Engineering Structures 184 (2019) 495–504 Contents lists available at ScienceDirect Engineering Structures journal homepage: www.elsevier.com/locate/engstruct Combined culm-slat Guadua bamboo trusses L. Villegas, R. Morán, J.J. García T ⁎ Escuela de Ingeniería Civil y Geomática, Universidad del Valle, Cali, Colombia A R T I C LE I N FO A B S T R A C T Keywords: Bamboo light structures Ring joints Truss Floor Roof This study pertains to the design and structural testing of trusses assembled with culms and slats of the bamboo species Guadua angustifolia (GA). The trusses were constructed using a novel design of steel clamps to accomplish the connections between the top and bottom chord culms and the diagonal and vertical slat elements. The clamps offer a better alternative to the conventional joining methods that typically lead to longitudinal splitting of the culms, since the clamps counteract the formation of longitudinal splitting and accommodate themselves to the geometric irregularities of the culms and slats. The experimental program included the monotonic static tests of pilot Howe truss prototypes with variations in the clamp joints. Next, four replicates of the proposed design (3 m long and 0.4 m high weighing 265.4 N) were tested under monotonic static load. The trusses were simply supported at the ends and loaded at three central joints. The stiffness of the trusses up to 9869 N (1.5 times the service load for floor use) was 631.3 N/mm (COV 3.6%), and no failure was shown at the proof load of 13,754 N. This pilot study demonstrates that these trusses are a good candidate to substitute the traditional materials to support the floors and roofs in low-cost, prefabricated housing projects. However, more experimental studies have to be accomplished to fully validate this application. 1. Introduction The alarming data regarding global warming, which is associated with the high energy required for the production of traditional materials, such as steel and concrete, has motivated the search for nonconventional and renewable materials [1–3]. One of them is Guadua angustifolia (GA), which is a species of bamboo that grows naturally in large areas of Central and South America. GA has a high longitudinal strength as compared to low carbon steel [4], and GA plantations play an important role in reducing erosion and controlling water cycles [5]. GA culms are cylindrical tubes with transverse diaphragms (natural nodes) located at variable distances along the length of the culm. The material is reinforced with strong cellulose fibers immersed in a weak and flexible lignin matrix [6,7]. The fibers are oriented axially in the internodes. Therefore, the bamboo mechanical properties are directional, with a high axial strength as compared to mild steel [8,9], but weak in transverse direction [5,10,11]. Preferential bamboo failure mode is longitudinal splitting, which is attributed to circumferential tension and shear stress on the fiber planes [11,12]. Globally, there are many full-culm bamboo structures, such as scaffolds, trusses, bridges, and houses [7,13–15]. However, bamboo construction remains a vernacular practice [16]. This might be attributed in part to the challenges to develop standard connections due to the tubular shape of the culms, the dimensional variations and the anisotropy of the material. Bamboo structures usually use coped joints at the culms ends to increase the area of contact, mortar injection to improve the transverse strength and through bolts to transmit the load [17–23]. These joints are prone to longitudinal splitting and their construction requires substantial workforce, increased construction time and cost and are not amenable to prefabricated processes. Other joints do not need holes, like the handmade used by Seixas et al. [24] for space structures, but cannot be easily implemented using prefabricated processes. Modern connections [25–27] are versatile and efficient but require relatively complex and expensive parts to be used in low-cost projects. One way to overcome the limitations of using full-bamboo culms is to produce engineered bamboo, which results from processing the raw culms into laminated composites, similar to glue-laminated timber products [28–35]. Engineered bamboo allows the use of standardized sections, which very much facilitates the construction process with techniques similar to those used in timber structures [28,36–38]. Unfortunately, engineered bamboo has a high environmental impact as compared to using raw culms [39]. ⁎ Corresponding author. E-mail addresses: laura.villegas@correounivalle.edu.co (L. Villegas), richard.moran@correounivalle.edu.co (R. Morán), jose.garcia@correounivalle.edu.co (J.J. García). https://doi.org/10.1016/j.engstruct.2019.01.114 Received 2 February 2018; Received in revised form 21 January 2019; Accepted 24 January 2019 Available online 01 February 2019 0141-0296/ © 2019 Published by Elsevier Ltd. Engineering Structures 184 (2019) 495–504 L. Villegas et al. of 681.1 MPa (COV 32%) and a circumferential tensile