Tunnels

A tunnel is a structure wrapped around a carriageway. In yucad3d it is a bore laid along an alignment: a sampled centre line, a cross-section, a lining, and the clearance envelope the bore has to contain. The road or railway running through it stays what it already is — an alignment with a corridor — and the tunnel wraps it.

Creating one

Tunnel › Create Tunnel Along Alignment lays a bore along the selected alignment. It samples the horizontal geometry and the vertical profile at the step you give, so the tunnel follows the road or the track in plan and in level. The alignment can be of any kind: a tunnel wraps a road alignment and a rail alignment equally.

Create Tunnel Around Corridor is the other way round. Select a corridor and the bore is sized to contain the carriageway that corridor actually builds, read from its own cross-sections rather than from a guess. Use this once the section inside is real.

Cross-section and lining

Three section types: bored (circular, the TBM case, set by the inner radius), horseshoe (arched crown on straight walls, the conventional case, set by width and height) and cut and cover (a rectangular box, also width and height). Three lining types: precast segmental rings, cast in place and sprayed. Precast is the one the model knows in detail: it carries the ring width, the number of segments in a ring, whether a key segment closes it, and the stagger angle that breaks the joints from one ring to the next.

Everything above is editable in the Properties dock after the fact, including the width and the height that size a horseshoe or a box.

Staying on the alignment

A tunnel records the alignment it was laid along and how it was sampled. Move that alignment, or edit its vertical profile, and the tunnel is marked out of date. It is not moved for you: a drawing that quietly rearranges itself is harder to trust than one that says what has changed.

Resync Tunnels to their Alignment puts every out-of-date tunnel back on its host, re-sampling with the same station range and step it was created with. The rebuild is one undo step and the tunnel keeps its identity, so anything selected or annotated against it survives. What is re-derived is the centre line only. The bore size, the lining and the gauge are your decisions; a refresh that re-derived those would redesign the tunnel every time the road inside it moved. Whether the bore still contains its gauge is the clearance check's question, below.

A tunnel built by hand, with no host alignment, reads nothing outside itself and is never flagged.

Checks

Clearance Check tests that the clearance envelope fits inside the bore. The Gauge outline property chooses what it is tested against: any clearance gauge the install has loaded — EN 15273, DB Ril 800, an AREMA plate, an operator’s own file — or the plain rectangle a tunnel carries by default, sized from the host alignment’s category and the project design standard.

A named outline is the real thing, not a box drawn around it. It is widened for the local curve radius, because a vehicle sweeps wider than its own section in a curve, and tilted by the applied cant, because the section is rigid with the track. Both vary along the run, so the check walks every station the tunnel was sampled at and reports the tightest point: the margin, the station, and the offset and height of the corner that does it. A gauge file that states no vehicle is taken as already including its widening, and the report says so rather than adding it twice.

Cant is read from the alignment when it carries one. A road alignment has none, and a rail design whose cant is computed rather than stored has none here either; the outline is then taken untilted, which the report states instead of assuming a value nobody supplied.

In the Sections tab the envelope is drawn as the outline it actually is, at that station.

Cover Check measures the ground over the tunnel crown against the existing ground surfaces and warns where the cover is thin or the crown daylights.

The ground it goes through

A tunnel does not meet one ground; it meets a sequence of them, and the support changes with each. So a tunnel carries geology bands: from station to station, a unit name, what was logged at the face, and the support class specified for that stretch.

Two classifications are computed from the log. RMR, Bieniawski 1989, is six ratings summed: intact strength, RQD, discontinuity spacing, discontinuity condition, groundwater, and an adjustment for how the discontinuities sit relative to the drive. Q, Barton at NGI, is (RQD / J_n) × (J_r / J_a) × (J_w / SRF). From those come the class and its name, the GSI, the stand-up time at this span, the span that would stand with no support at all, the bolt length, and the roof support pressure. Every one is the published table row or the published closed form, computed exactly, so any of them can be checked by hand against the source.

They are read at the excavation span, the width of the hole, not of the finished tunnel: the ground has to hold up what is dug, not what is later lined.

