ShellLab Form finding for concrete shells

Plan
Supports

Click nodes on the plan to add or remove supports. With the plan focused, arrow keys move and Space toggles.

Form
Lateral load

Carried by the gravity force densities, so the form leans. A shape study, not a seismic check.

Shell

Notes

  1. Form finding

    The plan is a pin-jointed grid in which each bar carries a force density q = N/L, negative in compression. With q fixed, equilibrium at the free nodes is linear in their coordinates (Schek, 1974), so x, y and z each come from one banded Cholesky solve. Perimeter bars take q times the edge tie, which stops unsupported edges from drawing in. The force densities are scaled once so the gravity form reaches the set rise, then reused for the lateral load.

  2. Loads and forces

    Self-weight is surface area × thickness × 25 kN/m³, shared equally between the nodes. The form does not depend on this scale, so a bar force is its unit-weight value times the node weight. Reactions balance the free-node loads, and the residual shown is the largest out-of-balance nodal force over the largest nodal load. The stress is a bar force over its strip of shell: a guide, not a design value.

  3. Carbon

    A1–A3 carbon is concrete volume × 340 kgCO₂e/m³ for C40, taken from ShellBench v1.0.0 (GB/T 51366-2019, Table D.0.1), with ShellForm’s ±25 % band. Reinforcement is left out, so the figure is a lower bound.

  4. Scope

    This finds forms; it does not design shells. Thickness is not sized, and buckling, bending under non-funicular loads, reinforcement and supports are not checked. Sized and checked designs are in the ShellForm explorer.

  5. Verification

    Six square cases, run through ShellForm v0.2.0 itself, are the test fixtures. The browser solver matches their coordinates to 1 × 10−13 m and keeps every form in equilibrium to a relative residual below 10−9. The tests run on each push.