Subsea Lifts

Four water forces act on a payload between deck and seabed: buoyancy, slamming, drag and added mass. What each does to the line load, how big they get, why the final approach sets the landing, and what depth does to the compensator.

By Peter Wang, COO · December 2025 · Updated September 2026 · 13 min read

RIGEL passive heave compensator lowering a GRP cover toward the seabed.

In brief

Four forces act on a payload in water. Buoyancy changes its effective weight, slamming hits at water entry, drag grows with the square of its speed through the water, and the added-mass force grows with its acceleration.
In the worked examples, a 15 × 10 m plate at 1.57 m/s sees about 216 kN (22 t) of drag on the steady-flow coefficient, or about 474 kN (48 t) on DNV-RP-N103’s wave-zone value, and a 10 × 10 × 20 m prism at 1.23 m/s² picks up about 908 kN (93 t) of added-mass force.
Sized for the case, a heave compensator cuts the crane’s share of the payload’s velocity and acceleration relative to the water, which pulls both forces down, and keeps the wire in tension down to the landing.
Submergence and depth move the compensator’s working point. Buoyancy and water pressure push it inward and are corrected by releasing gas; cooling on the way down pushes it outward and needs gas added. RIGEL, CYGNUS and ANTARES are all rated to 3,000 m and differ in how they re-centre: RIGEL is set on deck, CYGNUS can release gas, and ANTARES can release and add it.

A subsea lift runs from the deck, through the splash zone and down the water column to the seabed. Water is about 800 times denser than air and pushes back on the payload in four ways; which one dominates changes from the splash zone to the water column to the landing.

Four forces to keep apart

Force	Scales with	Where it matters	What it does
Buoyancy	Displaced volume, $\rho_w V g$	From water entry; changes fast while the load crosses the surface	Lowers the effective weight and makes it fluctuate: a slack-wire risk
Slamming	Impact velocity squared, at water entry	The splash zone	Short, sharp load spikes in the payload and rigging
Drag	Relative velocity squared, $\tfrac{1}{2}\rho_w C_D A_\perp \dot z^{2}$	Descent through waves and current	Resists the motion; 22 to 48 t on the example plate
Added mass	Relative acceleration, $\rho_w C_A V_R\, \ddot z$	Any oscillation; grows near the seabed	Adds inertia to every oscillation; about 93 t on the example prism

The splash zone stacks slamming, fast-changing buoyancy and wave-particle velocity into a few metres of travel; splash-zone crossing covers it, and DNV-RP-N103 gives the method. The two worked examples below use different bodies, so their results are not two forces on one payload and are not added together.

Drag

Drag grows with the square of the payload’s speed through the water around it, not its speed over the seabed. Crane and winch motion and the water’s own particle velocity both feed into that relative velocity $\dot z$ :

F_D = \tfrac{1}{2}\, \rho_w\, C_D\, A_\perp\, \dot z\, |\dot z|

Here $C_D$ is the drag coefficient and $A_\perp$ the area facing the flow; DNV-RP-N103 tabulates both by shape. The tabulated coefficients are for steady flow. In the oscillating flow of the wave zone the RP uses a coefficient typically two to three times higher, and without CFD or model tests it gives $C_D \ge 2.5$ for typical subsea structures.

Worked example: a 15 × 10 m plate

Take a plate lowered flat in 4 m waves with an 8 s period, as a sinusoid with amplitude $\zeta = H/2 = 2\ \text{m}$ . As a screening input, its peak velocity relative to the water is taken straight from the wave motion:

\dot z_{\max} = \zeta\, \omega = 2 \cdot \frac{2\pi}{8} \approx 1.57\ \text{m/s}

For a plate normal to the flow the coefficient is

C_D = 1.1 + 0.02 \left( \frac{15}{10} + \frac{10}{15} \right) \approx 1.14

and with $A_\perp = 150\ \text{m}^2$ and $\rho_w = 1{,}025\ \text{kg/m}^3$ :

F_D = \tfrac{1}{2} \cdot 1{,}025 \cdot 1.14 \cdot 150 \cdot 1.57^{2}

F_D \approx 216\ \text{kN} \approx 22\ \text{t}

That is with the steady-flow coefficient. With DNV-RP-N103’s wave-zone value of at least 2.5, the same plate sees about 474 kN (48 t), more than twice as much.

