing time and adhesive quantity were set as specified by the adhesive producer. A possible positive influence of grooves in the boards to be glued and rotary planning of the surfaces were also investigated. Figure 15: Maximum bending strength of selected specimens tested for panel action Quite low shear stresses are obtained from shear tests on edge. Due to crossing grain high shear values would be expected. The analysis of the shear failure indicates that most of the panels failed due to torsional shear between the boards as shown in Figure 16. It could be noticed that some of the boards are difficult to be glued, in particular side boards with growth rings close to parallel to the width of the boards. Additionally performed delamination tests support this hypothesis. It could be noticed that the requirements regarding delamination could not be fulfilled in many cases. Figure 16: Typical torsional shear failure - testing on edge 4.4 RESULTS FROM DELAMINATION TESTS Due to some observed shear failures it was decided to carry out shear and delamination test in order to control the quality of the surface gluing. In order to investigate if the low pressure during vacuum pressing has any influence on the delaminating behaviour of Pinus Radiata side boards, additional delaminating tests according to EN [5] were carried out. For these tests CLT and glulam specimens were produced using the recommended curing pressure (0.6 N/mm 2 ) and a significant lower pressure (0.14 N/mm 2 ). Figure 17: Definition of glulam and CLT specimens for shear and delamination tests considering various pressure conditions Good results regarding strength from shear block tests were achieved. Generally shear stresses of more than 10 N/mm 2 were achieved. In many cases low percentages of wood failures were however observed. Shear tests under an angle of 45 were also investigated but didn t lead to conclusive results. The delamination tests didn t yield the expected results and failed altogether. More work to successfully glue in particular the quite dense late wood portions of the side board must be undertaken. Also the question arises if such harsh tests are required for a product that will only be placed in favourable climate conditions 1 and 2 as specified by European standards. 4.5 POTENTIAL BOARD LAYUPS The results from this extensive investigation allow identifying suitable input material and panel layups. Varying many parameters and not having sufficient test results to statistically analyse them makes it impossible to derive characteristic values. Comparing the overall performance of some of the proposed layups with the performance of known products which are available on the European market however allows layups for 3-, 5- and 7-layer panels to be defined. The layups indicated in Figure 18 should lead to sufficient strength and stiffness properties as indicated

8 in figure 19 to be used in constructions as panels, walls or diaphragms. Figure 18: Strength and stiffness properties of selected panels tested on flat Figure 19: Potential panel layups and expected properties Figure 20: Minimum requirements for non-structural boards and other sawn side products to be used for load carrying CLT panels 5 CONCLUSIONS Panels produced from non-structural boards or materials that cannot be directly used for structural applications in principal yield very promising strength and stiffness properties in view of various structural applications. MGP boards may / should be used if high bending resistance and stiffness as for floors are requested. The results and measurements from this tests series also allow calculating the properties of the CLT in transverse direction. Furthermore loadbearing capacities of other layups may be estimated. This investigation showed that it is possible to produce CLT using Australian non-structural Pinus Radiata mainly. In order to get high performance MGP12 or wing boards have to be used for the surface layer. The nonstructural boards have to undergo a coarse grading in order to get a smooth production process and to ensure minimum quality requirements of the final product. If the bending strength is not critical (e.g. for walls) coarse graded nonstructural boards may also be used as surface layer. ACKNOWLEDGEMENT The authors acknowledge CHH Woodproducts Australia for the material supply and the financial support. The investigations have been possible thanks to the support and know-how of Schilliger Holz AG and Purbond AG. The final year students F. Egler and L. Deriaz carried out the investigations regarding assessing the input material. REFERENCES [1] AS : Timber Structures Code Part 1 Design Methods, [2] AS 4063: Timber - stress graded - in-grade strength and stiffness evaluation, [3] Boughton G. and Juniper P.: MGP Grade Design Properties, Australian Plantation Products and Paper Council, [4] Common Understanding of Assessment Procedure (CUAP), Solid wood slab element to be used as a structural element in buildings, prepared by OIB (Österreichisches Institut für Bautechnik), [5] EN 302-2: Adhesives for load-bearing timber structures, test methods, determination of resistance to delamination [6] EN 391: Glued laminated timber - Delamination test of glue lines, [7] EN 408: Timber structures - Structural timber and glued laminated timber - determination of some physical and mechanical properties EN 408:2010+A1:2012. [8] EN 14080: Timber structures - Glued laminated timber and glued solid timber - Requirements, [9] Schickhofer G., Bogensperger T. and Moosbrugger T.: BSPhandbuch, Holz-Massivbauweise in Brettsperrholz, Nachweise auf Basis des neuen europäischen Normenkonzepts, TU Graz in Kooperation mit dem Institut für Holzbau und Holztechnologie, [10] Steiger R. and Gülzow A.: Validity of bending tests on strip-shaped specimens to derive bending strength and stiffness properties of cross-laminated solid timber (CLT), Working Commission W18 - Timber Structures. [11] pren Timber structures Cross laminated timber - Requirements, 2011.