ITECH – Reclaimed Design
PROJECT TITLE
Reclaimed Design – An availability-oriented design methodology for reclaimed lumber
ARCHITECTURE AND RESEARCH TEAM
Clara Blum, Laura Marsillo, Gonzalo Muñoz Guerrero
LOCATION
Stuttgart University, Germany. Autodesk Technology Center, United States.
STATUS
Completed, 2024
Reclaimed Design explores an availability-oriented approach to structural design with reclaimed lumber, in which the characteristics of available material become active drivers of the design process. Rather than defining a structure first and sourcing standardized material afterwards, the project begins with an existing inventory of reclaimed elements characterized by variations in geometry, defects, and mechanical properties.
Developed as an integrated computational design-to-fabrication framework, the research combines material characterization, machine-learning-based strength prediction, equilibrium-based structural design, finite element analysis, and digital fabrication. The reclaimed lumber stock was characterized through non-destructive and destructive testing at multiple scales, and a machine-learning model was developed to predict localized compressive strength based on non-destructive testing values and defect location. This allowed each reclaimed element to be represented through a material gradient rather than as a homogeneous structural member. These predictions were incorporated into Rhino and Grasshopper as part of a digital material library, making material-specific information available directly to the structural design process.
The availability-oriented design system uses a step-by-step graphic statics process in which structural equilibrium and material allocation are resolved simultaneously. Rather than optimizing a predefined geometry and subsequently assigning material to it, member-specific constraints—including available length, grain direction, predicted strength, and geometric compatibility—inform the iterative generation of equilibrium geometry. When the available material cannot satisfy a structural request, the system can adapt its geometry and topology accordingly. Fabrication constraints, such as connection angles, joinery geometry, defects, and CNC manufacturing requirements, are also incorporated into this process, allowing material, structure, and fabrication to inform the design from the outset.
Karamba3D was incorporated into this workflow as soon as the system began generating viable structural configurations. Rather than treating finite element analysis as a final validation step, structural evaluation became part of an iterative feedback loop in which generated solutions could be continuously assessed and compared as the design space evolved. While the geometry itself was generated through a custom equilibrium-based method, Karamba3D provided an independent FEM environment directly within Grasshopper, allowing the team to systematically compare the axial forces predicted by their method with those obtained through finite element analysis. Structural evaluation could therefore inform the exploration and selection of generated solutions rather than serving only as verification of a final outcome.
The approach was demonstrated through the design and fabrication of a full-scale seven-meter truss. The system worked with a finite stock of 15 reclaimed timber members with varying cross-sections and predicted compressive strengths ranging from approximately 19 to 56 MPa, under a total applied load of 4 kN. From the same material inventory and structural requirements, the workflow generated 148 successful structural configurations, all of which were evaluated in Karamba3D. Across these solutions, the comparison showed an average discrepancy of approximately 0.1 kN between the axial forces predicted by our equilibrium-based method and the Karamba3D analysis, providing feedback on the reliability of the generative method while also informing the navigation of the emerging design space.
As material availability became more restrictive relative to structural demand, the system produced increasingly varied configurations while maintaining its underlying equilibrium logic. The comparison between the equilibrium-based predictions, material allocation, and Karamba3D FEM results became part of the criteria used to navigate these alternatives and select the configuration for fabrication. The selected truss consists of 15 reclaimed timber members, nine nodes, and seven triangular configurations. Its final geometry was therefore not the result of optimizing an abstract form and subsequently finding material to construct it, but emerged from the interaction between structural equilibrium, the actual dimensions and mechanical properties of the available reclaimed lumber, and fabrication constraints. Karamba3D provided the structural feedback necessary to evaluate this process throughout the exploration of the design space and support the transition from computational generation to the final built demonstrator.
Project Credits:
Research & Design:
Clara Blum Laura Marsillo Gonzalo Muñoz Guerrero
Thesis Advisors:
Tzu-Ying Chen Zuardin Akbar Jan Knippers Thomas Wortmann
Institutions:
Institute for Computational Design and Construction (ICD), University of Stuttgart Institute of Building Structures and Structural Design (ITKE), University of Stuttgart ITECH – Integrative Technologies and Architectural Design Research, University of Stuttgart
Full-Scale Demonstrator:
Autodesk Technology Center, Boston