The construction principle of zip-bent structures relies on a set of geometrical parameters and their mutual relations, as outlined in Fig. 3.
The construction of zip-bent elements starts from a developable reference surface, i.e., a surface with Gaussian curvature equal to 0 capable of being unrolled on a flat plane without distortions . This is generated through a series of parallel or non-parallel rulings. Each ruling represents a shared edge between two faces. Two main angles are identified as parameters for generating the reference surface. The bending angle (α) is the supplementary angle between two consecutive faces. When the bending angle of consecutive rulings is constant, the resulting developable surface can be described with the geodesics of a cylinder, which is the geometrical space explored in this paper. If the bending angle varies, the resulting developable surfaces are poly-cylindrical. The ruling angle (β) represents the angle between the specific ruling and its adjacent edge of a face, which defines the second direction of the plane that they both lie on. Variations of the ruling angles result in the twisting of the global form. If β is constant, then all the rulings are parallel and the resulting surfaces are cylindrical, while when β varies, the resulting surfaces are conical or poly-conical. The bending radius (r) of a zip-bent element is measured to quantify the bending capacity between a developable surface’s initial and final plane. It is used to measure the overall bending curvature of the zip-bent element.
Individual zip teeth are constructed geometrically, starting from each ruling. Each tooth is characterized by a zip tooth height (ht) and a zip tooth angle (γ). The negative space created between the two zip teeth is called a zip pocket. As a final zip-bent element comprises the two joint elements, the sum of the wood thickness (tw) and one veneer thickness provides the total thickness (tt).
Zip-bending as a material system
Zip-bending is a material system composed of two elastic layers such as wood and an intermediate adhesive layer, which binds the various parts into a unique element transferring forces and loads.
In this study, zip-bending experiments are conducted on battens obtained from engineered wood products with good dimensional stability, ensuring higher assembly precision than solid wood. The employed elements are T grade LVL (Laminated Veneer Lumber) spruce battens 39 × 95 × 3000 mm, with a total of 13 layers of 3 mm thick veneer glued together with parallel fibres to the main axis. The adhesive used to bond the zip-bent components is a polyurethane-based (PUR) wood glue with a curing time of about four hours.
The bending capacity of wood is determined by the material thickness, moisture content, and temperature [20, 21]. Wood flexibility can be estimated with the modulus of elasticity which, given the organic nature of the matabout four hours of curing time the moisture content and temperature [22,23,24]. Specifically, the increase of moisture and temperature in wood determines a decrease of the elastic modulus making the material more flexible, as seen in both industrial and crafting steam-bending techniques.
In the case of zip-bent components, the bending capacity of an element depends on the thickness of the top/bottom veneer layers. Previous studies used a layer thickness of 2–3 mm layer thickness to balance material flexibility and final strength [8,9,10,11,12]. This work employs a 6 mm thick veneer layer to enhance the post-assembly structural performance. This leads to stiffer elements that require thermal and moisture alteration during the production process to achieve considerable bending.
Different moisturizing methods have been compared to tune the flexibility of the zipped layers and prevent cracking during the assembly. An empirical test was performed, where a 6 mm thick 500 mm long LVL element was moisturized with three methods: soaking in water (10 min), steaming (10 min), and soaking and steaming (5 + 5). The elements were tested in a three-point bending with a constant load, where the midspan deflection was measured to compare the flexibility achieved with the three methods. The results highlighted that a combination of soaking and steaming increased the material flexibility more consistently than the other methods.
The achievable curvature radius of LVL zipped battens have been analyzed after the moisturizing process. Two different veneer thicknesses have been tested, 3 mm and 6 mm, on zipped battens that measured 39 × 95 × 1000 mm. The experiment compared the achievable curvature of zipped specimens with different veneer thicknesses and a constant moisture content MC = 18%. The specimens were elastically bent and the minimum curvature radius was measured from the geometrical construction of the circle tangent to the elastic curve formed by each specimen. The results have been expressed in curvature radius r and bending capacity r/tw, where tw is the total wood thickness (39 mm). The 3 mm veneer achieves a curvature radius of r = 260 mm when bending with zip teeth outwards and r = 280 mm in the opposite direction; the bending to thickness ratio is r/tw = 6.66 in the first case and r/tw = 7.14 in the second one. In the case of the 6 mm veneer the minimum bending radius is r = 435 mm with r/tw = 11.35 with outwards zip teeth and r = 570 mm, with r/tw = 14.61 in the opposite direction of bending (Fig. 4).
