Subsea Retrieval: Breakout, Ascent and Water Exit

Recovering an asset from the seabed is not an installation in reverse. The seabed holds on to it, its weight is uncertain, and a pull driven by the vessel’s heave can load the crane far past its SWL before the seabed lets go. What happens at breakout, and what a heave compensator did in our suction-caisson recovery case.

By Norwegian Dynamics Engineering · March 2026 · Updated September 2026 · 9 min read

Blender illustration of a subsea module hanging just above the seabed from an in-line heave compensator.

In brief

Retrieval adds loads and risks an installation does not have: the seabed’s grip, a weight changed by marine growth, trapped water and soil, and a structure weakened by years in service.
A suction caisson resists being pulled out, and it resists harder the faster it is pulled, because water cannot flow in under it fast enough. On a bare wire, the vessel’s heave sets that pace.
A passive compensator in line takes the heave as stroke, so the winch sets a slow, steady pull and the suction under the caisson stays low.
In our CONSTELLATION case, a CYGNUS 700 t / 5 m pulled a Ø6 m suction caisson out of firm clay at 120 m, at Hs 2.5 m, with a peak line load of 566 t, 87 % of the 650 t crane’s SWL. On a bare wire the line reached about 1,280 t, nearly twice the crane’s SWL.

An installation lowers a known object onto a prepared seabed. A recovery pulls up something that has sat there for years: held by the soil, covered in growth, often full of water and sediment, and possibly weaker than when it went down.

Why retrieval is harder than installation

The seabed’s grip. Suction under a caisson or a mudmat, and adhesion and friction along its skirts, add to the weight until the asset breaks free.
An unknown weight. Marine growth, trapped water and sediment add weight, and missing or corroded parts take some away. The as-found weight is an estimate with a range.
A weaker structure. Lift points and members may have lost strength to corrosion or fatigue, so the load they can take may be lower than at installation.
The release. When the seabed lets go, the grip that was holding the asset down is gone at once. The stretch stored in the wire and rigging throws the asset upward, and the line can go slack and then snatch.

Breakout

Suction caissons show the problem most clearly. They are normally removed by pumping water back in under the lid, which pushes them out. When that is not possible, the crane has to pull them out, and that is the case modelled here. As a caisson is pulled up, water has to flow in under its top through the valve. If it is pulled faster than the water can follow, the pressure inside drops and the seabed holds on harder, until the soil inside the caisson gives way. Friction along the skirt adds a part that changes little with speed.

On a bare wire, the pull comes from the vessel’s heave as much as from the winch: every wave pulls fast, and the soil’s resistance and the line load rise with it. If the caisson lets go under that pull, the line’s stored stretch throws it upward (Figure 1).

Figure 1. Pulling a suction caisson out of the seabed. (A) On a bare wire the vessel’s heave drives the pull; a fast pull builds suction under the caisson, because water cannot flow in through the top valve fast enough, and if it lets go the line snatches. (B) With a compensator in line, the unit takes the heave, the winch pulls slowly and steadily, and the suction under the caisson stays low. Schematic, not to scale.

A passive compensator in line changes that. The unit takes the vessel’s heave as stroke, so the pull on the caisson is set by the winch, slow and steady, and the soil’s resistance stays lower. In our case the caisson came out steadily, at under 0.1 m/s, with no sudden release. Where a seabed does let go at once, the unit’s damping limits how fast it lifts the asset away, instead of the line’s stored stretch throwing it upward.

Ascent and water exit

Once free, the asset is lifted through the water column, and the line carries its weight in water plus drag, often with soil still inside a caisson or water trapped in a structure. At the surface the splash zone works in reverse: buoyancy falls as the asset leaves the water, and the water and soil it carries add weight in air. DNV-RP-N103 asks for that entrapped water to be counted when the compensator’s stroke is checked on recovery. Splash zone crossing explains the loads at the surface.

Our suction-caisson recovery case

In a CONSTELLATION model, a construction vessel with a 650 t crane recovers a Ø6 × 10 m suction caisson (107 t in air, 93 t in water) by crane pull alone, through a Ø0.5 m top valve, from firm clay in 120 m of water, on a three-fall 90 mm wire, at Hs 2.5 m and Tp 8.0 s. The same rig and sea were run with a CYGNUS sized for the job, 700 t / 5 m, in line and on a bare wire.

