Placing tunnel elements to the millimetre: the metrology decision

What Fehmarnbelt asks of the positioning team

MacArtney has been supporting Femern Link Contractors (FLC) at the Fehmarnbelt Fixed Link between Denmark and Germany, supplying integrated subsea power and connectivity for the controlled positioning of tunnel elements on the seabed. The immersed tube tunnel runs approximately 18 kilometres and sits at depths of up to 40 metres. Standard elements measure 217 metres long and 42 metres wide, each weighing 73,500 tonnes. Every immersion floats an element out of the work harbour, tows it offshore, and lowers it into a shallow seabed trench. Three of the 89 elements have now been immersed and connected, with the operation delivered to what the project describes as millimetre-level precision.

Those are the facts on the table. What sits underneath them is a positioning and metrology decision that any contractor bidding a high-tolerance placement job has to get right before the first element ever leaves the basin. The interesting question is not whether millimetre precision is achievable. It is how you specify the measurement chain that delivers it, how you assure it, and where the real accuracy is won or lost.

Why the joint, not the trench, sets your tolerance

The phrase “millimetre-level precision” needs decomposing, because it does not apply uniformly across the operation. An immersed tunnel has a tolerance stack, and the numbers differ by an order of magnitude depending on which degree of freedom you are looking at.

Placing the element in the trench absolutely – its northing, easting and level against the project alignment – is typically a centimetric problem. Trench excavation and the gravel or sand foundation carry their own tolerances, and the foundation method (screeded gravel bed or sand-flow injection) sets the final level more than the placement gear does. The demanding number is at the joint. When the immersing element meets the previously placed one, the mating faces must align closely enough that the GINA gasket compresses evenly around the full perimeter and the road and rail alignment carries through without a step. That relative alignment is where single-digit millimetres matter.

This distinction drives the whole metrology design. You are running two positioning problems at once: an absolute one referenced to a project grid, and a relative one referenced to a fixed structure already on the seabed. They have different accuracy requirements, different reference frames, and different failure modes. Treating them as one specification is the first way projects get the error budget wrong.

The consequence of getting it wrong is not a rework line item. You cannot easily re-lift and re-set a 73,500-tonne element once the GINA gasket has taken initial contact and the joint chamber has been dewatered, because hydrostatic pressure then clamps the element against its neighbour with a force measured in thousands of tonnes. A misaligned set that has to be broken and re-attempted costs the environmental window, the pontoon spread, and schedule on a critical path where 89 elements queue behind it. The metrology is cheap relative to what a bad set costs.

What actually governs immersion positioning accuracy

Several things determine whether you hit the joint tolerance, and most of them are decided long before immersion day.

Geodetic control comes first. A tunnel that connects two national territories has to be built on one consistent construction grid, with a primary control network densified on both shores and tied to a common realisation. Any inconsistency between the datum used at the Danish end and the German end propagates directly into a longitudinal misclosure that only becomes visible when the two construction fronts meet. The alignment is defined in project coordinates, so every transformation from ETRS89 or a national frame into that grid has to be single-valued and controlled. This is engineering-survey geodesy, held to network accuracies far tighter than a hydrographic campaign, and it is the foundation the rest of the chain stands on.

The surface-to-subsea transition is where accuracy leaks. During float and tow, the element and its pontoons are positioned by GNSS, with heading from dual-antenna or gyrocompass and attitude from tilt sensors. As the element is ballasted down, survey towers protruding above the waterline keep the surface link alive – GNSS antennas on the tower tops – supplemented near shore by optical prisms tracked by shore-based total stations – give position and orientation while the structure itself is fully submerged. That link is finite. Once the element nears the seabed, fine positioning shifts to mechanical and acoustic references relative to the placed element: guidance beams, guide pins, mating jacks, taut wires and displacement sensors. Every handover between measurement systems is a chance to introduce a datum offset. The discipline is in reconciling them so that the relative and absolute frames stay coherent through the changeover.

Redundancy across independent chains is non-negotiable. No single measurement system should be trusted for a set of this consequence. The float-and-tow phase leans entirely on GNSS, which is why the case for a resilient PNT backup on the tow spread is stronger than it looks – a lost RTK correction stream at the wrong moment stalls the operation inside a fixed window. Below the surface, acoustic tracking of the element as it leaves GNSS view runs into the same close-range geometry and multipath problems that trouble USBL work in extreme proximity to structures, where the baseline collapses and reflected paths corrupt the fix. Independent chains – surface optical, subsea acoustic, mechanical taut-wire – let you cross-check rather than believe.

The subsea power and connectivity layer is the enabler. This is the part of MacArtney’s contribution that is easy to under-rate. Every sensor doing the fine positioning, every camera confirming the set, every displacement transducer on the guidance system needs power and a data path during the immersion window. If that infrastructure drops out while the element is hanging in the water column, you lose the instruments that tell you where it is at the exact moment you most need them. The umbilicals, subsea connectors and distribution hubs are not a utility bolted on at the end. They are on the critical path for the whole positioning function.

