A New Concept of Bio-based Prestress Technology with Experimental Proof-of-Concept on Bamboo- Timber Composite Beams

This paper presents a pioneering experimental proof-of-concept study to validate a novel concept of prestress technology that used only pure bio-based composite materials while achieved consistent prestressed stress distribution within the structure member, and provided in-situ flexibility, improved structural performance, and maximised the rate of utilisation of each material. Industrial level of facilities were used during this development. The prestress is achieved by pressurised/forced lamination of multiple components with different materials and geometrical properties. The prestressing process is activated during the pressure release stage during which the components are interacting with each other, creating different stress statuses that would favour the weaker and adverse the stronger components to maximise the strength exploitation of different materials. Using laminated bamboo and timber as an example * Corresponding author E-mail address: yudenggxust@gmail.com (Yu Deng); Minhe Shen: PhD candidate at Edinburg Napier University, a joint supervision PhD student with Guangxi University of Science and Technology. Page | 2 pair, twenty-two glulam, non-prestressed and prestressed laminated bamboo-timber composite beams were manufactured, tested, and analysed to provide an in-depth understanding of the structural behaviours of these novel structural members. Failure modes, yielding, ultimate and serviceability limit loads, and corresponding deflections, as well as the histories of strain development at key positions of the specimens were examined. The experimental study confirmed the feasibility, effectiveness and industrial scalability of the proposed technology. The novel concept provides a new approach for developing the prestress technology for bio-based materials, and this experimental study laid the foundation for its future analytical development and numerical studies.


