Finding leaks: seam testing and leak location surveys

A common assumption is that liner leaks are welding failures. Studies of installed linings have consistently found the opposite: the majority of holes in a finished geomembrane are punctures and installation damage, not defective seams. Which shapes what a sensible testing regime looks like.

Where leaks actually come from

• Puncture from stones in the subgrade, under sustained head, over time.
• Damage during placement of cover soil or drainage stone — plant tracking on the liner.
• Pipe penetrations detailed badly or not at all.
• Cracking at over-tensioned anchor details and wrinkles.
• Damage after handover: pumps dropped in, vehicles at the crest, cleaning equipment.
• Genuine seam defects — real, but a minority, and the one category that is systematically tested.

Air-pressure testing

The workhorse for wedge welds. A dual-track seam has an unwelded air channel running between its two welded tracks. Seal both ends, insert a needle, pressurise to the specified value, and hold. If pressure holds within the allowance, the seam is continuous along its entire tested length.

This is the only routine method that tests a whole seam rather than a point on it, which is exactly why specifications insist on dual-track wedge welding for production seams. One detail matters: at the end of the test, the far end is cut open to confirm the pressure was actually travelling down the channel. Without that check, a blocked channel reads as a perfect seam.

Vacuum box and spark testing

Extrusion welds have no air channel and are tested differently. A vacuum box — a rigid frame with a transparent top and a soft gasket — is placed over a soaped section of weld and a partial vacuum drawn. Any through-hole pulls air in and produces bubbles.

Spark testing is the alternative where a vacuum box cannot seal, such as on complex detail work. A conductive wire is laid into the weld and a high-voltage probe passed over the finished bead; a hole completes the circuit and sparks. Both are point-by-point methods — the operator has to cover the whole weld, and the record is only as good as the coverage.

Destructive coupons

The only test that measures bond strength rather than absence of a hole. A coupon is cut from a live seam at a specified frequency, divided into strips, and pulled in peel and shear on a field tensiometer. The requirement is not simply a number: the seam must fail in the parent sheet, not at the bond. A specimen that separates along the weld interface fails regardless of the load recorded.

The hole left behind is patched by extrusion weld and vacuum tested like any other repair.

Electrical leak location

The method that finds what seam testing cannot. It exploits the fact that an intact geomembrane is an insulator: a voltage is applied across it, and current can only flow where there is a hole. An operator sweeps the surface with probes and the current concentration identifies the defect.

Two versions are common. The water puddle method is used on exposed liner, with a wetted surface. The covered liner method surveys after soil or drainage stone has been placed, which is far more valuable because that placement is when much of the damage happens. A survey after covering routinely finds holes that were not there when the seams passed.

It is an additional cost, and on any project where a leak would be expensive or environmentally serious it is money well spent.

Finding a leak in an existing basin

An existing lining that has started losing water is a different problem. Sequence:

1. Confirm it is a leak. Separate seepage from evaporation using the floating-container test — several still days, no abstraction, compare the basin drop with the container drop.
2. Check the obvious first. Pipe penetrations, the anchor trench, the overflow, and anywhere plant or people access the basin. This finds most leaks.
3. Survey. If the basin can be drained, walk it and inspect systematically, particularly at wrinkles and old repairs. If it cannot be drained, an electrical leak location survey can be run on a water-covered liner.
4. Repair and test. Round-cornered patch of the same material, extrusion-welded, vacuum tested. Then repeat step one to confirm the loss has actually stopped — basins frequently have more than one hole.

The economics favour looking early. A puncture found in the first year is a patch; the same puncture found in year five has taken a great deal of water with it.

Related questions