AI-Designed TPMS Bone Scaffolds

ISEF Category: Biomedical Engineering

Subcategory: Biomaterials and Regenerative Medicine  ·  Difficulty: Advanced  ·  Setup: Home Setup  ·  Time: 1 to 2 Months

The Hook

Bone is not solid. Inside every long bone is a lattice of curved struts that hits an almost perfect balance between weight and strength. Engineers copy that pattern using math surfaces called gyroids. With a free Python library and a 3D printer, you can design lattices a human surgeon could actually use.

What Is It?

TPMS stands for triply periodic minimal surface. Gyroids and Schwarz primitives are examples. They are smooth, self-supporting shapes that repeat in three directions. Bones, butterfly wings, and some sea sponges all use TPMS-like patterns.

A Pareto front shows the trade-off between two things you cannot improve at once, like stiffness and porosity. By generating thousands of lattices in Python and computing each one's predicted stiffness, you can plot a clear front of the best designs.

You pick a handful of lattices on the front, print them on a low-cost resin printer, and crush them on a bathroom-scale load cell to validate. The bench data tells you whether the math front matches reality.

Why This Is a Good Topic

Generative design is a hot ISEF theme that you can run on a free Colab notebook. The math is approachable and the manufacturing step is realistic. You will learn computational geometry, optimization, and how to validate a model against a physical test.

Research Questions

• How does unit-cell size change predicted stiffness at fixed porosity?
• What is the effect of wall thickness on the measured Pareto front?
• Does the Python model predict measured compression force within 20 percent?
• To what extent does print orientation shift measured stiffness?
• Which TPMS family lands closest to published trabecular-bone modulus?
• How does post-cure time alter resin lattice stiffness?
• What is the effect of cell anisotropy on directional load response?

Basic Materials

• Resin 3D printer (sub-300-dollar consumer model).
• Standard resin and post-cure setup.
• Bathroom scale or 3D-printed load cell with strain gauge.
• Caliper for dimension checks.
• Linear stage for slow compression.
• Smartphone for time-lapse video.
• Laptop with Python.

Advanced Materials

• Engineering-grade photopolymer resin.
• Universal testing machine.
• Computed-tomography access for internal porosity.
• nTopology paid tier for advanced lattice editing.
• Calibrated load cell.

Software & Tools

• Python (NumPy and scikit-image): Generates TPMS lattices and converts to STL.
• nTopology free tier or MSLattice: Edits and slices lattices.
• Cura or Chitubox: Slices resin prints.
• Pareto front plotting in Matplotlib: Visualizes the design front.

Experiment Steps

1. Lock a single TPMS family for your first sweep and decide the porosity range.
2. Build a Python script that outputs both an STL and a predicted stiffness per design.
3. Decide your validation subset (for example, five points on the front) and replicate counts.
4. Calibrate the bathroom-scale load cell against known weights before testing.
5. Plan a print-orientation control so anisotropy does not contaminate the validation.
6. Compare measured stiffness to the predicted Pareto front and discuss deviations.

Common Pitfalls

• Treating a thin-walled lattice as if it would print cleanly when the resin can't resolve it.
• Skipping post-cure consistency, which changes stiffness sample-to-sample.
• Reporting one compression test per design and missing variance.
• Letting your Python prediction use a different cell size than the printed lattice.
• Plotting the Pareto front without normalizing for sample volume.

What Makes This Competitive

A competitive entry adds a published-data benchmark from trabecular bone, reports error bars on both predicted and measured stiffness, and validates at least five points along the Pareto front. Sensitivity analysis on the unit cell size and an anisotropy test push the project past a single-lattice demo.

Project Variations

• Replace gyroid with Schwarz primitive lattices and compare the fronts.
• Add a graded-density lattice and check whether the model predicts the gradient response.
• Print the same lattice in PLA and resin and compare absolute stiffness.

Learn More

• PubMed: Search gyroid bone scaffold Pareto reviews.
• NIH PubMed Central: Open-access papers on trabecular-bone mechanics.
• MIT OpenCourseWare: Course 3.054 Cellular Solids covers lattice mechanics.
• nTopology Learn portal: Free TPMS tutorials.
• Nature Scientific Reports: Open-access lattice design studies.