Tunnel › Rock Mass & Support Classes prints all of it band by band, and prints what the classification proposes next to what the design specifies. When those disagree it says so. That is not an error: choosing a heavier or a lighter class than the charts suggest is an engineering decision, and one worth being able to defend. A stretch with no class at all is called out rather than passed over.

Support classes

A support class is what gets installed: shotcrete thickness, mesh layers, bolt length and spacing, lattice girder or steel rib spacing, forepoling ahead of the face, the excavation sequence (full face, top heading and bench, or with the invert closed behind) and the round length.

The classes that ship are examples, and every one says so wherever it is printed. They are shaped like the A / B / C families Austrian and Turkish road practice uses, so the program has something to run with and a project has something to copy. They are not a transcription of any published table. Replace them with the project’s own by name, the same way turnout catalogues and clearance gauges are replaced.

Openings: cross passages, niches, exits

A tunnel is not one tube. Between two tubes run cross passages; into the wall go niches for equipment and lay-bys; out of the tube go emergency exits. All three are the same thing geometrically — a small bore meeting a main tunnel — so they are one kind of object with a setting for which it is, not three.

Tunnel › Add Cross Passage joins the selected tube to the nearest other one. Ask for several and it lays them at the spacing the project’s design standard requires, without your having to look that up; ask for more than fit and it stops at the portal rather than placing one beyond the end of the tube. Add Niche or Emergency Exit drives a single bore into one wall, to the side and depth you give.

An opening stores no geometry of its own. Where it runs is a station on one tunnel and, for a cross passage, the other tunnel, and both are read at the moment anything asks. Move either tube and the passage is already in the right place. There is nothing to refresh because there is nothing cached, which is why an opening never appears out of date.

What it takes out

Where a passage meets a main tunnel it adds no excavation: that ground was already taken out by the main drive. What it does is take a hole out of the main tunnel’s lining, and that hole is a quantity somebody pays for. The take-off deducts it from the support class band the opening stands in — the line that pays for that lining is the line that loses it — and shows the deduction as its own negative row so the arithmetic can be followed instead of trusted. An opening standing where no class has been decided has no line to come off, so it is reported apart rather than quietly charged to a class it is not in.

The deducted volume is the part of the bore lying between the tunnel’s clear face and its excavation face, through the one wall it enters by. A passage that went clean through both walls would take twice that, and a niche taking a through-bore’s worth would be wrong by a factor of two and look entirely plausible — so the two cases are counted separately.

Two things it declines rather than guesses. A bore skewed past 45° cuts a path through the shell that the simple correction no longer describes. A tunnel that is not a bored circle — a horseshoe, a box — is walled flat where an opening meets it, and treating the arch as a circle of the same half-width would produce a confident wrong number. In both cases the deduction is left out and the reason is printed.

A cross passage between two tubes that are too close together is refused outright, and the refusal carries the numbers: how far apart the two axes are, and how much of that the two bores already take. Twin running tunnels at 4.5 m centres with 3.3 m of bore each have nothing between them to drive through, and being told the three figures beats being told they touch.

Is there a way out?

Openings & Spacing Check is a rule check, not a geometry one. Past a certain length a road tunnel must have escape routes, spaced no further apart than the rule allows, and both figures come from the project’s design standard: RABT asks for escape above 400 m and one every 300 m, NFPA 502 for 1000 ft on both counts, the EU directive as Turkey and the UK transpose it for 500 m and 500 m. These are safety regulation rather than geometry, so a project plugs its own figures in through its standard data rather than having them derived.

The two portals count as ways out, because they are. A niche does not — it is a cupboard, not an escape — and counting it would pass a tunnel that has no way out at all. A failure is reported with the station the gap opens at, so it points at somewhere on the drawing rather than at the tunnel in general.

Take-off

Tunnel › Excavation & Support Take-off produces the quantity a tunnel contract is actually written against: per support class, over the chainage that class covers, the excavation, the shotcrete, the bolts by number and by total length, the mesh, the girders by number and by steel mass, the lining and the waterproofing. Give it a folder and it writes the same table as CSV, one file per tunnel. It runs headless with the same numbers.