Introduction
Prestress technology has been widely used in reinforced concrete structures [1][2][3][4][5][6][7][8][9][10][11] to reduce deflection, increase structural stiffness, exploit the materials' mechanical strength, and ultimately, to create a lighter structural system to span longer distance. It is always the timber engineers' dream to develop an effective, practical, pure bio-based prestressed technology with in-situ flexibility that allows members to be cut and adjusted in-situ. The most common approach currently followed by timber engineers and researchers is still more and less the same conventional technique for reinforced concrete, i.e., prestressed by high-stiffness/strength prestress tendons with spreading end plates as the anchors [12][13][14]. The difference in stiffness and strength between the prestress tendons and wood or bio-based materials has put these conventional approach followers in a dilemma, whereby adopting a low prestress level will limit the effectiveness of the prestress technology while switching to a high prestress level will result in high time-dependant relaxation. There is a need to find a new route to prestress bio-based materials in construction. Despite several interesting research attempts [15][16][17][18][19], the prestress technology for pure bio-based material is still far from mature and very few practical projects have adopted this technology.
There is a need to attempt new concepts when developing prestress technology for pure bio-based construction materials. Ideally, the prestress process can be incorporated into the fabrication and has the in-situ flexibility that allows the structural member to be cut short or adjusted as with normal timber beams.
For multi-material/species bio-based prestressed glulam beams, it is not necessary to apply the prestressed forces/stresses through the pre-/post-prestressed tendons. These prestressed forces/stresses can be applied during the glue-laminating process by pressing and gluing multiple parts with different curvatures or lengths together in a certain sequence. After that, when the pressure from the press machine is released, the parts in the glulam beam will interact with each other and redistribute the Initial stresses from one component to another. If the curvatures/lengths are carefully designed and assembled, part of these stresses in the weaker component(s) shall counteract with the stresses caused by the external loads to prevent the weaker parts from failing prematurely, before the strong component(s) exhaust its/their capacity, thus enhancing the overall strength of the composite beam. Although the stronger component(s) will take a greater share of the effects caused by the external loads after the prestressing (stress redistribution) process, their capacity/strength enable them to do so. This intrinsic capacity sharing mechanism maximises the overall utilisation of the material strength in different components.
Glulam beam is one of the best options to trial this new concept. The glulam [20][21][22][23][24] is commonly used in modern timber construction with excellent structural performance as well as thermal insulation, attractive textures, and very economic, sustainable materials. Manufacturing multi-material/species glulam beams with components that have different strength grades is an effective way to produce low cost, high strength structural members with optimised material exploitation. A good option is to combine the high strength engineered bamboo [25][26][27] and softwood timber to produce a sandwich glulam bamboo-timber beam [28][29][30][31][32][33]. The difference in mechanical properties and strength between engineering bamboo and softwood makes them a perfect pair to examine this new concept of bio-based prestress technology.
In the past decades, using pre-/post-tension cables and glued on or stress laminated steel or carbon fibre plates for strengthening structural-use timber or engineered bamboo members are among those most common approaches to strengthen these types of structural members [34][35][36][37]. Guo, et al [38,39] conducted a study on the short-term flexural behaviour for glulam bamboo beams with externally exposed prestress tendons under different pre-stressed states. The study revealed that flexural strength increases when the level of prestressing is increased, while the flexural stiffness stays relatively stable. Their study also concluded several disadvantages including complicated fabrication process, ineffective prestress effect and extra headroom required under the beam for the prestress tendon, which has prevented it from becoming widely used in practical engineering applications. Fibre reinforcement polymer (FRP), underpinned by its strength, lightweight density, and corrosion resistance, is an ideal material for strengthening bamboo and timber structural members [40][41][42][43][44][45]. However, its obsessively high (compared to bamboo and timber) elastic modulus could easily lead to high prestress loss due to creep and other fabrication limits such as anchor detailing, etc. Apart from the above attempts following the conventional approach, there are a few recent new studies that focused on developing alternative, pure bio-based prestress technology. By using moisture induced expansion of a series of compression wood inserts preinstalled in the pre-cut slots or holes to generate the prestressed (or pre-camber) conditions [15][16][17][18][19], these studies are the first laboratory attempts to develop pure bio-based prestress technology for timber beams. However, the uncertainty of moisture level during the regular service of a building and the damage to the structural integrity for accommodating the wood inserts are among the key obstacles preventing the application of this technology in practical engineering. This paper proposes an innovative pure bio-based prestress technology enlightened by the mouldability of bamboo and the customisability of the laminated bamboo production process. The key concept of this novel prestress technology is to actively utilise the post lamination interactions among the layered components with different geometrical parameters, for example in curvature or length, to prestress a certain or all layers through the internal stress redistribution that activates during the pressure release step in a forced lamination production cycle. The prestressed (redistributed) stresses in different layers could either counteract or append to the stresses caused by the external loads. An optimal design of these prestressed structural members would be the ability to achieve the following: 1) the weaker layers (such as softwood) will have the prestressed stresses that will counteract with the stresses caused by the external loads so that these layers will not fail prematurely; 2) while the stronger layers (such as engineered bamboo or hardwood layers) have the prestressed stresses that will add on to the stresses caused by the external loads, they are stronger layers and have the capacity to take a greater share of the loading. In this study, only one type of engineered bamboo, laminated bamboo, is used. Taking prestressed laminated bambootimber sandwich beam as an example to illustrate this process, firstly, the thin single layer of laminated bamboo will be glue-laminated in a curved arc-shape mould to form the curved lumbers. Then two layers of curved laminated bamboo lumbers will be sandwiched and glued with one straight softwood glulam beam as the central core. The top and bottom laminated bamboo layers should be curved in same direction with their arc centres positioned under the lumber. This three-part components will be pressed and glued together to form a straight sandwich beam by forced laminations (Figure 1).
During this procedure, the laminated bamboo layers are straightened out, with the returning moments (or flexural strain energy) stored in their bodies, while the timber core has no flexural deformation and stores no flexural strain energy during the forced lamination process. When the adhesive is cured and the pressure is released by the press machine, the returning moments within the two laminated bamboo layers will work together to prestress the timber core layer, i.e., bend the timber core layer upwards. That completes the prestress process. The stress states of the composite beam at different stages of the process are illustrated in Figure 1(b). When the external loads apply to this prestressed sandwich beam, the timber core layer will be pushed back towards its original position first, in which the prestressed stresses are released before it starts to take on external loads. It will increase the capacity of the timber core layer but at the cost of scarifying part of the laminated bamboo layers' capacity, as the prestressed stresses in the laminated bamboo layers will add to the stresses caused by the external loads. For an optimum design, it is expected that the laminate bamboo layers and timber core are failed at the same time after this adjustment. This technology is extremely useful for combining the low-grade timber with high strength materials such as engineered bamboo. As the prestressed stresses are distributed along the whole length of the beam, the composite beam can be cut short with no impact to the prestress status in the beam.
In this study, an experimental scheme is designed to investigate the impacts of prestress level, layout of the cross-section, and thickness of the laminated bamboo layer on the effectiveness of the proposed prestress technology. All the mechanical properties of the materials used in this study were evaluated by the small clear tests in tension, compression and bending to provide a more accurate parameter study. A total of 22 beams with different configurations were manufactured and evaluated by the four-point bending test. The manufacturing process, test procedure and experimental results are presented in the following sections.

