2017–2025 / Cambridge
Eight years of building and learning.
From my BA and MEng to a PhD on non-rigid linkages, Cambridge is where I developed the research behind the structures on this site.
Explore the research ↗Founding engineer · Beyond Reach Labs
I design deployable structures for space, and research how one structure can hold several stable shapes.
01 / Background
I’m a structural and mechanical engineer. I did my PhD at Cambridge and worked on Starship and Starlink at SpaceX. Now I design deployable structures for space at Beyond Reach Labs in New York.
Deployable space structures, mechanisms and linkages
Abaqus FEA, Python, MATLAB, C++
NX, SolidWorks, Fusion, Illustrator
A few points along the way
What I’ve worked on, and what I do outside work.
Select a star to explore. Smaller stars belong to the experience they orbit.
2017–2025 / Cambridge
From my BA and MEng to a PhD on non-rigid linkages, Cambridge is where I developed the research behind the structures on this site.
Explore the research ↗02 / Research
Some structures have more than one shape they hold on their own, with no motor keeping them there. This is multistability. It lets us build things that fold for transport, deploy for use or adapt to a new task.
My PhD asks how to design those shapes and the motion between them. I start with geometry: which shapes do the bar lengths and joints allow? Elasticity then shows where the structure can settle. The papers below turn this into a design method for linked frames.
Between geometry and motion
Explore the mechanisms behind the papers. Drag a bright joint along its compatible motion, choose a stable shape, or see how two frames share a connection.
Read the explanation ↓Part I, case (b): six zero-energy shapes on the sampled motion path.
Arrow keys explore; Enter settles.Drag the bright joint, or tap a shape. Geometric motion, not a force-response simulation.
Frame A (white) and frame B (teal) share the highlighted rigid edge. Frame B is uniformly scaled to match that connection. Move frame A above, or change frame B below.
Each frame keeps its own elastic link. This ideal geometric assembly does not check collisions or model actuation between disconnected branches.
Start with a rigid frame: in the ideal model, its bars allow no change of shape. Remove one bar and the linkage can move. Put a spring in its place, and the motion gains stable stopping points: here, six shapes where the spring is back at its natural length. Each is a stable state, a shape the frame holds on its own.
Force-controlled model: a small disturbance returns; a larger load switches states. Based on Part I’s energy formulation, with illustrative overdamped settling.
Worked example · Part I, case (b)
One elastic bar · six stable shapesThe energy plot shows how much energy the spring stores. The shape map follows the same movement using three measurements of the frame, rather than its position in space. Pick a numbered shape, here or on either plot, to explore it; both markers follow the motion. You can also drag the mechanism in the corner.You can also drag the mechanism above.
A larger structure can be built from these smaller frames. When the connections let each frame keep its own stable shapes, their combinations multiply. The idealised chain below adds one frame at a time: 6 × 6 × 6 × 2 = 432 possible stable combinations. In a physical design, some may be blocked by parts touching.
Building multi-stable structures based on Baranov truss, a three-part series with Keith Seffen in Proceedings of the Royal Society A. Open one for a short summary.
04 / Thoughts
Passages I’ve saved and reposted on Substack, newest first.
05 / Contact
Building something that has to fold, deploy or hold more than one shape? I’d like to hear about it.
Charles Gai
Light, places, and passing moments.
01 / 06Stable state
closure 4.702 · member 8.124 · energy 0.00