Post-tensioned containment walls

Fill a cylinder with liquid and the wall has to resist the pressure by going into hoop tension — it is pulled apart around its circumference, the way a barrel is held together by its bands.

Concrete is very good in compression and close to useless in tension. Ordinary reinforced concrete deals with this by accepting that the concrete will crack and letting the reinforcement carry the tension across the crack. In a building, that is a sound and economical design.

In a wall that may one day have to hold 160 000 m³ of LNG in, it is not acceptable at all. A cracked wall is not liquid tight, is not vapour tight, and lets moisture and chlorides straight in at the steel.

So the wall is pre-compressed.

A close view along a curved concrete tank wall showing a row of recessed stressing pockets receding into the distance. In the nearest pocket a rusted steel bearing plate carries an anchor head with a bundle of seven-wire strands protruding from it, the cut ends splayed slightly. The next pocket is empty, showing the open mouth of the duct. Beyond the wall, process structures and scaffolding stand out of focus, and two workers in orange hi-vis vests and white hard hats walk away along a grating walkway. — Each of those anchor heads is holding several hundred tonnes. The pockets are later filled, so almost everything that matters becomes invisible.

What prestressing actually buys you

The wall is squeezed in compression before any liquid arrives. When the hoop tension of the spill case is applied, it reduces that compression rather than reversing it.

	Reinforced concrete	Prestressed concrete
Under service tension	Cracks, steel takes the load	Stays in compression, no through crack
Water and vapour tightness	Depends on crack width control	Inherent while compression remains
Durability	Cracks are a path to the steel	No path, provided grouting is sound
Behaviour if overloaded	Visible cracking, ample warning	Little warning once compression is lost

That last row is the price. A prestressed wall gives you an excellent structure right up until the point it does not, and the warning signs are subtle. It is why the construction records on a post-tensioned tank are kept for the life of the asset.

On the left, an elevation of a cylindrical tank wall showing two tendon systems. Circumferential tendons run horizontally around the wall and are anchored at vertical buttresses that stand proud of the outer face; they resist the hoop tension from the liquid. Vertical tendons run from the base slab up to anchorages at the top ring beam; they resist bending at the restrained base and uplift from internal pressure on the dome. On the right, a cutaway of a bonded tendon shows, in order, the strand bundle, the cement grout filling the duct, the corrugated duct itself, and the surrounding concrete cover, with a vent shown at a high point of the duct. Below it, a list of the sources of prestress loss separates the immediate losses, which are friction along the duct, wedge draw-in at the anchorage, and elastic shortening of the concrete, from the long-term losses, which are creep, shrinkage and relaxation of the strand. — Two tendon systems for two different actions, and a cutaway of the one detail that decides whether the wall lasts forty years.

The two tendon systems

Circumferential (hoop) tendons run around the wall and carry the pressure of the contained liquid. They cannot run forever, so they are anchored at intervals — either at buttresses, vertical ribs standing proud of the outer face, or by overlapping loops that let successive tendons hand the force on to each other. Buttresses are the more common arrangement on LNG tanks and are what the photograph above shows.

Vertical tendons run from the base slab to the top ring beam. They are not there for the liquid pressure. They resist:

Bending at the base, where the slab restrains the wall from moving radially while the pressure and the temperature are both trying to make it move.
Uplift from internal pressure on the dome, which the roof passes down into the wall.
Seismic and wind overturning, in the few cases where those govern.

Losses: why the number on the jack is not the number in the wall

The force you apply at the jack is never the force that ends up in the concrete. Some of it goes immediately, and some leaks away over decades.

Immediate losses

Friction and wobble along the duct. The strand rubs where the duct curves, and also along the nominally straight lengths because no duct is perfectly straight. On a long circumferential tendon this is the dominant loss, and it is the reason tendons are stressed from both ends.
Anchorage draw-in. When the jack releases, the wedges bite and pull in a few millimetres. On a short tendon that small movement is a large proportion of the extension.
Elastic shortening of the concrete as it is compressed.

Long-term losses

Creep of the concrete under sustained compression.
Shrinkage as it dries.
Relaxation of the strand, which slowly sheds stress while held at constant length.

Together these commonly account for 20 to 25 per cent of the initial force. The design condition is not “compression at the end of construction” but compression at the end of design life, after all losses, with the tank in the spill case. Everything else is working backwards from that.

Grouting is the durability of the whole structure

In a bonded system, the duct is filled with cement grout after stressing. The grout bonds the strand to the concrete, but far more importantly it is the corrosion protection.

High-strength prestressing strand sits at roughly 70 per cent of its breaking load for its entire life. It is small in diameter, so a little section loss matters a great deal, and it has almost no ductile reserve: a corroded wire does not yield and sag, it snaps.

So a void in the duct — trapped air at a high point, or a pocket where bleed water collected and later drained away — leaves bare, highly stressed steel in a damp, enclosed space. It is the classic latent defect of post-tensioned construction, and it is invisible.

The defences:

Vents at every high point and low point of the duct profile, so air can escape and water can be drawn off.
Vacuum-assisted grouting or thixotropic, bleed-resistant grouts rather than plain cement-water mixes.
Verification after grouting — impact echo, or an endoscope through a drilled vent — on a sample of ducts, because pumping to refusal does not prove the duct is full.

Unbonded systems, where each strand is individually greased and sheathed, trade the bond for inspectability: the strands can be monitored, re-stressed and in some arrangements replaced. They are common in bridges and buildings; on LNG tanks the bonded system remains the norm, but the argument is a live one, and it is the same argument made about cast-in heating cables in frost heave and sub-soil heating.

Site control

Stressing is one of the few civil operations where the acceptance criterion is measured live, on the day, and the record cannot be reconstructed afterwards.

Elongation against calculation. The measured extension is compared with the value computed from the tendon profile and the assumed friction coefficients, typically with a tolerance of a few per cent. A tendon that reaches force but not extension is binding somewhere; one that extends too far may be slipping.
Calibrated gauges. The jack and gauge are calibrated as a pair, with certificates traceable to the stressing records.
Lift-off tests on a sample, to confirm the force actually locked into the anchorage.
Concrete strength at stressing, verified on cubes cured alongside the member rather than in a laboratory, because the anchorage zone sees very high local bearing stresses.
The stressing sequence, worked around the tank in balanced stages so the wall is not pulled out of round. The construction condition can be more demanding than anything in service.

Where this connects

The wall described here is the outer wall of the arrangement in LNG tanks: single, double and full containment.
Its cold behaviour, and why the reinforcement grade matters at temperature, is in what cold does to concrete.
Finding grout voids and delaminations without breaking anything open is covered in concrete NDT methods.

What to take away

Prestress exists to keep the wall in compression, which is what makes it tight rather than merely strong.
Hoop tendons take the liquid, vertical tendons take base bending and dome uplift.
Expect 20 to 25 per cent loss. The design case is the end of life, not the end of construction.
Grouting is the durability. A void is a latent, invisible, decades-long defect.
The records are part of the structure. Elongations, lift-offs and grout volumes are what a future assessor will have to work from.