Everything is derived from the same section the Sections tab and the 3D view draw, so a quantity can never describe a different tunnel from the one on screen.

Theoretical and paid excavation are separate figures. The Payline property is how far outside the theoretical excavation line the contract pays for. The volume to the theoretical line and the volume to the payline are reported side by side and never added together, because the difference between them is precisely what a unit-price contract is argued over.

A stretch with no support class is not a stretch with no support. It is a stretch nobody has decided yet, so it is kept out of the totals, given its own line, and called out. Undecided drive is not free drive.

Building the rings

A shield-driven tunnel turns its curves by rotating tapered rings. The two end faces of a tapered ring are not parallel: across the diameter one side is longer than the other by the taper, so setting a ring points the drive. Turn the ring about the tunnel axis and the same deflection is shared between plan and level — upright it steers sideways, on its side it steers up or down.

Two numbers say what a ring can do. A taper k on an outer diameter D deflects the drive by k/D; a ring of width w therefore turns a radius of wD/k and no tighter. The default 30 mm taper on a 6 m ring 1.5 m wide deflects 5 mrad and drives 300 m — a metro figure. The Ring taper property is in metres, like every other length on the tunnel; the create form asks for it in millimetres, which is how a taper is quoted.

Tunnel › Ring Build List reads the curve off the host alignment ring by ring and chooses, for each one, which of the ring’s key positions to build it at. The answer is a sequence, not a lookup: what ring 40 should be depends on where rings 1 to 39 left the drive. Give it a folder and it writes the whole list as CSV, one file per tunnel, with the station, the key position, the rotation and the offset from the alignment at each ring. Only a precast bored tunnel has rings to build; anything else is told so rather than given a list it cannot use.

The choice at each ring is made on position, not heading. Steering to get the heading right looks correct ring by ring and walks steadily off the line: on a tangent it picks a direction, corrects back to the same heading, and keeps every millimetre of the sideways step it took on the way. Steering to get back on the line gives that back, which is why the reported offset stays bounded over a long drive instead of growing with it.

Read the report for two things. The worst offset is what a surveyor would measure against the alignment; a few millimetres on a curve the taper suits, centimetres when the key positions are coarse. And if the alignment is tighter anywhere than wD/k, the report says the taper is not enough — the ring cannot turn that curve, and the answer is a bigger taper or a shorter ring, not a build list that pretends otherwise.

Overbreak and underbreak

The face never comes out on the theoretical line. Where the excavation went outside it there is overbreak, which somebody pays to have hauled away and filled back with concrete. Where it stayed inside there is underbreak, which is rock standing where the tunnel is supposed to be. This is the core of a tunnel payment certificate, and the two numbers are never added together: one is a quantity to pay for, the other is a fault to go back and take out, and a net figure hides both.

Get the measurement in one of two ways. Slice Scan into Sections takes every point cloud in the document and cuts it into sections along the selected tunnel at the step you give, keeping the points within a slab of each station. Import Measured Sections reads a surveyor’s CSV of station, u, v rows — u across the tunnel and v up it, both measured from the axis. Either way the result is a set of measured sections on the tunnel, and either is one undo step.

The comparison is made on rays from the section centre, which is what scanner software does and what a tunnel drawing is dimensioned against. A tunnel section is star-shaped about its axis, so each direction has one theoretical radius and one measured radius and the area between them is exact arithmetic. A measurement a uniform 100 mm outside a 3 m bore comes out at exactly π((R+0.1)²−R²) — the annulus, to the last bit, and the same whether the comparison uses eight rays or a hundred and twenty.

Volumes between measured sections use average end area, the same method the rest of the take-off uses. One section on its own measures an area and no volume at all, and the report says that rather than stretching one reading over a drive.

Where a bin of scan points holds several readings the profile takes their median. A scan of a tunnel wall also catches the ventilation duct, a cable tray and occasionally a person; a mean would be dragged by them and a maximum would let one of them stand in for the rock.

The Sections tab draws the measured face over the theoretical one in red, for the nearest measured section within half a metre of the station you are looking at. It joins the ends of the measured line only when the scan actually went all the way round — closing a half-measured section would draw a wall of rock that nobody measured.

