Tunnel Construction Simulation
An open simulator for mechanised tunnelling: published analytical models in the browser, GPU ground physics on the server.
Urban EPB shield (7 m class)
A 7 m EPB driven through real London ground: the Farringdon – Liverpool Street corridor from open BGS boreholes, with the 60 real buildings over the line. Chamber pressure held inside the DAUB window by the screw conveyor, foam conditioning, thrust and torque breakdown, ring building with Crossrail's 1.6 m ring of 7 segments + key, and surface settlement over the buildings, with the second bore. Shift pattern, downtime and tool interventions give a 90-day advance forecast.
Checked against: thrust and torque against four open field datasets, with torque in conditioned clay from the muck's measured shear strength, 0.98–1.14 against Sofia (EPB field data); the support window against the DAUB recommendations, reproducing their worked example (face support); the settlement trough reproduces Hyde Park, 8.1 against 8.3 mm, the second bore follows it, and volume loss rises when face pressure drops in London Clay (settlement); the chamber pressure gradient against the Botlek bulkhead, and the screw pressure calibrated on its screw gauges; the ring cycle and advance against the Crossrail drives, 11.8–12.5 against 12.3–14.9 rings per working day; best days from a day-to-day variation fitted on Crossrail and tested on Barcelona Line 9, whose average day the model still runs about 30 % slow (advance).
Urban slurry shield (Mixshield)
A 7.08 m Crossrail-class Mixshield on the Crossrail Thames Tunnel, Plumstead to North Woolwich, in real ground from open BGS boreholes (Chalk, Thanet Sand, the river above): submerged wall and air cushion, feed and discharge lines with a crusher at the suction, filter cake or slurry penetration at the face, and the dry-mass balance and deposit velocity in the circuit.
Checked against: settlement of both bores fitted to 17 measured sections of this drive, median ratio 1.10, with the volume loss consistent at an independent chalk site north of the river (settlement); the support window under the river, 3.0–3.6 bar, against the more than 3 bar Crossrail needed, with DAUB's factors and ±10 kPa deviation checked; rings per working day against this drive, 10.7–12.5 against 7.9–10.0, and the whole drive on two Dutch slurry tunnels, 0.78–0.98 of measured (Thames advance).
Ground physics on the GPU: tunnel face in sand
A material-point (MPM) model of a 5 m tunnel face in dry sand, recorded on the GPU server and played in the EPB demo's "Ground physics" panel: the face holds while the support pressure is above the limit, gives way when it drops below, with the failure chimney rising to the surface, and stops when the pressure is raised again. Run locally, the same panel takes the support pressure live.
Checked against: uncalibrated, reported as a bracket: Idinger et al.'s centrifuge tests measured a limit of 4.5–6 kPa at this cover, and the MPM gives 4.0–8.7 kPa depending on how the soil's friction cone is matched; Vermeer's analytical value is 8.4 kPa (MPM face in sand). The clay face and the EPB chamber are not shown: both are blocked on a solver limitation (clay, chamber). The earlier, unchecked Phase 0 benchmarks remain available.
Hard-rock open gripper TBM
A 9.58 m main-beam TBM with the Gotthard Base Tunnel machine data: disc cutterhead with buckets, grippers and the regrip cycle, rock bolts, mesh and ring beams installed by rock class, shotcrete on the backup, and a fault zone that slows the drive.
Checked against: machine data from Herrenknecht and the AlpTransit archive; disc-cutter penetration fitted on the Follo Line logs and tested, not refitted, on the Queens Water Tunnel (median 1.02); working-day advance bracketed by Gotthard's 8–10 m per calendar day (hard-rock projects).
Drill & blast road tunnel
A Norwegian T10.5 road tunnel driven round by round: the drilling pattern from published rules, a three-boom jumbo, charging, blast, ventilation, scaling, loader and trucks, bolts and sprayed concrete by support class, downtime, and vibration at a house above.
Checked against: hole counts against current Norwegian rounds; the round cycle against logged rounds at the Ulvin tunnel and a contractor's T10.5 estimate, 12.7 h against NFF's "about 12 h"; explosive use about a third below 353 logged Norwegian rounds of 2021–24, shown in the demo (drill & blast cycle).
SEM / NATM heading
The Crossrail Farringdon platform tunnel in real London ground: an 11 m sprayed-concrete enlargement of a TBM pilot in Lambeth Group and London Clay, undrained ground response, ring closure behind the face, and spiles in a water-bearing sand lens. A weak-rock heading with a fault zone stays selectable.
Checked against: predicted crown settlement, 10.4–13.7 mm, against the measured 10.3–16.5 mm and Crossrail's own 10 mm design analysis; the ground response against measured convergence in a weak-rock tunnel (SEM convergence).
Hard-rock double shield TBM
A 9.9 m double shield with the Follo Line machine data: telescopic and gripper shields, segment rings built while boring with pea-gravel backfill, and single-shield mode through the fault zone.
Checked against: Follo Line TBM1: penetration 6.4 against 6.32 mm/rev, within about ±27 % section by section; Queens Water Tunnel penetration, not refitted, median 1.02; advance 13–15 against 14.1 m a day (hard-rock projects).