Applying the last result, insofar as the most conservative one, to a zip-bent batten made up of two layers - one top and one bottom - the total thickness of the component is tt = 39 + 6 = 45 mm and, therefore, the bending capacity obtained, measured on the minimum curvature radius r = 570 mm, is r/tt = 12.66. In the literature, the value achieved is r/tt = 10 for solid oak wood panels with r = 250 mm, tt = 20 mm and a veneer thickness of 2 mm .
Determining the basic mechanical properties of the zipped LVL material is fundamental for a comprehensive study on the structural applications of zip-bending. In this work, a simplified test series was conducted, including:
Compression test parallel to the fibre direction;
A tensile test parallel to the fibre direction;
Three-point bending test perpendicular to the fibre direction.
Each test involved at least three specimens of both solid and zipped wood with a cross-section of 39 × 95 mm and 45 × 95 mm respectively. The acquired data is further used to verify and calibrate the Finite Element Method (FEM) simulation in Abaqus.
Four zipped wood specimens and three solid wood specimens with a length of 150 mm were tested. A compression test machine ToniPACT II with a 3000 kN load cell was used. All the experimental specimens were preloaded with 5 MPa pressure. The resultant curves demonstrate that the zipped specimens reached the strength of about 20 MPa, which is roughly half the strength of the solid specimens. In the case of the solid wood test, it is observed that all three specimens have almost the same elastic modulus with a perfectly corresponding linear segment in the stress-strain curve. However, a considerable gap is observed among the specimens in the necking phase, due to the non-uniform distribution of the wood defects within the individual LVL layers (Fig. 5).
In all the zipped tested specimens, a similar failure mechanism was observed - firstly, the outer layer is damaged, and afterwards, the failure develops towards the inner layers, resulting in their separation and individual bending or breaking (Fig. 6).
Experimental results show that the mechanical behaviour of the zipped elements depends on the type of glue, the quality of glueing application, and the correct curing process. The measured mean compressive strength was 21.03 MPa for the zipped wood specimens and 39.16 MPa for the solid ones, while the measured mean elastic modulus was 0.8 GPa and 6 GPa respectively.
Four zipped specimens and three solid wood specimens have been tested and compared. Two M16 steel bolts were used to facilitate the clamping of dog-bone specimens in a Zwick/Roell Z1200 testing machine with a 1200 kN load cell (Fig. 7). The measured resistance of the zipped specimens resulted in about 25% of the solid ones, where the mean tensile strength for the solid wood was calculated to be 42.48 MPa and 9.56 MPa for the zipped wood. An elastomer-like behavior is particularly observed in the specimens TZ-P-02 to TZ-P-04, which shows that the glue enters into action earlier than wood. However, wood showed to be dominant and had a higher elastic modulus for the specimen TZ-P-01, but it eventually failed after a slight deformation (Fig. 8).
Three-point bending tests were performed over three zipped wood specimens and three solid wood specimens of 800 mm length, with a span between the supports set to 576 mm. All flexural specimens were tested by a Zwick/Roell Z050 with a 50 kN load cell (Fig. 9).
The resulting flexural strength of the zipped wood specimens greatly varies due to inconsistent tolerances between the glued parts. This led to a high instantaneous release of energy during the applied load, where the glue between the zip teeth is immediately damaged and makes the specimen unstable. However, when the glue and wood were tightly bonded, the flexural strength of the zipped specimens reached 50% of the solid ones. To calculate the flexural strength of the zipped beams, the glue strength is neglected and it is assumed that each part of the zipped elements operates separately. Therefore, the flexural strength is calculated elastically using the following equation.
applied load of the hydraulic jack
length of the beam span
the distance of farthest edge from the neutral axis
moment of inertia for each zipper at the middle of the span
A flexural strength of around 50 MPa was found in the solid wood tests. Figure 10 shows that when the bottom veneer layers fail because of material inconsistencies, the specimen loses around 50% of its flexural strength.