Skip to content

Release 1 (v1.0.0)

Latest
Latest

Choose a tag to compare

View all tags
@dtt-github dtt-github released this 06 Nov 22:25
· 4 commits to main since this release
v1.0.0
4ec45d2

The first complete release of the open tunnel construction simulator: six browser demos, each checked against open published data, plus one recorded GPU ground-physics run.

▶ Open the live demo — runs in the browser, no install, no GPU needed.

What's in it

Demo Checked against
Urban EPB shield, Crossrail Farringdon – Liverpool Street Thrust and torque on four open field datasets; Hyde Park settlement; DAUB face support; Botlek chamber and screw pressures; Crossrail and Barcelona advance
Urban slurry shield (Mixshield), Crossrail Thames Tunnel Settlement on 17 measured sections (median 1.10) and three independent chalk sites; Thames and two Dutch slurry drives
Hard-rock open gripper TBM Disc-cutter penetration fitted on Follo, tested on the Queens Water Tunnel (median 1.02); Gotthard rates
Hard-rock double shield TBM Follo advance; Guadarrama's share of time boring; thrust by mode from the Brenner machines
Drill & blast road tunnel, Norwegian T10.5 A logged Ulvin round; a contractor's T10.5 estimate; 353 logged rounds for explosive use
SEM / NATM heading, Crossrail Farringdon Crown settlement 10.4–13.7 mm against 10.3–16.5 mm measured; convergence time law against the Fréjus tunnel
GPU ground physics (NVIDIA Newton MPM), in the EPB demo A recorded dry-sand tunnel face, bracketed by centrifuge tests

Behind them: a model library with a Python reference and a JavaScript twin (14 models, 94 + 21 tests on shared vectors), machine-class presets with a source on every field, and ground scenarios built from open BGS boreholes.

How to read the results

Every check is labelled independent, calibrated or bracket, and the open gaps are ranked, in the validation report. Known gaps ship disclosed, among them: hard-rock cutterhead torque is 1.25–2.4× the logged value on two projects; downtime rests on one EPB; drill & blast explosive use is about a third below logged practice.

Each demo has a realism checklist, and sources, labels and licences were audited for this release (audit).

Not in this release

  • The MPM clay tunnel and EPB chamber on the GPU (newton-physics/newton#5180).
    • Correction (7 Nov 2026): this release's notes and docs said the clay tunnel was too strong under GIMP integration. That came from a scaling error in our code (cohesion 2.744× too high), not from the solver. With it fixed and a smaller initial particle jitter, the clay tunnel reaches 0.92–0.98 of Kimura & Mair's stability ratio on main; see the changelog. The published demos are not affected: none uses the GPU clay or chamber, and the recorded sand face predates the GIMP switch.
    • Second correction (7 Nov 2026): the EPB chamber result we gave after the release (under GIMP it keeps 162 of 207 kPa while the cutterhead turns) was read after a 4 s excavation. Over 20 s the pressure falls to 51 kPa: the paste loses its pressure while the cutterhead shears it, under every solver setting we tried, so the chamber stays blocked (MPM chamber).
  • Single shield, inner linings, logistics and the further methods of Phases 3 to 5 (roadmap).

Get started

git clone --branch v1.0.0 https://github.com/dtt-github/tunnel-construction-simulation.git
cd tunnel-construction-simulation
npm run serve

Full list of changes: CHANGELOG. Code under Apache-2.0; third-party data keep their own licences (NOTICE).

We especially want measured data — TBM logs, settlement monitoring, face-pressure records, cutter wear. See CONTRIBUTING.

Assets 2
Loading