Figure 1. Drag on the example plate with the steady-flow coefficient (

C_D = 1.14

A_\perp = 150\ \text{m}^2

\rho_w = 1{,}025\ \text{kg/m}^3

) against its velocity relative to the water. The force grows with the square of the velocity, so halving the relative velocity quarters the drag. How much of the velocity a compensator removes on a given lift is a result of the lift model, not a property of the unit.

Steady-flow drag coefficients by shape (DNV-RP-N103)

Shape	$C_D$	$A_\perp$
Sphere	$0.5$	$\pi r^2$
Horizontal cylinder	$1.2$	$2 r L$
Vertical cylinder	$1.1$ at $L/2r = 0.5$ , $0.9$ from $1$ to $4$ , $1.0$ at $8$	$\pi r^2$
Cube, face on	$1.05$	$a^2$
Cone, tip into the flow	about $0.5$ , depending on the angle	$\pi r^2$
Rectangular plate, normal	$1.1 + 0.02\,(L/W + W/L)$	$L W$

Added mass

A payload that accelerates in water has to accelerate some of the water around it too. That water acts as extra mass whenever the payload speeds up or slows down, so it adds inertial force on top of the static weight:

m_A = \rho_w\, C_A\, V_R

Here $C_A$ is the added-mass coefficient and $V_R$ the reference volume, both tabulated by shape in DNV-RP-N103. Near the seabed the added mass grows through confinement, which is exactly where the landing happens.

Worked example: a 10 × 10 × 20 m prism

Take a square prism moving along its long axis in the same 4 m, 8 s waves. Its peak acceleration is

\ddot z_{\max} = \zeta\, \omega^{2} = 2 \cdot \left( \frac{2\pi}{8} \right)^{2} \approx 1.23\ \text{m/s}^2

At $L/W = 2$ the coefficient is $C_A = 0.36$ , and $V_R = W^2 L = 2{,}000\ \text{m}^3$ . So

F = 1{,}025 \cdot 0.36 \cdot 2{,}000 \cdot 1.23

F \approx 908\ \text{kN} \approx 93\ \text{t}

Added-mass coefficients by shape (DNV-RP-N103)

Shape	$C_A$	$V_R$
Sphere	$0.5$ , in every direction	$\tfrac{4}{3}\pi r^3$
Cylinder, across its axis (a pipe or spool lowered level)	$0.62$ at $L/2r = 1.2$ , rising to $1.00$ for a long cylinder	$\pi r^2 L$
Rectangular plate, normal	$0.58$ at $L/W = 1$ , rising to $1.00$ for a long plate	$\tfrac{\pi}{4} W^2 L$
Circular disc, normal	$2/\pi$	$\tfrac{4\pi}{3} r^3$
Square prism, along its length	$0.68$ , $0.36$ , $0.19$ , $0.08$ at $L/W = 1, 2, 4, 10$	$W^2 L$

What the worked examples assume

The relative velocity and acceleration are taken straight from a sinusoid of the stated height and period, not from a vessel response model. A real lift gets them from the coupled model: vessel motion, crane-tip response, winch command, wire dynamics, compensator state and the water’s own particle motion.
Drag and added mass do not peak at the same moment: on a sinusoid the velocity and acceleration peaks are a quarter period apart. Combined peaks come from a time-domain result.
The coefficients are infinite-fluid values with no correction for the free surface or the seabed.
Buoyancy and slamming are left out of both examples; in the splash zone they usually govern.