strength of 8.0 MPa (COV 26%, n = 10). The tests of slats were conducted under bending and torsion, according to the protocol proposed in [47]. The axial modulus obtained was 8787.2 MPa (COV 18%) and the circumferential-axial shear modulus 747.8 MPa (COV 21.4%, n = 10). These properties are similar to those obtained in previous studies [37] and are representative of a good quality GA material for construction. The clamps, curved washers and steel connectors (Fig. 1.b) were manufactured with ASTM A36 steel by students in the lab using a metal cutter, a plate bender and a drill press, with a relatively simple and fast procedure. The width of the GA slats and of the curved washers used to connect them was 40 mm. To promote the massive use of raw bamboo, easy and efficient construction methods have to be devised, for example, by developing technologies using prefabricated parts, as effectively implemented in the manufacture of timber houses [40]. Thus, in previous studies, we proposed the use of metal parts to connect culms and slats with great versatility [41,42]. First, we developed trusses made of GA slats that can be used for roof applications [41]. However, buckling of compression members was a concern for the larger spans. Later, we proposed using thin steel clamps to connect bamboo culms and improve the resistance to longitudinal splitting [42]. Steel clamps can accommodate a wide range of culm sizes, solving the problem of the joints customization. Based on the aforementioned joints, the objective of this manuscript is to propose a bamboo truss design that can potentially be used to support roofs and floors of low-cost houses using prefabricated processes. Pilot prototypes with variations in the joints were tested for the development of the proposed design. Next, four replicates of the proposed design were evaluated under static test and simulations. 2.3. Slat joints Fig. 2 shows a typical joint between slat elements and a culm. The slat elements were joined to the clamp lugs using 3.2-mm thickness plates (ASTM A36) that were connected to the slats through 9.5 mm diameter bolts grade 2 and inserted into 10 mm diameter holes drilled at 35 mm from the ends of the slats [48]. The bolt was tightened to apply compression along the thickness of the slat or the radial direction. The slats were outfitted with curved washers between the bolt head and the slat and between the nut and the slat to redistribute the pressure caused when tightening the bolt. For the elements under tension, curved washers were also inserted between adjacent slats. The curvature of the washers (radius of 62 mm) was nearly matched to that of the slats. As a simple means to control the bolt tightening when assembling this joint, the nut was tightened two turns after the parts came in contact. With two nut turns, the nominal radial deformation for a 10mm thick slat was 31.8%, which is lower than the failure deformation of the GA under radial compression [45,49]. The torque to tight the joint is in the range of 34–48 N-m. For the slat elements under tension, the critical failure mode is shearing on the longitudinal fiber planes. Previous experimental studies [41] indicated that the shear strength in joints with radial compression was 9.96 MPa. As this joint with radial compression is not covered in the Colombian standard, as a first approximation to establish preliminary allowable values to check this study’s designs, we used a safety factor of 3 that yields an allowable shear stress of 3.3 MPa. The safety factor is equal to that used by Paraskeva et al. [50]; however, we are fully aware that a more extensive experimental research has to be carried out to validate this proposal. 2. Materials and methods 2.1. Design specifications and truss geometry The configuration of the specimens to be tested were that of trusses intended to support the floor of prefabricated low-cost houses, which is based on 3 m × 3 m modules. The Colombian Building Standard [43] was used as a guide to verify the structural performance of the truss. For GA design, the strength verification is based on allowable stresses. Assuming that the truss will be used to support floors, a live load of 1766 N/m2 (180 kg/m2) and a dead load of 1373 N/m2 (140 kg/m2) were considered, which sum to a service load of 3139 N/m2 (320 kg/ m2). The trusses must be oriented parallel to each other to bridge the 3 m × 3 m space. If the separation of the trusses that conform the floor is 0.75 m, a distributed service load of W = 7063 N corresponds to each truss. The joints used in the trusses are not included in the Colombian standard. Hence, we verified the resistance of the joints of the proposed truss based on the results of preliminary tests presented in this paper and others reported elsewhere [41]. The trusses must also comply with a maximum deflection according to the application. The strength of the steel components was checked according to the provisions of the Colombian standard [43]. A Howe truss type was chosen, where the top and bottom chords were composed of GA culms, while each of the diagonal and vertical elements were formed with longitudinally superimposed GA slats about 40 mm wide (Fig. 1a). Upon the application of vertical loads to the three upper nodes, the shorter vertical elements are compressed and the diagonal elements are tensed. Thus, this truss type allows increasing the buckling load in the vertical members, which might be critical in some cases [41]. Clamps were used to transmit the load from the chords to the slat elements. In this joint design (Fig. 1b), both the external load and the forces from the web elements are applied to the lugs. Hence, the vertical loads are directly transmitted to the web elements, and the culms work only under the axial loading. 