Raw material processing
TC13-B European spruce, imported from Sweden, with C18 strength grade was used in this study. The original lumber size was 5100 mm×150 mm×50 mm before being sawn and planed into the desired glulam timber board size of 2440 mm×120 mm×26 mm as shown in Figure 2(a).
The raw bamboo poles were harvested from Jiangxi province in China with the age of four years. The bamboo poles were then cut into a minimum of 2.5 m to a maximum 2.8 m length based on their knot locations. After initial inspection for exterior defects and visual grading based on the diameter and wall thickness, the bamboo poles were ready for processing, as shown in Figure 2(b). To produce the laminated bamboo, the round poles were split into long strips as shown in Figure 2(c) and the outer green skin (epidermis) and inner yellow layer (pith ring) were planed off to produce the strips with 6 mm to 10 mm thickness and 18 mm high rectangular cross-section. These raw strips were then thermally treated with 110°C to 120°C steam for one hour, and then kiln dried at 30°C to 45°C for 2 days until the moisture content dropped to 8% to 12%. The treated bamboo strips, shown in Figure 2(d), were trimmed to 2,500 mm in length, as shown in Figure 2(e). The long edge surfaces were planed thinner to provide fresh gluing surface for edgewise lamination. Around 18 bamboo strips were grouped to give a minimum 120 mm width, which were then glued in a panel making machine using phenol-resorcinol adhesive as shown in Figure 2(f). The panels were pressed with 1.0 MPa vertical and 1.2 MPa side pressure for the period of 25 minutes at the temperature of approximately 650°C as shown in Figure 2(g). Each of these 18 mm thick, 120 mm+ wide panels were sawn into three equal slices as shown in Figure 2(h). The thinner and more flexible sliced panels are much easier to be moulded into curved laminated bamboo lumber. Each slice was sanded to 2440 mm×120 mm×5.2 mm in size before being moulded into curve laminated bamboo lumber.

Fabrication of glulam core and laminated bamboo arch
A water-based polyurethane structural adhesive for wood was used in this study for the laminated bamboo layer, glulam core and bamboo-timber lamination. It is a twocomponent adhesive; Component A is a compound of water-based composite polymer and polymeric polyol as the main agent and Component B contains isocyanate as the cross-linking agent. The mix-ratio in weight for the main and cross-linking agent is 100:20. As indicated in the product specification, the dry shear strength is larger than 10 MPa and the wet shear strength is above 6 MPa, and the wood breaking rate is greater than 80%. The rate of application on the component surfaces is 250 g/m 2 . The full cure time is 12 hours at room temperature by maintaining 1.2 MPa of pressure on the gluing surfaces.
When preparing the timber glulam core, the freshly planed timber boards were hand-brushed with the fully mixed water-based polyurethane structural adhesive and then four boards were stacked together and pressed for 12 hours using the hydraulic press machine, as shown in Figure 2(i). After the glulam cores were made, these core layers were planed and trimmed to 2440 mm×120 mm×104 mm for laminating with the two curved laminated bamboo lumbers.
A similar procedure was adopted for preparing the curved laminated bamboo layers with several pairs of purpose-built arc-shape mould inserts with different radians, as shown in Figure 2(j). There are two different thickness of these curved laminated bamboo arches. One is 26 mm thick and the other is 52 mm which requires 5 or 10 layers of thin slices, respectively, to be laminated together, as shown in Figure 2(k).
After applying the adhesive to each layer, the laminae were laid in-between the paired mould inserts, pressed and cured for 12 hours, as shown in Figure 2(l).

Assembling and prestressing process
Two options are considered in this study: to attach the laminated bamboo arch to the bottom only or to both top and bottom of the glulam core layer. But their assembling and prestressing process is very similar. First, the gluing surfaces are freshly planed or sanded before applying the adhesive. To prevent different parts slipping away when being pressed down, binding straps were used to temporarily hold the glulam core and laminated bamboo arch(es) together before loading it onto the press machine. These composite components were then pressed to firm contact with 1.2 MPa pressure, maintained for 12 hours. The ambient temperature was around 25°C. The curved laminated bamboo arches were loaded with "prestressing" strain energy when they were pressed to flat, waiting for the pressure to be released, during which the middle glulam core was prestressed by the laminated bamboo layer(s). The final products are shown in Figure 2(m).

