Structural Composite from FSC Certified Hybrid-Poplar

Transcription

1 Structural Composite from FSC Certified Hybrid-Poplar Frederick A. Kamke Dept. Wood Science & Engineering, Oregon State University Corvallis, Oregon USA Andreja Kutnar Primorska Inst. for Natural Sciences and Technology, University of Primorska Koper, Slovenia Abstract Clones of species from the Populus genus are cultivated in many countries, including the US. Historically, poplar clones have been grown as a raw material for pulp, although in recent years this resource has also been used for biofuels, oriented strand board, decorative millwork, and some specialty applications where low density and uniform grain pattern are desirable. Hybrid poplar grows quickly in temperate regions with growth rate up to 35 m 3 ha -1 y -1 after eight to 12 year rotations. Density is typically 300 to 375 kg m -3 and modulus of elasticity 7 to 8 GPa. Although hybrid poplar has been used for core stock in structural plywood, it lacks the mechanical properties necessary for building construction applications. Our research applied a hygro-thermal densification process to enhance the mechanical properties of hybrid poplar. The process is called viscoelastic thermal compression (VTC) and is specifically designed for processing thin wood laminates to be used in structural composites. Hybrid poplar (Populus deltoides x Populus trichocarpa) was obtained from a plantation certified by the Forest Stewardship Council in Northeast Oregon (GreenWood Resources, Boardman Oregon). The logs were 11 years old with a diameter of 25 to 30 cm. Hybrid poplar boards were processed to produce VTC wood with density of approximately 750 to 1200 kg m -3 and modulus of elasticity of 18 to 29 GPa. The VTC laminas were then used to manufacture a parallellaminated composite using VTC wood in the outer layers and untreated hybrid poplar in the core. The bending properties of the composites exceeded the structural design values required for loadbearing applications in wood frame construction. In addition, fracture toughness testing revealed good adhesive bond performance using a conventional phenol-formaldehyde adhesive. Keywords densification, hygro-thermal compression, composite. Paper WS-36 1 of 9

2 Introduction Clones of species from the Populus genus are cultivated in many countries, including the US. Historically, poplar clones have been grown as a raw material for pulp, although in recent years this resource has also been used for biofuels, oriented strand board, decorative millwork, and some specialty applications where low density and uniform grain pattern are desirable. Hybrid poplar grows quickly in temperate regions with growth rate up to 35 m 3 ha -1 y -1 after eight to 12 year rotations. Density is typically 300 to 375 kg m -3 and modulus of elasticity 7 to 8 GPa. Some hybrid poplar plantations have been certified by Forest Sustainability Council (FSC), which imposes strict standards on social, economic, ecological, cultural, and spiritual aspects of forest plantation management (Anon 2010). Although hybrid poplar has been used for core stock in structural plywood, it lacks the mechanical properties necessary for building construction applications. Our research applied a hygro-thermal densification process to enhance the mechanical properties of hybrid poplar. The process is called viscoelastic thermal compression (VTC) and is specifically designed for processing thin wood laminates to be used in structural composites. Results from previous studies for bonding simple laminated composites from VTC wood, and fracture testing of adhesive bonds, are reported. Materials and Methods Logs from low-density hybrid poplar (Populus deltoides Populus trichocarpa) were obtained from a FSC certified plantation located in northeastern Oregon USA. The logs were approximately 25 cm diameter and were from 11 year old trees. The wood was sawn into lumber (approximately 25 mm thick), air-dried, and then placed in an environment controlled room (20 C, 65% relative humidity) until equilibrium moisture content (approximately 12 %) was achieved. Boards were then planed into specimens of various thickness, ranging from 4 to 6 mm, depending on the degree of densification desired. Specimens had initial average oven dry density of approximately 0.38 g cm -3. Compression Procedure Specimens were compressed perpendicular to grain to achieve the desired degree of densification. The target compressed thickness was the same for all specimens. However, the initial thickness was different. Therefore, specimens with greater initial thickness were compressed with a greater degree of densification. Some specimens were used for determining the change of density and bending modulus before and after treatment. Other specimens were used to manufacture threelayer composites and evaluate adhesive bond integrity. Transverse compression of wood under saturated steam conditions at 170 C was performed in a custom-built pressurized vessel that was equipped with a heated hydraulic press. The cylindrical vessel was heated by an electrical jacket heater. Platen temperature was independently controlled using two electrical resistance heater rods in each platen. Steam was supplied by a dedicated boiler. Thermocouples monitored platen temperature and steam temperature. Steam pressure was monitored with a pressure transducer as well as a mechanical gauge. Temperature and steam pressure was controlled +/- 3 o C and +/ MPa, respectively. Force on the specimen was Paper WS-36 2 of 9