2.4. Culm joints The connection to the culms was effect by split clamps formed by two semi-rings [42,49], as depicted in Fig. 3a. The semi-rings were manufactured in four sizes to cover a range of culm diameters of 95–135 mm. The clamp was tightened around the culm with bolts (9.5mm diameter, grade 8) inserted into 10-mm diameter holes drilled in the lugs. The clamps were manufactured with steel plates (ASTM A-36) of 3.2-mm thickness and 25.4-mm width. This thickness provides good flexibility for the clamp to deform and follow the geometrical irregularities of the culm. The clamp applies external pressure to the culm that in turn generates hoop compressive stress. This compression counteracts the formation of longitudinal splitting. The clamp action is similar to the shrink fit technique widely used to set shafts to wheel hubs in rotatory machinery that are controlled by radial interference. For the clamp, the equivalent to circumferential interference (i.e., 2π times the radial interference) is proportional to the difference between the outer circumference of the culm and the curved central length of the semi-rings [42]. After tightening the bolts, a gap x remains between both the semirings (Fig. 3a). Thus, the circumferential interference Ic, which controls the clamping force, was calculated as: 2.2. Materials The bamboo material was obtained from the GA culms between three and five years of age, with about 12 m in length, growing in the Eje Cafetero (Colombia, South America) at an altitude of 1500 m. Culms were supplied by CO2 Bamboo Ltda. (Candelaria, Colombia). The material was borax treated and oven dried as described in literature [44,45]. The average humidity content was 16.9% (SD 1.6%, n = 5). Mechanical tests were performed to obtain representative values of the elastic constant of the GA material and the tensile circumferential strength. The edge bearing test [46] yielded a circumferential modulus 496 Engineering Structures 184 (2019) 495–504 L. Villegas et al. Fig. 1. a. Geometrical configuration for the proposed truss. b. Free body diagram of the joint A. Fig. 2. Joints used to transmit the load from the slat elements to the clamp lugs and critical shear planes for the elements under tensile load. Ic = Pc − 2La − 2x (1) implemented in the pilot trusses described in Section 2.6; both of them intended to provide redundancy to the joint and avoid early slippage. In the first variation (Fig. 3b), two drywall screws (4-mm diameter, 25.4mm long) were inserted through the ring of the clamp placed at diametrically opposite positions near the lugs. These screws can easily be installed after the clamp is tightened. The other option (Fig. 3c) was to insert a through bolt across the diameter (9.5-mm diameter) into 10mm diameter holes. where Pc is the perimeter of the culm and La is the interior curved length of one of the semi-rings. It is recommended to use circumferential interferences that are higher than 10 mm to avoid early slippage of the clamp and lower than 23 mm to avoid damage of the culm. For this range of interferences, the torque needed to adjust the clamp is in the range of 48–68 N m. Two variations of the aforementioned clamp-culm joint were also 497 Engineering Structures 184 (2019) 495–504 L. Villegas et al. Fig. 3. a. Simple clamp-culm assembly. b. Clamp-culm with two screws. c. Clamp-culm with a through bolt. Table 1 Maximum axial load capacity (Fmax) of the three clamp-culm configurations as shown in Fig. 3. D and t are culm diameter and wall thickness, whereas Ic is the circumferential interference. Group n D (mm) (COV) t (mm) (COV) Ic (mm) (COV) %H (COV) Fmax (N) (COV) 1 (only clamp) 10 2 (2 dw screw) 10 3 (1 through bolt) 5 110.6 (3.5%) 107.6 (0.3%) 105.5 (1.0%) 10.5 (14.1%) 10.3 (6.3%) 10.1 (8.2%) 17.2 (28.2%) 18.4 (28.5%) 10.4 (1.1%) 14 (6.3%) 14.5 (3.1%) 14.6 (3.2%) 18,275 (18.6%) 18,686 (33.1%) 15,414 (3.3%) 2.5. Experimental set-up and testing procedure The setup used to test the actual trusses is depicted, schematically and in actual rendition, in Fig. 5. Three levers, pivoting on radial bearings at one end and supporting the weight of bricks at the other end, transmitted the amplified weight to the three central nodes of the upper chord of the truss. Each lever was made up two steel angles of 50.8-mm × 3.2-mm and weighed