Knit Happens
Designing the Mechanics of
Machine Knitting
Accurate modeling of the mechanics of knits is challenging due to their complex microstructure and the many interlocking and sliding elements. This is true for even the simplest homogeneous knits, and significantly more so with real-world fabrics that may be constructed from multiple layers of knits with heterogeneous patterns, which most current work doesn’t capture. In this collaboration with the Architected Intelligent Matter (A.I.M.) Lab at the University of Houston and the Self-Assembly Lab at MIT, I leveraged the advances in both approaches and demonstrated a mechanical understanding of CNC-knit fabrics through numerical modeling and characterization experiments.
We began by homogenizing the anisotropic mechanical behavior of various yarns and fabrics, focusing on three types commonly used in CNC knitting: PET, cotton, and nylon. Each yarn has a distinct microstructure—monofilament, twisted bundle, or extruded filament—that is characteristic of knitting and influences both its behavior and the overall macroscopic response of the fabric.
At the fabric swatch level, we identified three primary factors influencing mechanical behavior: material type, pattern, and stitch length. By systematically varying these parameters and testing swatches of each combination, we aimed to clarify the role each factor plays in shaping the fabric’s overall response.
The response is analyzed in terms of stiffness and anisotropy. Notably, the knitting pattern emerged as the primary contributor to anisotropy, as it alters the fabric’s symmetry. In contrast, material choice and stitch length had a more pronounced effect on stiffness.
To further understand knits and maximize CNC-knitting capabilities, we developed a Finite Element model to simulate their behavior, using 2x2 stitch unit to capture knit patterns. With periodic boundary conditions, this model accurately represents the mechanical response of a full homogeneous swatch with over 90% accuracy.
CNC-knitting machines can vary the studied parameters within a single fabric, enabling the creation of heterogeneous fabrics with spatially distributed functionalities. To effectively control the behavior of heterogeneous fabrics, we began by analyzing transition zones between homogeneous swatches, whose properties are already well understood. We found that these transitions have a negligible impact on the macroscopic properties of the fabric and can therefore be disregarded. Heterogeneous fabrics can then be modeled as a patchwork of homogeneous areas.
This understanding was applied to the design and optimization of a variable stiffness knit sleeve that exerts uniform pressure on the arm.
Using a 3D scan of a muscular arm and the previously studied knit parameters, we generated a heterogeneous knit pattern including all three key variations. This approach can be extended to design other bespoke wearables, including compression garments, high-tech space suits, or custom-fitted clothing.
The sleeve was designed using Rhino and Grasshopper, optimized using characterization data and MATLAB, and knitted using a Stoll CMS 330.