3 measured by a load cell (Omegadyne Inc, Sunbury, Ohio, Model LC412-50K) that was located outside of the pressurized vessel. Force on the specimen was created using a hydraulic cylinder mounted outside of the vessel. Force was transferred from the hydraulic cylinder to the platens (18 cm x 18 cm) using a stainless-steel compression cylinder that was sealed inside of a compressible bellows. The load cell was calibrated to account for steam pressure and mechanical resistance of the bellows. During compression, deformation was measured with a linear variable differential transformer (LVDT) connected to the heated platens of the press. The LVDT (Macro Sensors, Model HSTAR , +/ mm) was calibrated for temperature and steam pressure within the range of conditions used for this experiment. Inside of the sealed vessel, the uncompressed specimens were first exposed to saturated steam for 3 minutes, followed by compression at 1.4 MPa for 3 minutes, then release of steam pressure, then release of compression force for 30 seconds, followed by application of compression force to reach target thickness, hold compressed thickness for 3 minutes, cool platens to 100 o C, and finally release of compression force and specimen removal. The compression stress required to reach the target thickness depended on the degree of densification. Thickness control was achieved using metal calibration bars. Cooling was accomplished by directing water through channels built into the platens. Cooling required about 90 seconds. Ten replications were used for each treatment. Bending and Density Tests of Lamina Before and after treatment the specimen weight and dimensions were measured. After treatment specimens were dried in a convection oven (103 C) over night, oven dry weight and dimensions were measured, and the mass loss, oven dry density and moisture content after compression were determined. Then the compressed specimens were placed into a controlled environment with 65% relative humidity at 20 C until constant moisture content was achieved. After conditioning, the specimen moisture content was again determined and the density, based on oven-dry weight, was calculated. Simple three-point bending tests were performed on specimens before and after treatment. All specimens were conditioned at 20 o C and 65% relative humidity before bending tests. Treated specimens had lower equilibrium moisture content (EMC), which was approximately 7%, while untreated specimens had EMC of 12%. The difference in MC contributed to some difference in measured bending modulus. Preparation of Bonded Specimens Two types of laminated composites were prepared one for evaluation of adhesive bond performance and another for demonstration of bending strength performance. Over-notch flexure (ONF) test specimens were selected to evaluate adhesive bonding in mode II fracture (shear). It is difficult to induce failure by shear when conducting bending tests using specimens with a rectangular cross section, since the tensile or compression strength of outer lamina can be reached before the ultimate shear strength of the adhesive. Therefore, deep composite beams were prepared as reported previously (Kutnar et al 2008a). The specimens were bonded with liquid PF adhesive supplied by Georgia Pacific Resins, Inc with viscosity 330 mpa Paper WS-36 3 of 9