approximately 54.9 N. The levers transmitted the load to the truss through square-thread screws, which allowed leveling the system before reading the deflections. Small bronze cylinders were used to reduce friction. The lever mechanical advantage was 8.6, i.e., each brick of 112.8 N caused an increment of load of 970.2 N over the specimen. The mechanical advantage of 8.63 of the levers was confirmed by using a load cell that measured the actual load transmitted to the specimen. No support was used to prevent the out-of-plane displacement. The Fig. 4. Test set-up to determine the axial load needed to draw the clamp from the culm. Tests, as described in [42], were performed to determine the axial load capacity of the three clamp-culm configurations (Fig. 3). Briefly, the test consisted of applying a force to one end of the culm, while the clamp lugs were supported on a steel pipe (Fig. 4). Table 1 presents the results of preliminary experiments [42], showing the average of the maximum axial load for each of the three joint configurations shown in Fig. 3. 498 Engineering Structures 184 (2019) 495–504 L. Villegas et al. Fig. 5. Experimental set-up used to test the trusses, a. Levering system, b. Top view. Fig. 6. Steel brackets used to transmit the load and to support the prototypes. The vertical displacements were measured at the points shown in Fig. 7, using four dial gauges with a resolution of 0.0254 mm (0.001″). Truss stiffness was calculated as the slope of the straight line fitted to the experimental load-displacements points, where the deflection was measured with indicator 2 (Fig. 7). The weight of the levers and the wooden planks was used as the first load increment. Next, each additional load increment of about 1942 N was accomplished by placing two bricks on the planks. During the test of the four replicates of the proposed design, the maximum load was 13,754 N, which was about twice the service load. As prescribed in ASTM E73-13, these proof tests were intended to assure that the truss supports a stated load and to determine maximum bronze cylinders that transmitted the load were visually located to coincide with the center of the upper chord. U-shaped brackets were used to support the end clamps and transmit the load to the lugs of the three upper central clamps (Fig. 6). At a load level of 10,000 N over the truss, the calculated contribution to the deflections of the steel profiles supporting the ball bearings was lower than 0.1%, while the contribution of the steel brackets was about 3.1%. Each load increment was accomplished by adding bricks of 112.8 N (COV 7.1%) weight over the wooden planks (26.5 N weight) that were connected to the levers by steel cables. Each load increase was held for 3 min approximately. The total time of one complete test was about 30 min. 499 Engineering Structures 184 (2019) 495–504 L. Villegas et al. Fig. 7. Positions to measure the vertical displacements marked with red points. 2.7. Description of the proposed design deflections. In some of the tests of the pilot prototypes, the load was increased until failure. The proposed design comprises single culms for the top and bottom chords (Fig. 8), three-slat elements for the end diagonals and two-slat elements for the vertical elements and inner diagonals. The conformation with single clamps retrofitted with two dry wall screws were used in the culm joints. To provide good confinement around the holes of the diagonal elements under tension, that are under shear stress, curved washers were inserted between adjacent slats. As the mechanism of failure of the vertical elements is not shear, only two washers were used, one at each side of the two-slat element. No restriction about the clamp location with respect to the natural nodes of the culms is prescribed in this design. Four replicates of the proposed design were built and tested. Table 4 presents the culm dimensions of the replicates and the range of circumferential interferences. With all the materials at hand, a truss can be assembled in approximately two hours. A preliminary estimation indicates that the cost of each truss is about US$50, including materials and workforce. This cost can be accurately calculated in a real production process. The weight of the truss is approximately 19.8 kg (67%) of GA and 9.7 kg (33%) of steel. 