Deck to seabed, one run

With ANTARES, one run uses its four modes:

On deck: Locked. The load is rigged and lifted off with the rod locked, which keeps the lifting height down.
Splash zone. Over the side and through the wave zone, with slamming and snap loads checked; the unit is stiffer and better damped for the crossing.
Lowering: Subsea. The unit is softer for the water column; drag and added mass follow the local relative motion.
Landing. The valves open and the EQ function is released for a soft set-down onto the seabed.

RIGEL and CYGNUS run one damping setting, chosen on deck for the lift. Whichever unit is used, a passive system isolates the load only from waves well shorter than its natural period, and it amplifies motion near that period. As wire pays out, the suspended system gets softer and its natural period grows, so it can reach the wave band at some depth; the lift model checks resonance along the descent.

ANTARES adaptive passive heave compensation, deck to seabed, modelled in CONSTELLATION. For the full sea-state screen of a lift like this, see operability screening.

Figure 2. The same run in numbers, from CONSTELLATION: payload depth and wire tension for the manifold on ANTARES (250 t, 4.5 m stroke), H_s 3.0 m and T_p 8 s in 150 m of water. The band spans the lowest to the highest tension in each 8 s wave cycle. The line carries about 189 t in air and settles near the 168 t weight in water for the lowering. In the splash zone the slings never go slack (DAF 1.13), and the manifold touches down at 0.11 m/s with a landing DAF of 1.00 before the seabed takes the load.

Forces of this size, a 93 t inertial force on top of the static weight, are why a heavy subsea lift is often run on a passive compensator sized for the case. It cuts the crane’s share of the payload’s relative velocity and acceleration, which pulls down both the drag and the added-mass force; the water’s own particle motion remains. Whether the wire stays in tension all the way down is checked in the lift model for the case. In our range, all three units are rated to 3,000 m: CYGNUS covers heavy lifts up to 10,000 t, RIGEL the standard lifts from 55 t, and ANTARES the lifts that need different settings for the splash zone, the water column and the landing.

Landing: the final approach

Touchdown energy grows with the square of the approach velocity, and the peak contact force depends on how much of that energy the soil and the structure absorb. Uncompensated, the approach velocity carries the vessel’s heave on top of the winch rate. A hard or bouncing landing risks damage to precision surfaces and seals on manifolds, trees and templates; buckled or over-penetrating mudmats; and a load left out of position for later tie-ins.

With a compensator sized for the sea state, far less of the vessel’s heave reaches the load, so its approach velocity comes much closer to the winch rate. We screen landings against a touchdown velocity of 0.5 m/s, the acceptance limit we apply under DNV-RP-N103. A passive unit reduces the vessel motion that reaches the load rather than removing it, so the velocity actually reached depends on the sea state and is confirmed per case.

Approach. Lower to a few metres above the seabed under normal compensation.
Final descent. On ANTARES, switch to Landing mode; pay out at a controlled rate.
Touchdown. Contact at 0.1–0.5 m/s, the residual heave taken up so the load does not bounce.
Set-down. Release the wire tension gradually; the compensator limits snap loads as the seabed takes the weight.

For a soft set-down, ANTARES’s Landing mode opens its valves and releases its EQ function. Where an exact lowering velocity has to be held, an active system provides that control. The motion physics behind a soft landing is in passive heave compensation.

What depth does to the compensator

Once the load is in the water, three effects move the compensator off its working point: the payload’s buoyancy, gas cooling and water pressure on the rod. A fourth, leakage, is kept out by design. RIGEL, CYGNUS and ANTARES are all rated to 3,000 m; they differ in how they re-centre. CYGNUS and ANTARES use their equilibrium (EQ) function, which on CYGNUS releases gas and on ANTARES also adds it. RIGEL has no EQ function: its charge is set on deck for the lift.