Figure 2. Peak line load in our suction-caisson recovery case, the same rig and sea with and without the compensator. The wire’s breaking load, 1,658 t, was not reached in either run. Without the compensator the line load passed both the crane’s 650 t and the wire’s 663 t SWL; the crane, with the lower SWL, is the limit that governs. One sea-state realisation, Hs 2.5 m, Tp 8.0 s, modelled in CONSTELLATION.

Result	Bare wire	With CYGNUS
Peak line load	about 1,280 t	566 t
Against the 650 t crane SWL	nearly twice	87 %
Against the wire’s 663 t SWL	nearly twice	85 %
Against the wire’s 1,658 t breaking load	77 %	34 %
Caisson pulled out of the clay	no: it rose about 1 m	yes: drawn out steadily at under 0.1 m/s, skirt clear of the seabed about 4.5 minutes into the run

From breakout on, the unit rode clear of its stops, mostly in the lower half of its stroke, while the winch drew the caisson steadily out of the clay. The run ends once the skirt is clear of the seabed, so the ascent and the water exit were not part of it. On the bare wire, the line load passed the crane’s SWL while the soil still held the caisson, which rose about 1 m. It is one sea-state realisation, and the CYGNUS was sized for this case; a real recovery is modelled on its own asset, soil and vessel.

The same case on film: a CYGNUS pulling a suction caisson out of the clay at 120 m, simulated in CONSTELLATION.

Choosing the unit

RIGEL keeps the setting it was given on deck. CYGNUS can release gas, and ANTARES can release and add it, so both can re-centre as the load falls after breakout. In our case the CYGNUS kept its breakout setting throughout: charged for several hundred tonnes, it would sit closed on its stop once the caisson hangs free at 93 t in water, so the ascent and the water exit need their own check. All three are rated to 3,000 m. For the most uncertain breakouts, active control adds flexibility; our VEGA active unit is in development.

Sources and assumptions

The recovery case is a CONSTELLATION model with rate-dependent suction-caisson extraction: friction along the outer skirt plus an internal under-pressure set by the flow through a Ø0.5 m top valve and capped by the lower of the soil’s reverse bearing capacity and cavitation, in clay with an undrained shear strength of about 42 kPa. The vessel’s motion is applied at the crane tip in six degrees of freedom. The run covers the pull-out: it ends once the caisson’s skirt is clear of the seabed.
One sea-state realisation (Hs 2.5 m, Tp 8.0 s). The CYGNUS 700 t / 5 m was sized for the case and kept one gas charge throughout.
The entrapped-water requirement is from DNV-RP-N103, formerly DNV-RP-H103 (April 2010, 4.7.7.4).

Frequently asked questions

Why is subsea retrieval harder than installation?

The seabed holds the asset, its weight has changed since it was installed, its structure may be weaker, and the release at breakout can throw it upward and snatch the line. An installation has none of these.

What is suction breakout?

A suction caisson or a mudmat resists being pulled out of the seabed. Under a caisson, water has to flow in through the top valve as it rises; if it is pulled faster than that, the pressure underneath drops and the seabed holds on harder. Under a mudmat on clay the suction lasts too long to wait out, so the lift has to break it. Friction along the skirts adds to it.

How does a heave compensator help at breakout?

It takes the vessel’s heave as stroke, so the winch sets a slow, steady pull and the soil’s resistance stays lower. In our case the caisson came out steadily, at under 0.1 m/s, with no sudden release. Where a seabed does let go at once, the unit’s damping limits how fast it lifts the asset away, instead of the line’s stored stretch throwing it upward.

Does the compensator need to change its setting during a recovery?

It depends on how far the load falls once the asset is free. RIGEL keeps the setting it was given on deck; CYGNUS can release gas, and ANTARES can release and add it. In our caisson case the setting was chosen for the breakout, so the unit would sit closed on its stop once the caisson hangs free, and the ascent and the water exit need their own check.

What did the recovery model show?

In our CONSTELLATION case, a CYGNUS sized for the job, 700 t / 5 m, pulled a Ø6 m suction caisson out of firm clay at 120 m, at Hs 2.5 m, with a peak line load of 566 t, 87 % of the 650 t crane’s SWL. On a bare wire the line reached about 1,280 t, nearly twice the crane’s SWL. It is one sea-state realisation, and the run covers the pull-out, not the ascent.