4 s. This is a commercial product designed for the face layer of oriented strand board. Each bonded surface had adhesive coverage of 155 g m -2. Six-layer fracture test specimens (Fig. 1) were bonded in a hot-press at 150 C using 700 kpa of pressure for 30 minutes. ONF fracture test specimens had the bondline of interest located between the two central layers (Fig. 1). When evaluating VTC wood bonding, the two central layers were VTC wood of the same degree of densification. The VTC wood layers were backed by untreated wood. The outer layers were again VTC wood of the same degree of densification as the central layers. Thus, these beams were symmetric and the sublaminates in arm 1 and 2 of the beam were symmetric as well. In the case of testing bond performance of the untreated wood, central layers were from the untreated wood, whereas the outer layers were made of the VTC wood with 132 % degree of densification. This produced unsymmetrical arm 1 and arm 2 in specimens for testing the PF adhesive bond performance of the untreated wood, which was considered in the calculations of fracture toughness G II (Kutnar et al 2008a). The pre-crack of ONF specimens was formed during bonding by inserting a teflon film into the first 55 mm of the interface to prevent bonding. After hot-pressing, the beams were cooled to room temperature and cut into two specimens 4 mm wide, 22 mm thick and 145 mm long. Prior to fracture testing, the specimens were conditioned in a controlled environment room at relative humidity of 65% and 20 C until a constant weight was obtained. Figure 1. Over-notch flexure specimen and bending test configuration; side view showing location of supports (bottom) and load point (top). Pre-crack notch at center left. Simple three-layer composites were prepared using various configurations of VTC wood and untreated wood lamina (Kutnar et al 2008c). The outer laminas were VTC wood with degree of densification of 63%, 98%, or 132%. The center lamina was untreated wood with thickness 6 mm. Bond preparation was the same as described for the ONF specimens. However, hotpressing was done for 6 minutes. These specimens were used for four-point bending tests to failure. The test span was 120 mm and the loading rate was 5 mm/min. ONF Fracture Test Procedure Paper WS-36 4 of 9

5 Fracture testing was performed using a standard three-point flexure fixture and a universal testing machine, however, the load point was off-center as shown in Figure 1. The specimens were loaded with a crosshead speed of 1 mm/min over a full span length 120 mm. The position of the external load for the ONF test was 40 mm (Fig. 1). The specimens were positioned in the fixture to achieve an initial crack length of 45 mm. The interlaminar fracture toughness G II was determined at each 2 mm increment of the crack as it propagated along the length of the bondline. Digital Image Correlation (DIC) was used to measure crack length. DIC is an optical method to measure two-dimensional deformation on an object surface. To create a distinct pattern for DIC on the specimen surface, the specimens were coated with white paint followed by a random speckled pattern of black paint (Kutnar et al 2008a). Results and Discussion Change of density as a result of VTC processing is shown in Table 1 in comparison to untreated specimens. Although VTC wood density of up to 1.4 g cm -3 can be achieved, the processing conditions used yielded maximum density of approximately 1.2 g cm -3, which was over 200% greater than the initial density. Moisture content (MC) immediately after VTC processing was different depending on the degree of densification. The higher density specimens lost less water during processing. After conditioning at 20 o C and 65% relative humidity, MC of all VTC wood was approximately the same at 7% (+/- 0.5%), whereas EMC of untreated wood was 12%. Table 1. Specific gravity before and after VTC processing for each treatment. Initial moisture content was 12%. Initial thickness (mm) Initial specific gravity Target specific gravity after VTC Actual specific gravity after VTC MC after VTC (%) Degree of densification (%) n.a. n.a. n.a Bending test results for the individual lamina are shown in Figure 2. Note that the relationship between MOE or MOR with specific gravity is nearly linear. Change of MOE and MOR were proportional to the degree of densification within the range of conditions studied. Bending test results for the three-layer composites are shown in Tables 2 and 3. MOE increased with the use of higher density VTC wood up to 98% densification, then no further increase was detected. There was no difference of MOR between any of the VTC composites. The untreated composite had significantly lower MOR than any of the VTC composites. The failure patterns were erratic. However, the VTC composites tended to fail with crack propagation along the interface between the lower tension layer and the center layer (Kutnar et al 2008c). Paper WS-36 5 of 9

6 Figure 2. Modulus of rupture and modulus of elasticity for untreated wood and VTC wood processed to 4 levels of densification. Linear regression line equations are shown as a function of specific gravity. Table 2. The MOE of 3-layer composites corresponding to the composite type and multiple range test with 95% LSD. Composite configuration Count Mean MOE [GPa] Standard Deviation Significant Groups Density [g cm -3 ] a b c c 0.54 MOE values of three-layer VTC composites are similar to those corresponding to commercially produced structural composite lumber products. Janowiak and Bukowski (2000) compared the MOE of solid wood yellow-poplar (Liriodendron tulipifera) with several composites made from yellow-poplar (Table 4). While yellow-poplar is a different genus than true poplars, they are both diffuse-porous hardwoods with similar anatomy. Yellow-poplar has greater density and is used commercially to produce structural lumber composites. The production of laminated veneer Paper WS-36 6 of 9

