3.1. Experiment Overview
Twelve connector specimens were evaluated for bending strength under four configurations: with and without structural adhesive (polyurethane resin) and with the connector’s strong and weak grain directions aligned to the CLT’s principal axis (
Figure 17). Each specimen was seated on a bearing plate and braced with a centrally mounted pantograph to eliminate initial deflection. Four-point loading was applied via a 2 000 kN testing machine at a monotonic displacement rate of 5 mm/min until either a peak load was reached or 50 mm of displacement occurred. Specimen deflection was recorded by a midspan displacement gauge.
Timber fitting connectors marked [ ’’ ] are oriented parallel to the CLT strong-axis direction; absence of [ ’’ ] denotes orientation along the weak axis. [ Bond ] indicates the application of structural adhesive. The test results confirm P max=12.0 kN (average peak load ), and a long term allowable bending strength Ma=1.54 kN·m (
Figure 18 and
Figure 19). Since the lamina side faces were uncrimped, shear failure of the fitting connector snagging area occurred after rolling shear occurred. Resorcinol-phenol co-condensation resin was used as the structural adhesive joining the laminae. Although a structural adhesive was used, only stiffness increased without any increase in bending capacity. There was also a unit that suffered shear failure of the timber fitting connector after flexural failure of the shear key. The stiffness of the structural adhesive was higher than the stiffness of the timber fitting connector, so there no combined proof stress at the end of the test. Since cracking occurred only in the weak axial direction of the CLT layer, the bearing capacity would have increased if the fittings were in the strong axial direction (
Figures 20). Notably, seating the connector on a strong-axis lamina would markedly increase bending capacity. Unlike through-bolts or long screws—which create continuous load paths and require no layer-by-layer orientation fitting connectors in CLT concentrate stresses at each lamina. Therefore, connector design must be studied per layer to avoid weak-axis alignment at the interface. Moreover, for buildings like this one with predominantly unidirectional loading, it may be more rational to adopt a glulam-style layup—stacking laminations all in the fiber direction—instead of cross-lamination, thereby eliminating weak-axis layers and simplifying connector requirements.
Table 2.
Assessment Results.
Table 2.
Assessment Results.
3.3. CLT Cost-Reduction Strategy and Carpentry Execution
At the time, CLT was a novel construction method, and contractors were hesitant to procure materials directly without intermediaries. As the client, I assumed the contractor role—purchasing CLT panels and fittings directly from the manufacturer—and performed the carpentry work myself. By separating the CLT works contract from the main building contract and eliminating intermediary fees, CLT framing costs fell to approximately about $200/m², roughly one-third of the original estimate. This reduction was achieved despite post-disaster price inflation following the Kumamoto earthquakes, which had driven up material and labor rates.
All CLT arch modules and superimposed open-work beams were prefabricated at the factory (
Figure 21 and
Figure 22). Arch CLT (t=150 mm, S60 cedar) and joint jigs were CNC-cut from 3.0m × 12.0m × 0.15 m “mother boards” using DXF layouts provided by the design team, achieving just 6 % waste (20.28 m³ raw to 19.01 m³ installed). Connector mortises were undersized by 1.0 mm to account for post-adhesive swelling; mock-ups confirmed that light “kigoroshi” hammering collapsed cell walls to ensure tight fit and improved glue penetration.
On-site, a crew of five carpenters completed erection in a single day on a narrow lot crisscrossed by overhead power lines. Modules measured 2.75–2.85 m in width and weighed 1.5 t each. A few carpenters guided each arch into place using a nylon sling and hydraulic jacks, then secured interfaces with long screws. Lightweight CLT enabled manual positioning when necessary (
Figure 23).
I organized a team of local friends and former collaborators to undertake every phase of the project—from well drilling through carpentry to interior finishing—allocating each task to match individual expertise (
Figure 24). By eliminating hierarchies and embracing a flat, collaborative structure, we achieved a creative disaster recovery that honored traditional craftsmanship and maximized efficiency.
Figure 24.
Project Roles: Traditional vs. Author-Led Reconstruction.
Figure 24.
Project Roles: Traditional vs. Author-Led Reconstruction.
Figure 25.
Public Bath Interior and Street-Facing Exterior (Shigeo Ogawa).
Figure 25.
Public Bath Interior and Street-Facing Exterior (Shigeo Ogawa).