Suction caissons: the suction needed, and the limit

A suction caisson is an open-bottomed steel cylinder with a closed lid. It is lowered to the seabed, penetrates some distance under its own weight, and is then driven the rest of the way by pumping water out of the enclosed volume: the pressure difference across the lid pushes the caisson down, with the seabed as the reaction.

The attraction is that no pile driving is needed, the operation is quiet, and the same pump run in reverse retrieves the caisson at decommissioning. The constraint is that the suction cannot be increased indefinitely — past a certain value the soil plug inside the caisson lifts instead of the caisson going down, and the installation stops with the caisson short of design depth and no way to recover.

Both the requirement and the limit are computed from the same undrained strength profile. That is the interesting part of the problem.

Every number in this article is recomputed from its inputs by a script that runs on each build of this site.

The caisson and the groundPermalink to “The caisson and the ground”#

A caisson 5.0 m in diameter, 12.0 m long, in 25 mm steel, so the inner diameter is 4.950 m. The clay has an undrained shear strength rising linearly from 10 kPa at the seabed:

$$s_u(z) = 10 + 1.5\,z \ \text{kPa}$$

giving 19.0 kPa averaged over the penetration and 28.0 kPa at the tip. Take an adhesion factor $\alpha = 0.65$ on both the inside and outside walls, and a bearing factor of 7.5 on the tip annulus.

The areas follow from the geometry:

The suction requiredPermalink to “The suction required”#

Penetration resistance is skin friction on both walls plus end bearing on the annulus:

$$F_{\text{friction}} = 0.65 \times 19.0 \times (188.50 + 186.61) = 4633\ \text{kN}$$ $$F_{\text{tip}} = 7.5 \times 28.0 \times 0.3927 = 82.5\ \text{kN}$$

for a total of 4715 kN. The submerged weight of the caisson provides 1200 kN of that, and the rest must come from the pressure difference acting over the plug area:

$$s_{\text{req}} = \frac{4715 - 1200}{19.24} = 182.7\ \text{kPa}$$

Note how small a part the end bearing plays — 82.5 kN out of 4715, under 2%. A caisson is a thin-walled object and its penetration resistance is almost entirely friction, which is the opposite of a driven pile of the same diameter and is why the suction required is modest enough for the method to work at all.

The suction that ends the installationPermalink to “The suction that ends the installation”#

The suction acts on the underside of the lid, and it also acts on the soil plug. If the plug’s resistance to being lifted is exceeded, the plug moves up into the caisson rather than the caisson moving down, the tip stops advancing, and the pumping is achieving nothing.

The plug resists by friction on the inside wall and by reverse end bearing across its base:

$$F_{\text{plug}} = 0.65 \times 19.0 \times 186.61 + 9.0 \times 28.0 \times 19.24 = 2304.6 + 4849.5 = 7154\ \text{kN}$$ $$s_{\text{limit}} = \frac{7154}{19.24} = 371.8\ \text{kPa}$$

The margin is

$$\frac{371.8}{182.7} = 2.04$$

which is comfortable. Installability is exactly this ratio, and it is the number a design should report, because neither of the two suctions means anything on its own.

Why the two numbers move togetherPermalink to “Why the two numbers move together”#

Look at what appears in both calculations: $\alpha$, $s_u$ at mid-depth, and the inside wall area. The inside friction is a resistance to be overcome in the first calculation and a resistance that helps in the second.

That coupling has a consequence worth stating. If the strength profile is underestimated, the required suction falls and the limiting suction falls too, and the margin barely moves. The margin is far more sensitive to things that appear on one side only:

• Outside friction appears only in the requirement. Anything that reduces it — installation-induced remoulding, a lubricated coating — improves installability directly.
• Reverse end bearing on the plug appears only in the limit, and it is the largest single term in it, at 4849.5 kN out of 7154. The bearing factor of 9.0 used here is the standard deep value; if the plug is shallow relative to its diameter, or if the clay is fissured, it is lower, and the limit falls with it.
• Submerged weight appears only in the requirement, as a credit. A heavier caisson needs less suction, which is why ballast is sometimes added deliberately.

So the sensitivity study that matters is not on $s_u$, which is the parameter everybody worries about. It is on the terms that are not shared — the same lesson as in the CPT chain, where the uncertainties that cancel are less interesting than the ones that do not.

What the calculation does not coverPermalink to “What the calculation does not cover”#

Three things, each of which has stopped a real installation:

Layering. A sand lens inside a clay profile raises the required suction sharply while it is being penetrated, and it also allows water to flow into the caisson, which can make the required pumping rate impossible even where the pressure is achievable.

Seepage. In any permeable soil the suction drives an upward flow inside the caisson, which reduces the effective stress in the plug and therefore reduces its resistance. In sand this effect is what makes the method work — it lowers the tip resistance too — and it is also what makes the plug limit much harder to compute than it is in clay.

Set-up. The strengths above are the ones acting during installation, when the clay next to the wall is remoulded. Capacity under load, months later, is governed by the strength after the excess pore pressures generated by installation have dissipated, which is higher — often much higher. Designing the anchor on the installation strengths would be conservative to the point of absurdity; designing the installation on the set-up strengths would leave the caisson stuck 3 m above design level.

Which is the general point. The same soil, the same caisson, two different questions, two different strength profiles — and the parameter that answers one of them is the wrong parameter for the other. The anchor load this caisson has to carry is a third question again.