Material properties from small clear tests
In order to calibrate the mechanical properties of the timber and bamboo used in this study, 12 small clear specimens each were prepared for compression, tension and bending tests. At the time of test, the moisture content of the laminated bamboo specimens was  , and the specimens were tested at room temperature of . The small clear tests followed the ASTM D143 (2014) standard [46]. The test results are reported in Table 1 and Table 2

Specimen details
A total of 22 beams were manufactured as detailed in Table 3. Two of these 22 beams were the timber glulam beam (TGB), cross-sectional type (1) as shown in Figure   3(a). Both beams were for comparison purposes, and neither were prestressed. Two beams were single laminated bamboo-timber beam (SLBTB), cross-sectional type (2) as shown in Figure 3(b). One of the two beams was not prestressed, i.e., the glulam core was laminated to the straight laminated bamboo layer. This is the control beam for  Table 4.

Test setup
The flexural properties of these 22 beam specimens were evaluated by the fourpoint bending test. A hydraulic loading system, YBD300-160, with a maximum capacity of 300 kN was used to apply the load. The loading rate was 3 mm/min with displacement control. The test started in the elastic stage to plastic stage until rupture.  Table 5 and the Appendix ( Table 6). In summary, the laminated bamboo layer could improve the flexural behaviour of the composite beam by transforming brittle tension failure to more ductile failure accompanied by more evenly distributed mini-fractures to disperse the energy.
Strengthened by the proposed prestress technology, the ductility of the laminated bamboo-timber composite beams was enhanced further. Especially, after unloading, the prestressed laminated bamboo-timber composite beams would normally recover largely to the original shape which was powered by the residual prestressed stresses within the laminated bamboo layers. This would be very convenient for repairing and structural strengthening after being damaged.  Figure 7(a)-(d), at the point where the structural stiffness changed the most during the test. It was less accurate than the ultimate values, but they are reasonably accurate indicators for structural stiffness at the elastic range.

Yielding, ultimate loads and corresponding deflections
To compare the performance of the prestress effect of the laminated bambootimber composite beams, the above indicators were used. All the test results of the specimen beams that failed in dominant mode ⑦: failure with timber defects, were removed from Table 5 to improve the regularity. The rest of the test results were grouped by the type of cross-section and level of prestress, specified in Table 3.  controlled by the curvature of the laminated bamboo layers in this study, should be gauged to use this redundant capacity to prestress the timber to a stress status that will counter-act with the external loads and delay its failure to synchronise with laminated bamboo rupture, i.e., the capacity of both materials is fully utilised. Figure 8 presents the load-strain relationship measured by the strain gauges mounted along the height of the side surface at mid-span for all the specimen beams.