2.6. Description and test results of pilot prototypes Pilot prototypes with the clamp-culm joints described in Fig. 3 were developed and tested. These tests were mainly used to assess the facility of construction and improve the redundancy of the joints and truss stiffness, which resulted to be a critical factor for design. We next present a short description of these pilot prototypes. GA elements, clamps and other connectors were reused in most of the cases. A description of the joints used in each prototype and a summary of the results is presented in Table 2. In two of the groups, two clamps were used to join the culm to the slat elements: one connected to the slat elements and the other intended to stop the slippage of the connected clamp. In two groups, through bolts were used, which did not increase the stiffness of the truss. Thus, as a through bolt is relatively difficult to install, this option was not further considered. One test with two dry wall screws in the clamps showed a relatively good stiffness of 636 N/mm. A wide range of 11,811–23,466 N was obtained for the maximum load. One of the prototypes did not fail at the maximum capacity of the set up (23,466 N), whereas two others were not loaded to failure to preserve them for other tests. Eleven specimens failed by clamp slippage, characterized by an important degree of axial displacement of the clamp (∼1 cm) with respect to the initial position, as depicted in the first photo of Table 3. Two specimens failed by shear of the end diagonals (second row of Table 3) along the areas shown in Fig. 2. In one case the upper culm, weakened by a through hole, showed excessive local deformation near the support (third row of Table 3). In two cases the upper chord failed by shear (fourth row of Table 3). These culm failures may be attributed to the propagation of minor fissures caused during previous tests. No sudden collapse of the truss was observed in any of the experiments. The maximum load was always higher than the service load, as shown in the last column of Table 2. In the trusses with a single clamp without dry wall screws (row 1 of Table 2) the only mechanism of load support is friction. Thus, they failed by clamp slippage, two of them at a load of 11,811 N. The insertion of two drywall screws blocks the clamp from the culm and creates another mechanism of load support complementary to friction, adding redundancy to the joint. Based on these experiments, we suggest using at least 30 mm between the clamp and the end of the chords under compression to provide good confinement to the material. This distance should be increased to 50 mm for the chords under tension to prevent the complete extraction from the culm. As the dry wall screws add redundancy to the joint and can be easily installed, the option with a single clamp and two dry wall screws was finally adopted to build and test four replicates. 3. Theoretical model and structural verification of the proposed design The internal forces in the truss might be readily calculated by hand. However, to have a prediction of the stiffness considering the eccentric connections between the web elements and the chords, a model was developed using ANSYS 16.0 (SAS IP, Inc, Chyenne, US). The model is composed of beam elements and included finite deformations (Fig. 9). Off-set distances were prescribed to consider the eccentricity of the joints between the web elements and the chords. A tubular cross section of 120-mm diameter and 10-mm thick was considered for the culms, while a rectangular cross section 40-mm wide and 10-mm thick was considered for all the slats. A Young’s modulus of 9500 MPa was prescribed for GA, according to NSR-10 [43]. Spherical joints were used to transmit only forces at the ends of the web elements. This model does not include complex sources of nonlinearities present in the real structure. However, it provides a reasonably good approximation of the first part of the load-deflection curve, which is approximately linear. Three vertical loads W/3 were applied at the central upper nodes while the ends of the truss were considered to be simply supported. The maximum compression force (−1.25 W) occurs in the central portion of the upper chord, while the maximum tension (1.06 W) occurs in the end diagonals (Fig. 9). For the service load, Table 5 presents the structural verification of the design. For the clamp joints, a safety factor was calculated as the ratio of the mean axial load capacity reported in Table 1 over the load transmitted by the clamp. The maximum deflection predicted for the service load of 7063 N is 11.4 mm, which is lower than the admissible deflection of 12.5 mm, as prescribed by the Colombian Standard [43]. The consideration of the 500 Engineering Structures 184 (2019) 495–504 L. Villegas et al. Table 2 Summary of the results of the pilot tests. Type of joint Truss stiffness range (N/ mm) Range of maximum load (N) Type of failure Range of ratio of maximum load over service load Single clamp (n = 4) 416–585 11,811–17,639 Clamp slippage (n = 4) 1.67–2.50 Double clamp (n = 3) 429–600 11,811–23,466 Clamp slippage (n = 1) Two other did not fail 1.67–3.32 Single clamp, through bolt (n = 5) 354–493 11,811–21,523 Shear failure (n = 2) Local deformation end clamp (n = 1) Culm failure (n = 2) 1.67–3.05 Double clamp, through bolt (n = 2) 400–618 19,580, both Clamp slippage (n = 1) Shear failure (n = 1) 2.77 Single clamp, two dw screws (n = 1) 636 13,654 No failure 1.93 3.6%), which is approximately 8.4% higher than that predicted with the theoretical model (578.3 N/mm). No out-of-plane displacements were visually noted. For the service load of 7063 N, the experiments show a maximum deflection of about 10.5 mm, which is lower than that of the maximum deflection