Effect	What it does to the compensator	The answer
Buoyancy	The payload’s net weight drops in water, so the unit’s equilibrium moves inward, at worst to full retraction	Release gas to re-centre it; on CYGNUS, once the gas has cooled. CYGNUS and ANTARES can both do this.
Cooling	The gas cools on the way down and its pressure falls, most of it quickly through the thermocline; the stroke lost depends on gas volume, preload and descent	Add gas from the onboard bottle. ANTARES can; CYGNUS, which can only release gas, cannot.
Water pressure	About 1 bar per 10 m acts on the rod and pushes it in. On a 250 t lift with a 180 mm rod at 1,000 m the water pushes the rod in with about 26 t, as if the load had dropped to 224 t. On a unit whose gas volume is about three times the oil it displaces over the full stroke, that moves the equilibrium from mid-stroke to about a sixth of the stroke	Release gas as the unit goes deeper. CYGNUS and ANTARES can both do this.
Leakage	Water getting in, with lost performance and corrosion	Design, testing and maintenance: double seals on two sealing surfaces, external-pressure testing at FAT, compartments that contain a leak, scheduled seal replacement

Water pressure example: 1,025 kg/m³ seawater gives about 10 MPa at 1,000 m, or about 256 kN on a 180 mm rod.

The effects partly cancel. Cooling happens mostly in the thermocline, while the pressure shift keeps growing with depth. CYGNUS’s one-way bleed covers a deep lift if it holds its release until the gas has cooled and has the stroke to ride off-centre until then. ANTARES can add gas from its onboard bottle, so it re-centres at every stage of the descent. RIGEL rides the descent on the charge set on deck, so the lift model checks that the net shift stays inside its stroke.

Frequently asked questions

Why is a subsea lift harder than a lift in air?

The water pushes back in four ways: buoyancy changes the effective weight, slamming hits at water entry, drag resists with the square of the relative speed, and the added-mass force makes the payload effectively heavier whenever it accelerates. Which one dominates changes from the splash zone to the water column to the landing.

How large can drag forces get on a subsea lift?

In the worked example, a 15 × 10 m plate at 1.57 m/s relative velocity sees about 216 kN, or 22 t, with the steady-flow drag coefficient, and about 474 kN, or 48 t, with DNV-RP-N103’s wave-zone value of 2.5. Drag grows with the square of the velocity: halve the relative velocity and the force drops to a quarter.

What is added mass, and how big is it?

The water a payload has to accelerate along with itself. The 10 × 10 × 20 m prism in the worked example picks up about 908 kN, or 93 t, of inertial force in 8 s waves, and added mass grows near the seabed through confinement.

Which force dominates the splash zone?

Slamming and fast-changing buoyancy on top of the wave-particle velocity. That combination, packed into a few metres of travel, sizes the compensator more often than any other stage of the lift.

How does heave compensation protect a subsea lift?

Sized for the case, a compensator cuts the crane’s share of the payload’s velocity and acceleration relative to the water, which pulls down both the drag and the added-mass force, and keeps the wire in tension so slack-and-snap events are limited. A passive unit isolates the load only from waves well shorter than its natural period, so the lift model checks resonance as the wire pays out. In our range, all rated to 3,000 m: CYGNUS for heavy lifts, RIGEL for standard lifts, ANTARES for lifts that need different settings for the splash zone, the water column and the landing.

Which compensator suits deep water?

RIGEL, CYGNUS and ANTARES are all rated to 3,000 m; what differs is how they re-centre as the depth changes. Buoyancy and water pressure push the compensator inward and are corrected by releasing gas; cooling on the way down pushes it outward and needs gas added. The effects partly cancel: cooling happens mostly in the thermocline, while the pressure shift keeps growing with depth. RIGEL rides the descent on the charge set on deck, so the lift model checks that the net shift stays inside its stroke. CYGNUS releases gas and covers a deep lift if it holds its release until the gas has cooled. ANTARES can also add gas from its onboard bottle, so it re-centres at every stage.

Working the numbers on a subsea lift?

Send the payload, its dimensions and weight, the water depth and the sea states. We’ll model the lift in CONSTELLATION and recommend the compensator and its size.

Send your lift case →