This thesis addresses the challenge of scaling 3D printed (3DP) glass in architecture by enabling reversible, dry-assembled systems that eliminate adhesives and support circular construction. It builds on prior research (TU Delft and MIT) in interlocking glass units, which require interlayers to manage tolerances, transfer loads, and prevent contact that could induce tensile stress.

A novel surrogate-assisted optimization workflow is developed to define target interlayer properties for specific geometries and load conditions. This material-agnostic method enables flexible selection of interlayer materials based on availability and performance criteria. Further, a kirigami-inspired aluminum interlayer is explored in this study for its tunability, ease of fabrication, scalability, and transport efficiency.

Three parametric structures- a freestanding wall, a catenary vault, and a form-found compression shell were modeled and tessellated based on 3D printed glass constraints. The optimization workflow, combining Grasshopper, ANSYS, PyAnsys, and OptiSLang, is demonstrated through the catenary vault. Finite element simulations confirmed the structural performance of the assembly and strategies to stabilize a fully glass interlocking vault, while compression testing of fabricated kirigami interlayers revealed key areas for improving the surrogate model accuracy. Overall, the workflow enables a flexible, performance-driven approach to interlayer design and material selection in interlocking glass vault assemblies