The environmental window bounds everything. A 217-metre element hanging from pontoons is an enormous drag body. Cross-current across the belt loads it, sets up a catenary in the lowering lines, and induces motion the positioning system has to resolve in real time. The immersion has to be completed inside a window defined by current, tide and weather. Real-time current measurement at element depth, not just at the surface, is what tells you whether the window is genuinely open.

Where people get this wrong

The failures on high-tolerance placement work are rarely exotic. They come from a handful of specification decisions taken too late or too loosely.

1. Collapsing absolute and relative accuracy into one number

Writing “±X mm placement accuracy” into a specification without separating the absolute set on the trench from the relative alignment at the joint produces a document that cannot be met economically or verified sensibly. Absolute placement of tens of millimetres is realistic and adequate; relative joint alignment of a few millimetres is what the gasket and the alignment continuity demand. Two numbers, two reference frames, two verification methods.

2. Under-specifying the primary control network

Teams focus on the exciting real-time positioning and treat the geodetic control network as a solved problem inherited from the design. On a cross-border alignment, the network accuracy and the datum consistency between the two shores set a hard ceiling on what any downstream system can achieve. If the control is soft, no amount of RTK or acoustic precision recovers it, because you are measuring accurately to the wrong place.

3. Trusting a single measurement chain through the handover

Relying on one system to carry position through the GNSS-to-subsea transition removes your ability to detect an error before it becomes a set. When surface optical, subsea acoustic and mechanical taut-wire chains disagree beyond a pre-agreed tolerance, that disagreement is the alarm. A single chain gives you a number with no way to know if it is lying.

4. Treating subsea power and comms as a background utility

The instinct is to specify the sensors carefully and buy the power and connectivity on price. That inverts the risk. A loss of power or telemetry to the fine-positioning sensors during the lowering is a worst-case failure for the operation, and it deserves the same failure-mode analysis a DP contractor applies to a WCFDI. Redundant paths, hot-swap capability and uninterruptible supply belong in the specification, not the wish list.

5. Leaving the go/no-go criteria undefined until the barge is over the trench

The most avoidable failure is deciding, in the moment, what “good enough to dewater the joint” means. The acceptance thresholds for each degree of freedom, the cross-check tolerances between systems, and the abort criteria all have to exist in writing before mobilisation. Once the element is in the water column on a closing window, judgement made under pressure is not a control.

How to decide and how to assure it

The following are the testable positions we would hold a positioning and metrology package to on any high-tolerance immersion or heavy-placement job.

• Split the tolerance specification by degree of freedom. State absolute placement and relative joint alignment separately, each with its own reference frame and verification method. Typical industry practice on immersed tunnels holds absolute placement to the order of tens of millimetres and joint alignment to single-digit millimetres – carry those as distinct budget lines, not one blended figure.

• Verify the primary control network independently before it is used. Establish the network to a defined engineering-survey accuracy, tie both shores to a single datum realisation, and prove it with an independent baseline check rather than accepting the design network on trust. Document the transformation from the national or ETRS89 frame to the construction grid as single-valued.

• Require two independent measurement chains for every degree of freedom at the joint. Surface optical tracking of tower prisms, subsea acoustic positioning and mechanical taut-wire or displacement sensing should cross-check against a stated tolerance. Any disagreement beyond that tolerance is an automatic hold.

• Run an FMEA on the subsea power and connectivity before you run one on anything else. Treat loss of power or telemetry to the fine-positioning sensors during lowering as a critical failure. Specify redundant umbilicals or distribution paths, hot-swap connectors, and backup supply that rides through a switchover without a data gap.

• Fix the go/no-go criteria in the positioning procedure before mobilisation. Write acceptance thresholds per degree of freedom, inter-system cross-check tolerances and abort criteria into the procedure, and rehearse them. The integrity decision to dewater the joint must be traceable to logged, time-stamped measurements – an audit trail, not a recollection.

• Measure current at element depth in real time and define the window from it. Model the environmental window from current and tidal data, then confirm it live at depth during the operation rather than inferring conditions from surface readings.

• Use ROV visual confirmation as an independent final check. Before dewatering the joint, an ROV should confirm gasket contact, guidance engagement and the absence of debris on the mating face, closing the loop between the instrument numbers and what is physically there.

For governance, lean on the disciplines that already exist even though no single standard covers immersion metrology end to end. The geodetic control belongs to national and FIG engineering-survey practice. Surface positioning and its quality control map onto IOGP and IMCA GNSS positioning guidance. Acoustic close-range work draws on established subsea positioning and metrology practice from the spool-metrology world. Trench and as-built survey sit within IHO S-44, while the survey management system itself should be auditable under ISO 9001. Millimetres at the joint are not delivered by a single clever sensor. They are delivered by a control network you can trust, a measurement chain with no unchecked handovers, and an enabling infrastructure specified as though the operation depends on it – because it does.

---

Based on: MacArtney Supports Tunnel Element Positioning at the Fehmarnbelt Fixed Link