Microstructural Manipulations

This research project aimed at developing a computational design process based on a continuous information flow from scanning the individual anatomical features of wood pieces to the eventual digital fabrication. The project consists of connected vertical veneer strips that are assembled as an undulated wall element. The main fiber direction of the quarter sawn pine veneer is oriented parallel to the long axis of the vertical strips. As a consequence, the significantly lower stiffness perpendicular to the grain allows forming horizontal curvature in each strip. This local curvature together with the larger undulation of the larger system contributes to the significant compressive strength of this extremely thin walled construction. The structural behavior and internal forces were analyzed using a special finite element analysis application for anisotropic materials. The following development phase was based on the hypothesis that the majority of the material’s compressive strength can be contributed to the thick and dense latewood cells. Thus, proportionally to the internal force intensity, parts of earlywood cells can be removed without having a major effect on the overall structural capacity.

Latewood features a much denser cell arrangement with considerably thicker cell walls as compared to the relatively fast growing early wood. The visible grain of wood, in this case the aforementioned quarter sawn Pine veneer, results from this anatomical difference between latewood and earlywood cells. Hence an optical scanning process was developed on which the following fully integrated computational design and fabrication process is based.

First, a Finite Element Analysis off the overall system is conducted and the strain and stress intensity is mapped on a dense registration mesh. At the same time, the elements from which the overall system will be assembled are computationally defined and referenced to the actual veneer elements intended for construction. Once each veneer element has received an ID tag that identifies its location in the final assembly, the pieces are scanned on a purpose-built optical scanner. Based on pre-calibrated threshold parameters an algorithmic procedure converts the high resolution pixel data into vector information that separates the earlywood form the latewood regions. In a subsequent step, the computational process correlates this data with the registration mesh of the structural analysis and derives a specific cutting pattern for each sublocation, whereby the stress intensity defines the cut-outs’ size and density. In a last step the resultant cutting patterns are formatted for immediate use with a precision laser cutter, which can then be employed to remove and directly recycle the dispensable early wood. In this way, digital fabrication responds directly to the unique anatomy of each wood piece.

The entire information flow and algorithmic processing is fully automated and operates with a high degree of precision. First tests have indicated that the remarkable patterns of removed and directly recycled cutouts significantly reduce the material’s mass but only have a marginal effect on the load bearing capacity. More importantly, it begins to show how digital fabrication can now be conceived of as fully embedded in an information feedback from material scan to computational design, analysis and manufacturing.

Performative Wood Studio (Achim Menges)
Jose Ahedo, Harvard University Graduate School of Design, 2009