Strain analysis for the mid-span cross-section
As shown in Figure 8, the strain across the cross-section of the specimen beams maintained linearity before the yielding. After yielding, the strain across the crosssection exhibited an increasing degree of nonlinearity but most of the nonlinear behaviours appeared at the very end of the rupture test. From the test results, it can be seen that in majority of the test period, except the short period before rupture, the plane section remain plane assumption is still valid.
As shown in Figure 8(a), the strain distribution across the cross-section of the TGB mostly remained linear, and the plastic and rupture stages are relatively short. This means that the TGB suffered brittle failure, a less favourable failure mode in practical engineering applications.
Compared to the TGB, as shown in Figure 8 The ultimate load of the SLBTB beam, as shown in Figure 8(b) was also increased significantly due to not only using stronger laminated bamboo in tension but also better utilisation of the timber layer in compression. The strain gauge 4 and 5 readings in 12518 , which were 39.76% and 25.22% higher than the non-prestress LBTSB, respectively. The rate of utilisation of the laminated bamboo layer was significantly increased but the tension layer was about to reach its maximum strength while in compression there was room for the strain to develop further.
Considering that the maximum ultimate tensile strength (13609 ) is much lower than the compression one ( 20957 − ), when prestressed level increased to medium, the initial prestressed strains increased to 3003  , there was not much further room for the tensile strain to develop when compared to the low-level prestressing. The imposed strains at rupture for the medium-level prestressed LTBSB at top and bottom mid-span were about 12880 − and 10559 , respectively. Clearly, the tensile strain was the governing factor. After adding on the Initial strain, these ultimate actual strains were  15883 − and  tension with a small increase but the ultimate strain in compression decreased instead.
This combined effect leads to limited increase of the ultimate load with the medium level of prestress. For the LTBSB with a high level of prestress, the initial prestressed strain was 4153  , and the imposed strains at rupture in compression and tension were 11147 − and 9325 , respectively. Both strains were lower than the corresponding test results from the LTBSB with a medium level of prestress, which indicated that these beams were over prestressed as there are less room for the imposed strains to be added onto the initial strain. After adding on the Initial strain, the ultimate actual strains were 15300 − and 13478 , nearly the same as the medium level prestress. and 51.03% increase, respectively, when compared with the non-prestressed results.
The prestress effect was nearly doubled when compared with LTBSB with thinner laminated bamboo layers. In conclusion, the prestressed technology proposed a new approach to balance the uneven capacities by offsetting the strength from the stronger material to the weaker one. The best structural performance of the prestressed bambootimber composite beam can be achieved by fine tuning the thickness ratio between the laminated bamboo and timber layer, prestress level and cross-section layout. The optimum design should aim to maximise the rate of utilisation of all types of materials at various locations.

Serviceability limit loads
As with other practical timber designs, the deflection of laminated bamboo and timber beams should satisfy the serviceability limit conditions specified in Eurocode 5 (EN1995) [47], i.e., the serviceability limit deflection of the beams should satisfy the For the sake of comparability, mm l 07 . 8 300 = is taken as the deflection limit for the serviceability limit state analysis, the corresponding serviceability limit loads of each beam, which were recorded from the tests at the moment the beam deflection reached 8.07 mm, are shown in Table 3 and Figure 10. The serviceability limit loads are largely governed by the structural stiffness at the elastic stage, as the elastic modulus of the laminated bamboo and timber are similar. That is also the reason, as shown in Figure   10, why changing the cross-section layout (SLBTB and LBTSB) did not have a significant impact on the serviceability limit loads but changes to the overall dimensions of the cross-section (ELBTSB) did. Figure 10 also indicates that increasing the level of prestress will also increase the serviceability limit load.

Conclusions
This well-rounded experimental proof-of-concept study on the prestressed laminated bamboo-timber composite beams has proven that the novel concept of the proposed prestressed technology for pure bio-based composite structure members is effective with potential to be improved and optimised. The research findings are underpinned by the material validation with small clear tests, structural performance evaluation with four-point bending tests, and component analysis with strain gauge data.
The following key conclusions are drawn based on the outcomes of this experimental study: 1) The experimental proof-of-concept study proved the concept of forcelaminating multi-species components with different geometrical parameters together and activating the prestressing process during the pressure release step is feasible and effective.
2) The specimens were fabricated in an industrial level manufacturing facility by slightly modifying the existing lamination process. It proved the adaptivity and scalability of the proposed technology.
3) A comprehensive analysis and categorisation of the failure modes of different types of beams were conducted with the attempt to reason the failure mechanism with different design parameters, such as prestress level, laminated bamboo to timber ratio, cross-section layout, etc.

4)
The structural performance of the prestressed laminated bamboo-timber beams, indexed by the yielding load, ultimate load, serviceability limit load and corresponding deflections, was evaluated and compared with the nonprestressed composite and traditional glulam beams. The prestressed effect has proved to be effective but could be influenced by the level of prestress, laminated bamboo to timber ratio, cross-section layout, etc.

5)
An in-depth component analysis underpinned by the extensive strain gauge records was conducted to reveal the interaction between the laminated bamboo layer(s) and timber core. The analysis also laid out several key assumptions for future analytical and numerical studies: a. plane section remains plane assumption is valid at most of the periods before rupture b. cross-section neutral axis shifting caused by strain softening must be considered in the future analytical development and numerical studies c. the actual stresses and strains of a composite beam can be estimated by the superposition of the prestressed stresses and strain with the stress and strain caused by the external loads.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
(b) Illustration of the stress states at different process stages  k) ELBTSB-3a Figure 9 Strain distribution on the upper and lower surface of the left span of the testing beam