Within architecture, wood products must be stable, predictable and consistent construction materials, typically resulting in flat or straight members. While upholding these construction values, this ongoing research offers an alternative laminated-timber approach that produces specific curved forms without the use of any formwork. Variable laminated timer (VLT) adjusts the grain direction (GD) and moisture content (MC) of “active layers” within the lamination assembly. Combined with the number and placement of active laminated timber layers (LL), the three variables generate a range of possible curvatures from a flat pressed lamination. This work aims to bypass needed formwork or moulds, like the ones used by Charles and Ray Eames and incorporate the anisotropic structure of wood, as well as the hydroscopic characteristics of wood as design variables to achieve specific curvatures. Unlike other successful curved laminated timber assemblies, VLT utilizes one species of dimensional lumber in the assembly for a more accessible transition from flat to curved wooden panels.

The subjects confirmed the direction of the resulting curvature is corresponds directly to the rotated GD angle in the lamination assembly. (fig. 02) By misaligning and angling the active layers, the swelling, stretching and shrinking of T is no longer constrained by the neighboring perpendicular stiffness of L and therefore can effect the global shape of the lamination. During the lamination process, layers are glued and pressed together and the glue increases the MC of both wood layers. As the laminations are released, the MC of the wood decreases as it dries and returns to the average of 10%, causing the wood in the lamination to effectively shrink. The most prominent and unconstrained shrinkage is along the T axis of the lamination layer with the rotated GD. The result is a global, self-curving process that conforms to a cylindrical monoclastic surface along the L axis of the active lamination layer with the rotated GD. Beyond the self-curving process of a same-species bilayer, the small-scale prototypes confirmed the GD angle is a reliable variable that determines the direction of the finished lamination’s monoclastic curved form. Two laminations of the same size and species, assembled on a flat surface, can have drastically different monoclastic forms according the angle of grain direction in the assembly (fig. 02).

Further, the subjects responded generally as expected to varying increased levels of MC. With higher MC levels during lamination, the subjects generally had a higher degrees of curvature after drying. However, there were some inconsistencies, including higher MC levels with less curvature. The relationship between MC and degrees of curvature requires either more testing or a different strategy for adjusting the MC. The added moisture amount was tested and determined as 10%, 12.5%, 15%, 17.5% and 20% of the to-be-active layer’s weight, and then sprayed onto the active layer.

VLT Samples:

This self-curving principle was tested with large scale samples using a series 10mm x 130mm profile western pine dimensional lumber so to simulate the industrial CLT process of laminating dimensional lumber. In addition to confirming the direction of the monoclastic curvature by altering the GD angle of the active lamination layers, the samples also aimed to produce differing degrees of curvature by altering the MC of the active layers. The thought and hope was the subjects were too small and thin to accurately test the MC, were larger wood elements like dimensional lumber would be saturated and respond more consistently. In order to test the relationship between MC and degrees of curvature, a series of different MC levels were obtained in active layers before lamination: 10% (average MC), 15% and 20%. Unlike the subjects, the active layers were soaked until the desired MC was reached, and once saturated, the elements were laminated with a non-active layer that had a normal MC of 10%. The resulting laminations (fig. 05) confirmed both the directional control of monoclastic curvature and the degree control of curvature through managing the MC of the active layer. Expectedly, the most extreme degrees of curvature resulted from the higher MC of 20%.

However, while the general degree of curvature was consistently related to the MC of the saturated rotated layer, the monoclastic curvature did not follow the shape of a cylinder but rather a cone, with some variation in-between. Further investigation is needed to explain the conic irregularity in the resulting curvature. It should also be noted that small gaps between lumber were noticed in the VLT with a saturated MC of 20%, meaning there are some structural compromises in favor of curved form for these particular VLT panels. Although curved surfaces do offer a more robust section modules, it is imperative all members are laminated together to function structurally.