of 12.5 mm, as prescribed by the Colombian Standard [43] for floors. The differences in the experimental curves might be explained by the inherent variations of elastic constants and dimensions of natural materials. As the main sources of nonlinearities, i.e., slippage at the joints, permanent plastic deformation of the clamp lugs and deviations of the axis of the elements with respect to a straight line, were not included in the ANSYS model, the theoretical curve is linear, while the experimental curves show decreasing stiffness with load increments. The nonlinear effects mainly arise at higher loads levels, which might off-set distances in the connections of the web elements to the chords reduces about four times the stiffness as compared to a model with centric connections. 4. Results and discussion No replicate of the proposed design showed any sign of failure when loaded at 13,754 N, which is approximately twice the service load. This load level supported by the four replicates of the proposed design is higher than the failure load of two of the examined trusses without drywall screws (row 1 of Table 2), that showed slippage at load levels of 11,811 N, demonstrating that the drywall screws increase redundancy of the joint. The load-displacement curves showed reduction of the stiffness with load increments (Fig. 10). At a load level of 9869 N (approximately 1.4 the service load), the stiffness was 631.3 N/mm (COV 501 Engineering Structures 184 (2019) 495–504 L. Villegas et al. compared to using laminated bamboo. The slats can be cut from those culms that do not conform to the geometrical constrains as prescribed in the construction codes (e.g., conicity) or have substantial longitudinal splitting due to previous treatments (such as immunization, drying, etc.). Therefore, this system reduces material wasting, and it could be implemented in prefabricated processes. Additionally, these trusses provide applications to increment the utilization of a sustainable non-conventional material like GA at a competitive cost. In turn, more use of GA in construction might create economic incentives to increase the plantations. GA forests (guaduales) play a key role in preserving the environment as they control water cycles, reduce erosion and capture CO2. As compared to our previous proposed trusses made only out of GA slats [41], the construction of the combined culm-slat truss is simpler and less vulnerable to failure by lateral instability and buckling of the upper chord under compression, since the tubular shape is well optimized by nature for this application. In a real application to support floors, the trusses can be transversally connected with the GA slats to conform a fully tridimensional structure. The transverse connections also prevent out-of-plane displacement and reduce the buckling load of the upper chord. We are in the process to assemble and test a full scale 3 m × 3 m floor prototype that we expect to describe in future publications. One limitation of the study is that the mechanical tests consisted of monotonic static load, which does not allow detecting the response of the structure under cyclic forces. In addition, the developed linear model is only useful to estimate the truss stiffness under service loads but it has a limited value to understand the influence of many sources of non-linearity that arise at higher loads. Thus, experimental studies with the joints have to be carried out to obtain characteristic load-deflection curves to be input into the computational model to reproduce the nonlinear behavior of the structure under different loading scenarios. The span of the prototypes developed in this study was chosen to accommodate a typical modular unit oriented to low-cost housing. More tests have to be conducted to expand this design to longer trusses that might be used for other applications, such as pedestrian bridges. Additionally, more studies have to be performed to analyze the increments of deflections with time. Table 3 Type of failures observed in the pilot tests. Type of failure and description Clamp slippage Shear failure in two-slat diagonal elements Excessive local deformation in through-bolt clamps Shear failures of the upper culm explain higher differences in the curves for larger deflections. This study shows that the combined culm-slat GA trusses could be a sustainable alternative for a wide variety of applications. However, other experimental studies need to be conducted with a higher number of specimens to characterize the joints under various conditions and geometrical configurations for determining allowable levels of stresses for design. In addition to reducing the propagation of longitudinal fissures and creating ductile modes of failure, the construction process with the clamp joints is relatively simple. This is because no complex parts are required and there is no need to use mortar injection. This truss system is composed of raw culms and slats. Thus, the energy consumption to use the material is kept to a minimum, as 5. Conclusions The trusses proposed can be a sustainable alternative to support roofs and floors of low-cost housing projects using prefabricated processes. However, more experimental studies have to be conducted to fully validate the use of these trusses under construction codes. Other applications of these trusses might be short pedestrian bridges in rural areas, but more studies on the topic are needed to accomplish good for longer spans Two dry wall screws Two slats Three slats Fig. 8. Representation of the proposed truss. 