PerfoStruct, a method for designing lightweight engineered wood structures based on a planar material perforation strategy that uses topology optimization and quadtree subdivision.
Context
Developments in digital design, structurally engineered-wood and fabrication technologies as well as trends towards low-carbon construction have reinforced architects’ and engineers’ interest in the new possibilities of lightweight wood structures.
Design Goals
By introducing topology optimization for structural composite lumber we intend to further the research into lightweight wood construction. The goals of this research are:
- To improve the performance to weight ratio of laminated strand lumber (LSL) for a structural beam application in architecture and construction
- To design a method for material removal based on a perforation strategy that uses topology optimization and quadtree subdivision
- To reduce weight, increase performance, and provide design variability for architects and engineers.
Topology Optimization
We introduce a design strategy for structural elements based on topology optimization. We use discretization and planar subdivision to generate perforations responding to principal stresses. We hypothesize this method to increase the performance of an equally weighted beam when perforated according to standardized techniques.
Finite Element Analysis
In this study, we leveraged the benefits of surface finite element analysis (FEA) in Millipede, a Grasshopper tool, and volumetric FEA of linear-elastic behavior in COMSOL Multiphysics. Von Mises stresses and displacements at a parametric applied load of 400 Newtons are calculated to evaluate the PerfoStruct method.
Mechanical Testing
Physical tests were performed in conjunction with the planar and volumetric FEA simulations to evaluate mechanical properties of the various beam designs beyond the linear-elastic region. Each sample was fabricated using a laser cutter and mechanically tested with a three-point bending test on an Instron. The accuracy of the two FEA methods are discussed by comparing it to scale model physical testing.
Artifact Design
We remove material from each slice of the beam by creating a single hole inside each subsurface of the partitioned slice. The form of the perforation is based on a parametrized geometry and depends on two factors: i) the mean density of the subsurface, and ii) the direction and magnitude of the first component of the mean principal stress of the subsurface.
.
Fabrication
A beam measuring 1.8m by 0.15m (length x height) made of five sheets of 0.03m (0.15m total depth) structurally perforated LSL was fabricated as a full scale proof of concept of the structurally optimized perforation system for lightweight wood construction. The beam was made through CNC milling.
Full-scale artifact
We showed that PerfoStruct has the ability to produce lightweight and of high structural performance designs for beams that are easy to fabricate using off-the-shelf materials and tools. An overall volume reduction of 40% was achieved and the CNC fabrication methodology was validated.
Findings
Our results suggest that this form finding method creates beams of at least equal performance with those based on standardized methods, while introducing shape variety and design flexibility. By controlling the shapes distribution and size we gain control over the failure mode. We found that PerfoStruct enables us to introduce gradual failure in place of acute failure which can have applications for seismic design and design in other fields where advanced warnings of structural failure are desired.
Credits
Spyridon Ampanavos (topology optimization, surface discretization algorithm, literature review)
Nicole Bakker (COMSOL simulations, fabrication small-scale beams, mechanical testing, literature review)
Peter Osborne (topology optimization, fabrication full-scale artifact,
literature review)
Acknowledgements
This research was conducted as part of SCI 6359: Interface Design: Integrating Material Perceptions taught by Professor Sawako Kaijima at the Harvard Graduate School of Design, Spring 2018.
Special thanks to Andreas Haggerty and Elaine Kristant from the Harvard SEAS Active Learning Labs for providing access to the Instron machine and COMSOL software.
Awards
This project was awarded the Harvard Graduate School of Design Peter Rice Prize for Excellence in Structural Design in 2018.