Construction

A boring machine 53 feet across stalled 177 feet under the Yangtze and could not be reversed, but the crews did not abandon it and drove a second shield at it from the far bank

By SEP 10, 2026 11:50 AM 4 MIN READ
Yangtze boring machine cutterhead pressed into tunnel docking chamber, two boring machines

A tunnel under a very large river, with a machine at the end of it that has stopped.

It is 53 feet across and it cost about 50 million dollars.

It sits 177 feet below the riverbed, with the whole Yangtze pressing down on the ground above it.

It cannot be backed out and it cannot be reached from the surface.

Everything behind it is finished tunnel and everything in front of it is saturated ground.

Standard practice would be to leave it there and reroute around it.

Instead somebody started a second machine on the far bank.

Why a stuck shield is normally abandoned

A shield machine is not a drill that can be pulled out. It builds the tunnel behind itself as it goes.

Concrete segments are erected inside the tail of the shield, ring by ring, so the finished tube ends exactly where the machine is.

Reversing means dismantling that tube, which removes the only thing holding the ground open.

Under a river the ground is saturated and under pressure, so opening the face without support invites water and soil into the tunnel.

Reaching it from above is worse. A rescue shaft would have to be sunk through the riverbed itself.

Which leaves the third option, and it is the strange one. You dig a second tunnel to it.

What the tunnel is and why it mattered

The crossing links two industrial districts in eastern China across roughly 4 miles.

It is built to dual six lane highway standard, which is why the bore is so large.

A 53 foot diameter shield is at the upper end of what has ever been built, and machines that size are made for one job and one alignment.

The stall happened in early 2023, midway through the excavation, and there was no repair option available under that water pressure.

The project could not be rerouted either, because both approaches on land were already built and the alignment was fixed by them.

So the choice was not rescue or abandon. It was rescue or restart the whole crossing.

The docking and the numbers behind it

The second machine drove from the opposite bank on a heading calculated to intercept the first.

That is harder than it sounds, because each machine steers by pushing against the rings it has already built.

Guidance for a shield comes from a laser and target arrangement that measures position continuously against a survey line, and it corrects by steering the thrust rams.

Over a drive of that length the error budget is measured in inches, and it usually gets consumed.

Here the two shields met with a vertical difference of about 0.08 inches and no horizontal deviation at all.

The docking was completed and announced in January of this year.

That accuracy is the actual achievement, and it is rarely reported anywhere.

What the rescue proved and what it does not

The engineering is new, and no comparable underground docking at this diameter has been documented anywhere else.

Two shields meeting face to face also means two sets of tolerances have to cancel out rather than add up.

It also demonstrates that a stranded shield is not automatically a write off, which changes the arithmetic on very large machines.

What it does not prove is that this is repeatable at sensible cost, and the account of it is a story about a recovery rather than a method.

The second machine was itself an enormous purchase, so the saving is against restarting a crossing rather than against a normal drive.

The original failure is also thinly described, and the technical coverage does not settle what actually stopped the first shield.

Until that is published the lesson stays half a lesson, because nobody can design against an unnamed fault.

What a boring machine really risks

Every long drive is a bet that the ground ahead matches the boreholes taken before the work started.

Obstructions, boulders and old foundations are the classic causes of a stall, and they turn up in city ground constantly.

A machine that stops short of its exit is an expensive object in a place nobody wants to dig, as happened when one stopped 23 feet short of its shaft under a Toronto street.

Getting the machine in at all is its own problem, which is why crews cut a cavern 147 feet down and bolted 263 pieces together under a Sydney cricket ground.

None of that is unusual. What is unusual is finishing the drive from both ends and meeting in the middle by accident of necessity.

The tunnel still has to be fitted out and opened, so the recovery is a milestone rather than a finished road.

Hugo Rojas Editor

Hugo Rojas is an editor and science writer who turns complex research into clear, engaging stories. With a sharp eye for detail and a love for the natural world, energy, and technology, he brings big ideas down to earth for every reader.