Quantities and output

Segment Schedule reports, per tunnel, how many rings the run takes, how many segments that is in total, and the concrete volume of the lining.

Generate Lining Rebar builds a reinforcement cage for a bored lining from the bar diameter, the ring spacing, the number of longitudinal bars and the cover. The cover is honoured: the cage sits inside it, as two layers when the lining is thick enough to hold them and as one at mid-lining when it is not. The cage lands under the tunnel and goes through the same bending schedule as the rest of the project's reinforcement.

Export Segment (STEP) writes one precast segment as a solid, for the mould maker. Give it a path; without one it writes tunnel_segment.step beside the project.

Getting it out: IFC, LandXML, geotechnics

A tunnel used to leave this program as one shell. Lining, support and waterproofing are different trades, different payment items and — when IFC 4.4 lands — different classes; a single blob carries none of that, and a receiving system cannot invent it. So the export is split, and the split is carried in the classes and the property sets now, before the schema that will name them exists.

• Lining — intended IfcTunnelPart / LINING. IFC 4.3 add2 has no tunnel entity, so it exports as a proxy carrying the intended class, and the export’s loss report says so. That line is an honest record of a gap in the schema, not a gap here: when 4.4 lands, only the writer’s class list changes.
• Support — IfcCovering, one product per geology band, with SHOTCRETE as its object type. A real 4.3 class: shotcrete is a covering on the excavated face.
• Waterproofing — IfcCovering / MEMBRANE. The enum item exists in the schema; it was read off the bindings rather than hoped for.

Bolts, mesh and steel sets have no geometry in this model, so they ride in the support covering’s property set rather than as invented solids. A solid nobody computed is worse than a number somebody did. The support’s pset also carries whether its class is a shipped example or a cited table, so a figure nobody can cite does not arrive somewhere else looking like one that can.

LandXML carries the tunnel’s section and its geology bands, inside <Feature> — LandXML’s own extension point, since the schema has no tunnel element. The axis is deliberately not repeated there: it is the host alignment and already travels by the ordinary alignment export, and two curves in one file is a reconciliation nobody asked for.

Export for Geotechnical Analysis writes the section and the excavation stages as a flat definition a PLAXIS or RS2 pre-processor can read: the refined section polygons, then one stage per geology band in the order the drive meets them, each with its classification, its roof pressure, and what is installed in it. There is no standard for this, so the format is ours, it says its units on the first line, and it says plainly that it is ours — pretending to be a PLAXIS file would be worse. A stretch with no support class decided arrives as SUPPORT NONE_SPECIFIED rather than silently.

The tunnel in the Composer

A cast-in-place or sprayed lining is a continuous section, so it belongs on the corridor road rather than in a producer of its own: the Subassembly Composer’s palette carries a Tunnel category with five primitives — the lining arc, the invert fill, the service walkway, the cable duct under it, and the clearance gauge. Build a corridor from those and the tunnel inside it is the real cross-section, not an approximation of one; and if the carriageway or the track is built at the same place, what runs inside the tunnel is a fact rather than a guess.

The precast ring is not there, and will not be. Discrete rings set at staggered key positions cannot be said as a swept section, so they keep their own producer — the ring build list, above. Forcing them into a subassembly would cost more than it bought and would lie about what a ring is.

The gauge primitive draws its envelope in links and carries no shape. A clearance envelope is a constraint, not a material; giving it a shape would let it be taken off as a quantity, and a tunnel would be billed for the air inside it.

What a tunnel is not, yet

The module models the bore, its lining, the ground it goes through, the rings it is built from, the openings driven into it, what was actually excavated against what was drawn, the way all of that leaves for another system, and the continuous lining as a corridor section. It does not yet carry portal structures — headwall, wing walls, the cut in front of them — nor pre-support geometry: a forepole umbrella is named in the support class and in the export, but no pipe is drawn. Nor does it register or align a point cloud, drive a scanner, or fill a scan shadow by interpolation: it compares what it is given.

The ring maths is the geometry of steering, not the structure of the ring: no load case, no joint stiffness, no bolt force. The ITA and DAUB design situations are a separate body of work and nothing here pretends to any of it.