502 Engineering Structures 184 (2019) 495–504 L. Villegas et al. Table 4 Range of diameter (D), circumferential interference (Ic) and thickness of the four replicates of the proposed design. Upper chord Lower chord Replicate 1 2 3 4 1 2 3 4 D (mm) Ic (mm) t (mm) 119–121 16–22 10.2–10.8 112–118 11–23 9.5–10.4 112–117 10–22 9.9–10.7 110–112 11–23 9.0–9.4 114–119 12–20 10.6–11.2 114–117 10–21 10.1–10.7 110–115 12–18 10.1–10.8 111–112 20–25 9.6–10.0 Fig. 9. Compression and tensile load and elements numeration. Table 5 Structural verification of the proposed design. The identification of the elements and clamp joints is presented in Fig. 9. Element ID Clamp Joint ID Identification Internal Force (N) Admissible load (N) Criteria to calculate the admissible load C1 C2 C3 C4 T1 T2 T3 −6639 −3532 −8829 −2331 7847 2543 6639 −10,074 −11,411 −10,074 −11,411 12,474 8316 39,301 Buckling load for columns under provisions of Colombian Norm [43]. Diameter equal to 100 mm and thickness of 10 mm Shear in connection. Shear area = 2 * 35 * t * n mm2, n number of slats, t = 9 mm, admissible shear stress = 3.3 MPa Tension element, admissible stress of 13.9 MPa Identification Force (N) Safety factor Criteria to calculate the safety factor J1 J2 J3 J4 J5 6639 6639 2190 0 0 2.81 2.81 8.53 NA NA Mean maximum value of 18,686 N reported in Table 1 for clamp-culm joints retrofitted with two dry wall screws Acknowledgments We thank Jaime Buitrago for his critical review. This research was financially supported by grant C.I.138-52110517 of the Universidad del Valle. R. Morán is grateful to Administrative Department of Science, Technology and Innovation COLCIENCIAS for financial support through a national doctoral grant. References [1] De Flander K, Rovers R. One laminated bamboo-frame house per hectare per year. Constr Build Mater 2009;23:210–8. [2] Ghavami K. Bambu: um material alternativo na engenharia. Rev Eng Eng Ed Téc LtdaSão Paulo 1992;492:23–7. [3] Ghavami K. Application of bamboo as a low-cost construction material. Bamboos Curr. Res. Proc Int. Bamboo Workshop Cochin India; 1998. [4] Tam CPT. Comportamiento estructural de la guadua angustifolia. Uniones en guadua. Ing. E Investig., vol. 24, 2004, pp. 3–7. [5] Arce Villalobos OA. Fundamentals of the design of bamboo structures. Ph.D. Thesis Eindhoven University of Technology, Holland; 1993. [6] Janssen J. Bamboo in building structures. PhD Thesis, Eindhoven University of Technology; 1981. [7] Janssen J. Designing and building with bamboo. Technical report No. 20. Beijin: International Network for Bamboo and Rattan; 2000. [8] Ciro HJ, Osorio JA, Vélez JM. Determinación de la resistencia mecánica a tensión y cizalladura de la Guadua ansgustifolia Kunth. Rev. Fac. Cienc. Agropecu. Univ. Nac. Medellín; 2005. p. 2709–15. Fig. 10. Comparison of the experimental and theoretical load deflection curves for the four replicates of the proposed design. stiffness and strength for longer spans. The use of two drywall screws, one in each semi-ring, provided joint redundancy. These two screws were easily installed, and they did not reduce the resistance of the structure, as the circumferential compression caused by the clamps counteracts the propagation of longitudinal splitting. The clamp connection used to join the culms can be easily installed and creates ductile modes of failure. Additionally, it might be easily implemented in prefabricated processes. 503 Engineering Structures 184 (2019) 495–504 L. Villegas et al. [32] Luna P, Olarte AM, Takeuchi C. Theoretical and experimental analysis of structural joints of glued laminated pressed bamboo guadua for a housing project. DYNA 2014;81(184):110–4. [33] Rodríguez JJP, Tam CPT. The structural behaviour of laminated-guadua panels under parallel plane loads. Ing E Investig 2012;32(2):18–22. [34] López L, Correal J. Exploratory study of the glued laminated bamboo Guadua angustifolia as a structural material. Maderas Cienc. Tecnol. ISSN 0717-3644 Vol 11 No 3 2009, vol. 11, Dec. 2008. p. 171–82. [35] Echeverry JS, Correal JF. Cyclic behavior of laminated Guadua Mat sheathing-toframing connections. Constr Build Mater 2015;98:69–79. [36] Sharma B, Gatóo A, Ramage M. Effect of processing methods on the mechanical properties of engineered bamboo. Constr Build Mater 2015:95–101. [37] Xiao Y, Chen G, Feng L. Experimental studies on roof trusses made of glubam. Mater Struct 2014;47(11):1879–90. [38] NTC 2500, Instituto Colombiano de Normas Técnicas - ICONTEC, Uso de la madera en la cosntrucción; 2010. [39] Zea Escamilla