7 lumber (LVL) and parallel strand lumber (PSL) increased the MOE by 26%, whereas laminated strand lumber (LSL) was about the same when compared to solid wood. On the other hand, application of VTC wood in the face layers in the 3-layer VTC composites increased the MOE, when compared to the three-layer untreated composites (0-0-0). In the VTC composites, VTC wood increased the MOE by 25%; by 44%, and composites by 48%. Note the density of the commercial structural lumber composites is greater than any of the VTC composites, except the VTC composite. This comparison demonstrates that the VTC process offers the possibility of manufacturing structural wood composites from non-structural timber species. Table 3. The MOR of 3-layer composites corresponding to the composite type and multiple range test with 95% LSD. Composite configuration Count Mean MOR [MPa] Standard Deviation Significant Groups Density [g cm -3 ] a b b b 0.54 Table 4: Average MOE of different wood-based composites according to composite type and density (Janowiak and Bukowski, 2000). Composite Density 1 [g cm -3 ] MOE [GPa] Solid wood, yellow poplar LVL PSL LSL Based on oven-dry weight and volume at 12% MC. Fracture toughness results are shown in Table 5, along with p-values for Least Significant Difference (LSD) analysis. Increasing the degree of densification improved fracture toughness. The VTC process did not diminish the ability to create an adhesive bond using a conventional PF adhesive system. Observations of the adhesive bondline by fluorescence microscopy revealed very little penetration of adhesive into VTC wood, with decreasing penetration with higher degree of densification. Untreated hybrid poplar had deep penetration. Complete wood failure occurred on all ONF specimens, indicating no failure of adhesive-wood interface. One might speculate that VTC wood exhibits stiffness more similar to pure PF resin, and therefore, less stress concentration occurs even if adhesive penetration is low. Paper WS-36 7 of 9

8 Degree of densification Table 5. Average values of G II at different crack lengths. Average G II [N/mm] 45 mm 50 mm 52 mm Average Average p-value G II p-value G II [N/mm] [N/mm] p-value 0% (untreated) % % % Conclusion VTC wood may be adequately bonded using conventional phenol-formaldehyde adhesive to itself or to untreated wood. Dramatic increase of bending properties have been demonstrated, with increase of over 200% for both MOE and MOR when compared to untreated hybrid poplar wood. When used to manufacture a simple three-layer composite, the overall density of the VTC composite is similar to commercial structural composite lumber products. These results were achieved using FSC certified hybrid poplar that otherwise does not have mechanical properties suitable for structural applications. The results presented here are based on small laboratory specimens. The same results may not be reproducible on a commercial scale. However, laboratory studies support the potential for commercial application of VTC technology for use in structural composites. References Anon Forest Stewardship Council. accessed July 20, Janowiak J.J. and S.W. Bukowski Toughness properties for several composite lumber materials. Forest Prod J 50(5): Kamke, F.A Densified radiata pine for structural composites. MADERAS: Ciencia y Tecnología' journal (Wood: Science and Technology), 8(2): Kutnar, A.; F.A. Kamke; J.A. Nairn and M. Sernek. 2008a. Mode II fracture behavior of bonded viscoelastic thermal compressed wood. Wood and Fiber Sci. 40(3): Kutnar A., F.A. Kamke, M. Petrič, and M. Sernek. 2008b. The influence of viscoelastic thermal compression on the chemistry and surface energetics of wood. Colloids Surf. A: Physicochem. Eng. Aspects, 329: Paper WS-36 8 of 9

9 Kutnar, A.; F.A. Kamke; and M. Sernek. 2008c. The mechanical properties of densified VTC wood relevant for structural composites. Holz als Roh und Werkstoff 66(6): Kutnar, A.; F.A. Kamke; and M. Sernek Utilization of plantation wood in high performance wood-based structural composites. In: Proc. Wood Adhesives 2009, South Tahoe, Nevada, Sep , 2009, in press. Sernek M.; Resnick J; and F.A. Kamke Penetration of liquid urea-formaldehyde adhesive into beech wood. Wood Fiber Sci. 31(1): Paper WS-36 9 of 9