E, Habert G. Environmental impacts of bamboo-based construction materials representing global production diversity. J Clean Prod 2014:117–27. [40] Midon M, Pun C, Tahir H, Kasby N. Construction manual of prefabricated timber house. FRIM Technical Information Handbook. Malaysia: Forest Research Institute Malaysia; 1996. [41] Villegas L, Morán R, García JJ. A new joint to assemble light structures of bamboo slats. Constr Build Mater 2015;98(Nov.):61–8. [42] García JJ, Benitez C, Villegas L, Morán R. Thin steel rings as a feasible alternative to connect bamboo culms, presented at the 17th International Conference on NonConventional Materials and Technologies (17thNOCMAT 2017), Mérida, México; 2017. [43] AIS, NSR-10. Reglamento colombiano de Construcción Sismo Resistente NSR-10. Bogotá, Colombia: Asociación de Ingeniería Sísmica; 2010. [44] ISO 22157-1, Bamboo – Determination of physical and mechanical properties; 2004. [45] García JJ, Rangel C, Ghavami K. Experiments with rings to determine the anisotropic elastic constants of bamboo. Constr Build Mater 2012;31:52–7. [46] Moran R, Webb K, Harries K, García JJ. Edge bearing tests to assess the influence of radial gradation on the transverse behavior of bamboo. Constr Build Mater 2017;131(Jan):574–84. [47] Moran R, Ghavami K, García JJ. A new method to measure the axial and shear moduli of bamboo. Proc Inst Civ Eng - Struct Build 2017;170(4):303–10. [48] Villegas L. Thesis: Desarrollo y evaluación de una nueva unión para estructuras de latas de guadua, Universidad del Valle, Cali, Colombia; 2014. [49] Morán R, Muñoz J, Silva HF, García JJ. A bamboo beam-colum connection capable to transmit moment, presented at the Non-conventional Materials and Technologies NOCMAT, Mérida, Yucatán, México, 26-Nov-2017. [50] Paraskeva TS, Grigoropoulos G, Dimitrakopoulos EG. Design and experimental verification of easily constructible bamboo footbridges for rural areas. Eng Struct 2017;143(Jul.):540–8. [9] Sharma B. Seismic Performance of Bamboo Structures. Doctoral dissertation, University of Pittsburgh, Pittsburgh, USA; 2010. [10] Sharma B, Harries K, Ghavami K. Methods of determining transverse mechanical properties of full-culm bamboo. Constr Build Mater 2013;38:627–37. [11] Richard M. Assesing the performance of bamboo structural components, presented at the Doctoral dissertation, Pittsburgh, USA: University of Pittsburgh; 2013. [12] Moreira L, Ghavami K. Limits states analysis for bamboo pin connections. Key Eng Mater 2012:3–12. [13] Ugarte J, Habusta M. Bamboo the vegetal miracle. Costa Rica: IAT; 2011. [14] Vélez S. Grow your own house. London: EMAP; 2013. [15] Salas D. Actualidad y futuro de la arquitectura de bambú en Colombia, Tesis de Doctorado, Universidad Politénica de Cataluña, Barcelona, España; 2006. [16] Harries KA, Sharma B, Richard M. Structural use of full culm bamboo: the path to standardization. Int J Archit Eng Constr 2012;1(2):66–75. [17] Camacho V, Páez I. Estudio de conexiones en guadua solicitadas a momento flector. Universidad Nacional de Colombia, Bogotá, Colombia; 2002. [18] Davies C. Bamboo connections. London: Bath University; 2008. [19] Hidalgo O. Manual de construcción con Bambú. Estudios Técnicos Colombianos Ltds. Construcción rural 1. Bogotá, Colombia; 1981. [20] Clavijo S, Trujillo J. Evaluación de uniones a tracción en guadua. Santafé de Bogotá, Trabajo de grado, Universidad Nacional de Colombia. Facultad de ingeniería. Departamento de Ingeniería civil., Bogotá, Colombia; 2000. [21] Lamus FA, Takeuchi C. Determinación de la rigidez de un tipo de conexión vigacolumna en Guadua angustifolia. Épsilon 2009:193–204. [22] Sassu M, De Falco A, Giresini L, Puppio ML. Structural solutions for low-cost bamboo frames: experimental tests and constructive assessments. Materials 2016;9(5):346. [23] Lamus Báez FA, Plazas Bernal MA, Luna Tamayo P. Resistencia de una conexión empernada solicitada a cizalladura doble paralela a la fibra para estructuras de guadua angustifolia. Tecnura 2014;19(43):52–62. [24] Seixas M, Ripper JLM, Ghavami K. Prefabricated bamboo structure and textile canvas pavilions. J Int Assoc Shell Spat Struct 2016;57(Sep.):179–88. [25] Albermani F, Goh GY, Chan SL. Lightweight bamboo double layer grid system. Eng Struct 2007;29:1499–506. [26] Widyowijatnoko A, Fitranto R, Intan R. Proposing Joints for Bamboo Tensegrity, presented at the 10th Wolrd Bamboo Congress, Damyang; 2015. [27] Minke G. Building with bamboo. Basel, Germany: Birkhauser Verlag AG; 2012. [28] Sharma B, Gatóo A, Bock M, Ramage M. Engineered bamboo for structural applications. Constr Build Mater 2015;81(Supplement C):66–73. [29] Sharma B, Bock M, Gatóo A, Mulligan H, Ramage M. Engineered bamboo: state of the art. Constr Mater 2015:57–67. [30] Xiao Y, Yang RZ, Shan B. Production, environmental impact and mechanical properties of glubam. Constr Build Mater 2013;44(Jul.):765–73. [31] Xiao Y, Shan B, Yang RZ, Li Z, Chen J. Glue Laminated Bamboo (GluBam) for Structural Applications. Materials and Joints in Timber Structures. 2014. p. 589–601. 504